Investor engagement can curb cross-border externalities, but the best channel depends on politics and firm mobility: direct stewardship of multinationals limits leakage when foreign regulators cannot act, yet it can discourage foreign regulation when they could, whereas pushing for local regulation can induce complementary policies abroad and reduce global harms.

Main Finding

Investor "voice" can reduce global externalities arising from multinational firms, but the better channel—corporate stewardship (changing firm behavior directly) versus policy advocacy (shaping local regulation)—depends on cross‑border mobility and foreign political frictions. When foreign regulators cannot realistically regulate (high political frictions), stewardship is relatively more effective because it reduces production leakage and raises abatement globally. When foreign regulation is politically feasible, local advocacy can "crowd in" foreign regulation and improve global outcomes, while stewardship can unintentionally "crowd out" foreign regulation and sustain policy miscoordination.

Key Points

• Setting and margins

  • Two‑country model of a multinational firm that creates global externalities (paper framed as carbon emissions).
  • Firm chooses abatement (reduces global externality) and relocation (shifts production across jurisdictions). Relocation generates leakage: it shifts emissions/activities across borders without reducing global harm.
  • Regulators set unilateral local emissions taxes; each regulator faces political frictions and values domestic output and domestic welfare (so internalize externality imperfectly).

• Two investor channels

  • Policy advocacy: investors increase the regulator’s internal weight on the externality (pressuring for higher local tax).
  • Stewardship: investors increase the firm’s internal weight on the externality (firm behaves less like a pure profit maximizer and more like a partial internalizer across its global operations).

• Mechanisms and comparative statics

  • A higher local tax raises local abatement but induces relocation to lower‑tax jurisdictions (leakage). Thus advocacy can be undermined by relocation mobility.
  • Stewardship raises abatement in all locations and reduces the regulator’s need to rely on taxes; this narrows cross‑country tax differentials and thus reduces leakage (effect stronger when relocation is easier).
  • If foreign political frictions are so high that foreign regulation is impossible, advocacy is weakened by leakage while stewardship is especially valuable.
  • If foreign regulation is possible (but costly), advocacy can increase the likelihood and stringency of foreign regulation (strategic complementarity across regulators) and thus reduce global externalities. Stewardship, by lowering the marginal benefit of foreign regulation, can reduce the probability of foreign regulatory adoption (crowd‑out).

• Geographic scope of stewardship

  • If stewardship places heavier weight on domestic vs foreign harm (e.g., investors focus on visible/domestic impacts), it can itself create incentives for firms to relocate harm abroad, generating leakage driven by partial (local) stewardship.

• Policy and empirical implications (high level)

  • The optimal emphasis between stewardship and advocacy depends on where firms operate and on foreign political feasibility of regulation.
  • Investor actions have feedback effects on regulators’ incentives; stewardship and advocacy are not independent tools.
  • Empirical predictions: advocacy is less effective where relocation is cheap; stewardship reduces leakage when foreign regulation is absent; stewardship can be associated with lower propensity for foreign regulation when it is politically borderline; geographic bias in investor concern correlates with cross‑border shifting of harm.

Data & Methods

• Methodological approach

  • Analytical, game‑theoretic model with two countries and a multinational firm.
  • Regulators choose local taxes taking into account firm relocation responses and political costs of regulation.
  • Firm chooses global abatement and allocation of production; investors influence either regulator objective (advocacy) or firm objective (stewardship) via exogenously changing weights on the externality.
  • Comparative static and equilibrium analysis characterizes how investor channels affect local taxes, relocation, abatement, adoption of foreign regulation, and global externalities.

• Key assumptions

  • Externality is global and symmetric in its global damage, but regulators differ in political frictions and valuations of domestic output.
  • Relocation is costly but feasible; abatement is costly and reduces global harm.
  • Investor voice is modeled as changing objective weights (capturing engagement, public pressure, policy campaigns, shareholder stewardship) rather than changing financing or cost of capital directly.
  • Political frictions are parameterized so that foreign regulation may be impossible, possible but costly, or easy.

• No primary empirical data in the paper (the contribution is theoretical). The paper discusses empirical patterns and testable predictions and situates results within related empirical literature (on leakage, investor engagement, and cross‑border policy interactions).

Implications for AI Economics

• Mapping to AI externalities

  • Replace "emissions" with AI safety/harm risks that are global (e.g., misuse, model misalignment, systemic risks, illicit outputs). Firms can mitigate risk via internal safety practices (analogue of abatement) or shift risky activities (compute, data practices, model deployment, R&D) to jurisdictions with laxer rules (analogue of relocation/leakage).
  • Regulators set local AI rules/standards and face political constraints; cross‑jurisdictional fragmentation creates scope for regulatory arbitrage (move compute, incorporate abroad, host models in friendly jurisdictions).

• Practical takeaways for investors and policymakers in AI

  • When foreign jurisdictions are unlikely to regulate AI risks (high political frictions, low enforcement), investor stewardship of firm‑level safety standards is particularly valuable because it reduces global leakage and raises safety across all operations.
  • When multiple jurisdictions could realistically adopt AI safeguards, investor policy advocacy in jurisdictions with regulatory capacity can "crowd in" regulation abroad (through strategic complementarity), enhancing global coordination. In such contexts prioritizing advocacy can catalyze broader regulatory adoption.
  • Stewardship can have unintended effects: broad stewardship that reduces firm pressure on regulators or makes regulators feel less need to act may reduce the chance of desirable public regulation—important in borderline political contexts.
  • Geographic focus matters: investors who pressure firms only on risks in the investor’s jurisdiction (domestic focus) may induce firms to move risky activities abroad. For AI, that could look like shifting model training, data collection, or deployment to less regulated countries.
  • Combined strategies are context dependent: in some environments a mix—stewardship in jurisdictions where regulators are weak and advocacy where regulation is politically viable—will do best in reducing global AI risk.

• Empirical and policy research directions for AI economics

  • Empirical tests: examine how investor engagement (stewardship measures) and investor policy advocacy correlate with firm‑level measures of AI safety practices, cross‑border relocation of compute/R&D, and the timing/likelihood of local regulatory adoption. Use quasi‑experimental variation (staggered policy changes, exogenous investor campaigns) and difference‑in‑differences on affected vs unaffected firms/locations.
  • Metrics to gather: firm disclosures on safety controls, cloud usage and data center footprints, cross‑border staff and subsidiary shifts, public records of investor advocacy (policy letters, signatories, lobbying contributions), and regulators’ political‑friction proxies.
  • Policy design: international coordination (e.g., standards, mutual recognition, model export rules) reduces leakage and changes the relative returns to stewardship vs advocacy. Investors should align strategies with likely regulatory trajectories: advocate where it can tip policy adoption, steward where regulation is infeasible.

Overall, for AI economics the paper highlights that whether investors should focus on changing firm behavior or pushing for regulation depends critically on cross‑border mobility of risky activity and on the political feasibility of regulation abroad. Investor strategy and policymaking interact—so coordination across investors and between private and public actors matters to mitigate global AI externalities.