Generative AI strengthens market institutions and improves liquidity where regulation is strong; in poorly governed markets it can deepen information asymmetries. The net market benefits of GenAI therefore hinge critically on institutional quality.
The impact of generative AI on institutional efficiency: Regulatory and trading evidence from financial markets
Using a cross-market panel and multiple quasi-experimental techniques, the paper finds that generative AI adoption improves institutional efficiency and thereby increases liquidity and stabilizes volatility—effects that materialize mainly in markets with strong governance while low-governance markets may see amplified informational imbalances.
This paper examines how generative artificial intelligence (GenAI) affects institutional efficiency in financial markets through the combined roles of regulatory and trading institutions. While AI adoption in finance has expanded rapidly, most studies emphasize predictive accuracy and trading performance, leaving the institutional mechanisms through which GenAI influences markets less explored. We conceptualize GenAI diffusion as an institutional shock that reshapes information production, monitoring capacity, and enforcement quality, thereby affecting market efficiency through both regulatory and trading channels. Using a cross-market panel dataset and a novel proxy for GenAI adoption, we analyze its relationship with institutional efficiency and market outcomes. Our empirical approach employs fixed effects regressions, interaction models, instrumental variable estimation, and difference in differences designs to address endogeneity concerns. The results indicate that GenAI adoption significantly improves institutional efficiency, particularly in markets with strong governance and regulatory capacity. Institutional quality acts as a key moderating factor, amplifying efficiency gains while limiting the benefits in weaker environments. We also find asymmetric effects: in low governance markets, GenAI may intensify informational imbalances. Gains in institutional efficiency further transmit GenAI’s impact to trading dynamics by increasing liquidity and stabilizing volatility. Overall, the results underscore the importance of institutional context in shaping the economic consequences of generative AI.
Summary
Main Finding
Generative AI (GenAI) adoption improves institutional efficiency in financial markets overall, but the effects depend strongly on institutional quality. In well-governed markets GenAI amplifies efficiency gains (better monitoring, higher liquidity, and more stable volatility). In weak-governance settings GenAI can worsen information asymmetries and concentrate informational advantages among advanced actors.
Key Points
- Research question: How does GenAI diffusion affect institutional efficiency in financial markets via regulatory (RegTech/SupTech) and trading channels?
- Core result: Positive association between GenAI adoption and a composite Institutional Efficiency measure (regulatory quality, enforcement, investor protection, market transparency), with meaningful heterogeneity by governance.
- Moderation: Institutional quality is a key moderator — high-quality institutions amplify benefits; low-quality institutions attenuate or reverse them (H2, H3).
- Trading effects: Gains in institutional efficiency translate to better market microstructure — increased liquidity and reduced/stabilized volatility (H4).
- Asymmetry/risk: In weaker institutional environments, GenAI’s opacity, scale economies, and high fixed costs can increase informational asymmetries and concentration of informational rents.
- Conceptual contribution: Treats GenAI as an “institutional shock” that reshapes information production, monitoring capacity, and enforcement quality, integrating regulatory and trading institutions in one framework.
- Hypotheses tested: H1 (GenAI ↑ institutional efficiency), H2 (effect stronger with higher institutional quality), H3 (GenAI ↑ information asymmetry where institutions are weak), H4 (GenAI improves liquidity and stabilizes volatility).
Data & Methods
- Data scope: Cross-market panel covering multiple financial markets over 2010–2024.
- Main variables:
- Dependent: Institutional Efficiency (InstEff) — a multidimensional index reflecting regulatory quality, governance effectiveness, rule enforcement, investor protection, and market transparency.
- Key explanatory: Novel proxy for GenAI adoption (constructed from GenAI-related regulatory developments and market-level indicators of GenAI use — paper describes it as a “novel proxy”).
- Moderators/outcomes: Cross-country institutional quality measures and market microstructure indicators (liquidity, volatility, information efficiency).
- Sources: Internationally recognized financial, institutional, and technological databases (paper describes using aggregated regulatory developments, microstructure data, and institutional indices for comparability).
- Identification and estimation:
- Panel fixed effects regressions to control for time-invariant heterogeneity.
- Interaction models to assess moderation by institutional quality.
- Instrumental variables estimation and difference-in-differences designs to address endogeneity and provide causal leverage.
- Robustness checks across specifications (heterogeneity analyses by governance strength; alternative outcome measures).
- Empirical strategy aim: Separate regulatory channel (RegTech/SupTech improvements) and trading channel (information aggregation and execution) while addressing endogeneity of GenAI adoption.
