Digital Twins for Competition Enforcement Simulation

1. Introduction

Digital Twins for Competition Enforcement Simulation refers to the use of computational replicas of markets, firms, consumers, supply chains, platforms, algorithms, and regulatory environments to simulate how competition-law interventions may affect market outcomes.

A digital twin in competition enforcement is not merely a statistical model. It can represent a continuously updated virtual environment in which an authority can test hypothetical scenarios such as:

• whether a merger is likely to substantially lessen competition;
• whether a dominant platform's conduct forecloses rivals;
• whether an algorithm facilitates tacit coordination;
• whether access to data or infrastructure is competitively essential;
• how a remedy such as interoperability, data portability, divestiture or access obligations may operate;
• whether an undertaking's proposed commitments actually restore competition.

The concept is particularly relevant to modern digital markets, where conventional enforcement may struggle with rapid technological change, zero-price services, network effects, multi-sided platforms, algorithmic pricing and complex ecosystems.

2. Meaning of a Competition-Enforcement Digital Twin

A competition-enforcement digital twin can be understood as a virtual representation of a relevant competitive environment that allows an authority to reproduce, manipulate and observe market behaviour under alternative legal or economic scenarios.

A simplified model can be represented as:

Real Market → Data → Digital Twin → Simulated Intervention → Predicted Competitive Outcome → Enforcement Decision

The twin may contain representations of:

1. Firms
2. Consumers
3. Competitors
4. Suppliers
5. Platforms
6. Algorithms
7. Pricing systems
8. Distribution networks
9. Data flows
10. Regulatory constraints

For example, an authority investigating a dominant online marketplace could construct a digital twin containing:

platform → sellers → consumers → ranking algorithm → advertising system → payment system → competing marketplaces.

The authority could then simulate what happens if the platform:

• gives preferential ranking to its own products;
• removes self-preferencing;
• permits interoperability;
• requires data portability;
• increases commission rates;
• restricts third-party sellers;
• changes its advertising algorithm.

3. Why Digital Twins Matter to Competition Enforcement

Traditional competition enforcement frequently relies on historical evidence.

Digital twins introduce a more prospective and experimental approach.

Traditional enforcement

What happened?

Digital-twin enforcement

What would happen if this conduct continued?

and:

What would happen if the authority imposed Remedy A instead of Remedy B?

This can be particularly important where markets are changing rapidly.

For example, an authority may not have sufficient historical observations to determine the effects of a new AI-driven pricing system. A digital twin can simulate alternative pricing environments using observed behavioural parameters.

4. Core Components

A. Market Twin

The first component is a representation of the relevant market.

It may contain:

• market shares;
• demand elasticities;
• switching rates;
• customer segmentation;
• entry barriers;
• network effects;
• supply relationships;
• geographic constraints;
• innovation variables.

The authority can then simulate competitive interactions among market participants.

B. Firm Twin

Each major undertaking can be represented as a virtual agent.

The model may incorporate:

• pricing;
• production;
• investment;
• advertising;
• innovation;
• acquisition strategies;
• contractual restrictions;
• capacity;
• distribution decisions.

A dominant firm can therefore be tested against hypothetical competitors.

C. Consumer Twin

Consumer behaviour is critical in digital markets.

A consumer model could incorporate:

• price sensitivity;
• switching costs;
• privacy preferences;
• loyalty;
• search behaviour;
• platform preferences;
• multi-homing;
• susceptibility to personalization.

For example:

If a platform makes switching easier, how many consumers would actually leave?

This distinction between formal switching possibility and actual switching behaviour can be important in assessing market power.

5. Algorithmic Twin

An especially important feature is the ability to replicate algorithmic decision-making.

The authority could model:

• recommendation algorithms;
• ranking systems;
• pricing algorithms;
• advertising auctions;
• matching systems;
• procurement algorithms;
• credit-allocation systems.

This allows investigators to ask whether apparently independent algorithms generate systematically anti-competitive outcomes.

6. Simulation of Algorithmic Collusion

Digital twins could be used to examine potential coordination between algorithms.

For example:

Firm A Algorithm ↔ Market Data ↔ Firm B Algorithm

The authority could run repeated simulations to determine whether:

• prices converge;
• discounts disappear;
• deviations are punished;
• market volatility falls;
• supracompetitive prices persist.

Importantly, simulation would not itself establish an infringement.

It would instead help identify:

• suspicious patterns;
• causal mechanisms;
• investigative leads;
• counterfactual scenarios.

Actual enforcement would still require legally admissible evidence.

7. Merger Enforcement

Digital twins may be particularly valuable in merger control.

An authority can construct a pre-merger market twin and a post-merger market twin.

Scenario 1 — No merger

A
↓
B
↓
C
↓
Competitive prices

Scenario 2 — Merger

A + B
↓
Reduced competitive constraint
↓
Higher prices / reduced innovation

The authority could compare the two environments.

