Digital markets compete on innovation and algorithms, not just price, and traditional antitrust tests miss these dynamics. Computational antitrust — combining simulations, structural inference, and large-scale platform data — can improve merger review and remedies if regulators secure data access and technical capacity.

Main Finding

Digitalization has shifted competition from primarily price-based rivalry to innovation-driven competition. Traditional antitrust frameworks, built around static price and quantity analysis, often miss the dynamic, multi-sided, and data-driven aspects of innovation competition. Computational antitrust — the use of modern computational, data-science, and simulation tools together with economic theory — can substantially improve antitrust analysis by making innovation central to market assessment, merger review, and remedies.

Key Points

• Shift in competitive locus: Firms increasingly compete through product quality, algorithms, data aggregation, network effects, and continual feature innovation rather than just price.
• Limits of traditional antitrust: Static price-focused tests and narrow market definitions struggle to capture dynamic entry, innovation incentives, multi-sided platforms, and algorithmic interactions.
• What computational antitrust offers:
  • Dynamic modeling of innovation incentives and strategic R&D investments.
  • Counterfactual simulations of mergers and policy interventions using agent-based models and structural estimation.
  • Causal inference on the effect of platform changes, product launches, or algorithm updates on innovation and welfare.
  • Large-scale empirical analysis using logs, clickstreams, patents, code, and other digital traces.
• Legal and institutional context: Recent literature and case law show increasing attention to non-price harms and potential precedents, but enforcement tools and evidentiary standards often lag technical realities.
• Proposed solutions (overview): Integrate innovation metrics into market definition and merger analysis; adopt dynamic merger simulation and ex ante assessments; deploy algorithmic transparency, monitoring, and sandboxing; use computational counterfactuals to inform remedies.

Data & Methods

• Typical data sources:
  • Transactional and pricing logs, ad-auction data, clickstream and engagement metrics.
  • Patent filings, R&D expenditures, software repositories, API logs.
  • Network and platform usage data, user cohorts, and firm-level financials.
  • Legal filings, pleadings, and case documents for doctrinal analysis.
• Methods and tools:
  • Structural econometric models that embed innovation decisions and dynamic incentives.
  • Agent-based and multi-agent simulation to model strategic interactions among firms and users over time.
  • Causal inference (difference-in-differences, synthetic controls, instrumental variables) applied to natural experiments and platform changes.
  • Machine learning and NLP for document analysis (e.g., contracts, communications), anomaly detection, and demand/choice modeling.
  • Network analysis to capture platform and ecosystem effects, and market power propagation.
  • Monte Carlo and counterfactual simulation to estimate welfare impacts of mergers or policies.
• Strengths and limitations:
  • Strengths: ability to model dynamics, handle high-dimensional data, produce richer counterfactuals, and surface non-price harms.
  • Limitations: data access and confidentiality, identification challenges for causal claims, model specification risk, computational cost, and interpretability of complex models for courts and regulators.

Implications for AI Economics

• Research direction: AI economists should prioritize models that endogenize innovation, algorithmic strategies, and dynamic complementarities across multi-sided markets.
• Empirical practice: Invest in assembling and curating rich digital datasets (logs, patents, repos) and in methods that combine structural inference with ML for robust counterfactuals.
• Policy design: Provide regulators with computational tools (simulators, monitoring dashboards, interpretable ML) to evaluate likely innovation effects of mergers and conduct ex ante assessments.
• Institutional change: Antitrust agencies will need technical capacity (data engineers, computational economists), frameworks for secure data access, and standards for model validation and explainability suitable for legal settings.
• Broader welfare measurement: Move beyond price-centric metrics to include quality-adjusted consumer surplus, long-run innovation externalities, and ecosystem-level impacts when assessing competition.
• Practice implications for firms: Expect increased scrutiny of innovation strategies and algorithmic interactions; build auditability and transparency into AI systems to ease compliance and oversight.