Implications for AI Economics
- Institutional context matters: The economic effects of GenAI are conditional on governance and regulatory capacity. Models and policy analyses of AI adoption must incorporate institutional heterogeneity rather than treating AI as a homogeneous productivity shock.
- Distributional and market-structure concerns: GenAI may increase scale advantages and information concentration, particularly where enforcement is weak. AI economics should analyze how technological fixed costs and data/control of models create asymmetric returns across firms and countries.
- Policy direction: To realize GenAI’s systemic benefits (liquidity, volatility stabilization, stronger supervision), complementary investments in regulatory capacity, transparency, and governance are critical. Regulators should prioritize explainability, data access fairness, and supervisory RegTech deployment to avoid widening informational gaps.
- Research avenues:
- Micro-level causal evidence on firm/institution adoption and competitive dynamics (market shares, rents).
- Long-run effects on market concentration, entry, and competition in trading and financial intermediation.
- Mechanisms: disentangling surveillance/monitoring improvements from trading-side informational aggregation in driving market outcomes.
- Design of institutional safeguards (transparency, auditability, data-sharing) to mitigate asymmetric effects in weak-governance settings.
- Broader message: Evaluations of AI’s economic impact should jointly model technology and institutional capacity — policy and welfare implications depend on the institutional complementarity (or lack thereof) with AI adoption.
Assessment
Paper Typequasi_experimental
Evidence Strengthmedium — The paper uses multiple complementary quasi-experimental methods (FE, DID, IV, interaction models) and a cross-market panel, which increases credibility; however, it remains observational, relies on a novel proxy for GenAI adoption whose measurement error could bias results, and causal claims depend on IV validity and DID parallel-trends assumptions that are not guaranteed across heterogeneous markets.
Methods Rigormedium — The authors apply a range of standard econometric techniques and robustness checks (fixed effects, IV, diff-in-diff, heterogeneity analyses), indicating careful empirical work, but the absence of random assignment, potential omitted variables, possible invalid or weak instruments, and limited detail on tests for key identification assumptions constrain the methodological rigor.
SampleA multi-market panel dataset covering financial markets across jurisdictions over time, linking a novel market-level proxy for GenAI adoption to measures of institutional efficiency (information production, monitoring, enforcement) and market outcomes (liquidity, volatility); sample covers both stronger- and weaker-governance markets but exact countries/years and sample size are not specified in the summary.
Themesgovernance adoption innovation
IdentificationPanel fixed-effects with market and time controls, difference-in-differences exploiting variation in timing/intensity of GenAI adoption across markets, instrumental-variable estimation to address endogeneity of adoption, and interaction models to identify heterogeneous effects by institutional quality.
GeneralizabilityFindings apply to financial markets and may not generalize to other sectors (manufacturing, services)., Results depend on the validity and granularity of the GenAI adoption proxy; measurement error limits external validity., Heterogeneity across market types and legal regimes may limit applicability to specific countries or asset classes., Early-adoption period dynamics may differ from later stages as GenAI tools and regulations evolve., Potential selection: markets with better institutions may self-select into adoption, limiting inference for low-governance contexts.
Claims (7)
| Claim | Direction | Confidence | Outcome | Details |
|---|---|---|---|---|
| GenAI adoption significantly improves institutional efficiency. Governance And Regulation | positive | high | institutional efficiency (information production, monitoring capacity, enforcement quality) |
0.48
|
| Institutional quality acts as a key moderating factor, amplifying the institutional-efficiency gains from GenAI adoption in markets with strong governance and regulatory capacity. Governance And Regulation | positive | high | interaction effect of GenAI adoption and institutional quality on institutional efficiency |
0.48
|
| In weaker governance environments, the benefits of GenAI adoption for institutional efficiency are limited. Governance And Regulation | negative | high | magnitude of GenAI effect on institutional efficiency in low-governance markets |
0.48
|
| GenAI adoption may intensify informational imbalances in low-governance markets (asymmetric adverse effects). Decision Quality | negative | high | informational imbalances / information asymmetry |
0.48
|
| Gains in institutional efficiency from GenAI adoption transmit to trading dynamics by increasing market liquidity. Market Structure | positive | high | market liquidity |
0.48
|
| Gains in institutional efficiency from GenAI adoption transmit to trading dynamics by stabilizing market volatility. Market Structure | positive | high | market volatility / volatility stabilization |
0.48
|
| The economic consequences of generative AI in financial markets depend critically on institutional context (regulatory and governance capacity). Governance And Regulation | mixed | high | overall economic consequences (efficiency, liquidity, volatility) conditional on institutional context |
0.48
|