Potential variables include:

• unilateral effects;
• coordinated effects;
• diversion ratios;
• entry;
• innovation;
• network effects;
• efficiencies.

This could complement traditional merger simulation models such as Bertrand or differentiated-products models.

8. Digital Twins and Counterfactual Analysis

Counterfactual analysis is central to competition law.

The relevant question is often:

What would the market have looked like absent the allegedly anti-competitive conduct?

A digital twin can create alternative worlds.

World 1

Actual market.

World 2

Market without exclusionary conduct.

World 3

Market with interoperability.

World 4

Market with data portability.

World 5

Market after structural separation.

The authority can compare outcomes across these simulated environments.

9. Remedy Simulation

One of the most important applications is remedy design.

Suppose a dominant platform has been found to engage in exclusionary conduct.

The authority could simulate:

Remedy A

Fine only.

Remedy B

Behavioural commitment.

Remedy C

Interoperability.

Remedy D

Data portability.

Remedy E

Structural separation.

The twin could estimate effects on:

• entry;
• prices;
• innovation;
• consumer switching;
• market concentration;
• platform quality.

This could reduce the risk of imposing a remedy that appears theoretically appropriate but produces unintended competitive consequences.

10. Digital Twins and Market Definition

Digital twins may also assist market-definition analysis.

Instead of looking exclusively at observed substitution, an authority could simulate:

What happens if the price or quality of Product A changes?

For digital services, "price" may be zero.

Consequently, the twin can incorporate non-price dimensions such as:

• privacy;
• quality;
• advertising intensity;
• data collection;
• latency;
• functionality.

This is particularly relevant to platforms where consumers pay with attention or data rather than money.

11. Digital Twins and Essential Facilities

Suppose a dominant cloud provider controls infrastructure necessary for competitors.

A digital twin can simulate:

What happens if access is denied?

and:

What happens if access is offered on regulated terms?

Variables could include:

• competitor entry;
• prices;
• capacity;
• innovation;
• migration;
• switching costs.

This can help authorities evaluate theories involving essential facilities, refusal to deal and access discrimination.

12. Digital Twins and Abuse of Dominance

A digital twin can simulate different forms of potentially abusive conduct.

Examples

Self-preferencing

Platform → Own service → privileged ranking

versus

Platform → Neutral ranking

Tying

Product A → mandatory Product B

versus

Product A → independent choice

Exclusive dealing

Supplier → Platform A only

versus

Supplier → multi-platform distribution

Predatory pricing

Price below cost → rival exit → price increase

The twin helps test the causal chain underlying the theory of harm.

13. Digital Twins and Dynamic Markets

Competition authorities increasingly confront markets in which today's structure may not predict tomorrow's structure.

Digital twins can model:

• innovation;
• technological disruption;
• entry;
• exit;
• investment;
• consumer migration;
• technological standards.

This makes them particularly useful for dynamic competition.

The relevant question becomes:

Does the conduct merely affect current prices, or does it change the future competitive trajectory of the market?

14. Major Legal and Competition-Law Case Laws

The following cases do not necessarily concern "digital twins" themselves. Rather, they establish legal principles that provide the doctrinal foundation for using simulation, counterfactual analysis, economic modelling and technologically sophisticated evidence in competition enforcement.

1. United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001)

The Microsoft litigation concerned Microsoft's conduct relating to operating systems, browsers and distribution channels.

The case is highly relevant to digital-twin enforcement because the court examined:

• network effects;
• barriers to entry;
• platform economics;
• exclusionary conduct;
• technological integration;
• effects on future competition.

Relevance

A digital twin of a platform market could reproduce the competitive relationships examined in Microsoft.

For example, an authority could simulate:

operating-system dominance → browser distribution → consumer adoption → developer incentives → reinforcement of platform dominance.

Principle: Competition analysis in technology markets must account for network effects and dynamic competitive conditions.

2. United States v. Google LLC, 2024 WL 4281218 (D.D.C. 2024)

The Google search case is particularly relevant to modern simulation-based enforcement.

The litigation examined Google's distribution arrangements and their effect on search competition.

A digital twin could model:

default placement → user behaviour → search volume → data accumulation → quality improvements → stronger defaults.

This creates a feedback loop.

Relevance

The case illustrates why enforcement authorities may need to model competitive effects extending beyond immediate prices.

Principle: Distribution arrangements can reinforce market power where scale, defaults and data produce self-reinforcing competitive advantages.

3. European Commission v. Google (Shopping), Case T-612/17

The Google Shopping litigation concerned Google's treatment of competing comparison-shopping services.

The underlying theory involved:

• dominance;
• search results;
• preferential positioning;
• traffic;
• foreclosure.

A digital twin could model:

ranking preference → traffic diversion → rival deterioration → reduced competitive pressure.

Relevance

This is almost an archetypal environment for simulation.

An authority could compare:

1. neutral ranking;
2. self-preferencing;
3. different ranking algorithms.

Principle: Algorithmic ranking can have competitive effects through control over access to users and traffic.

4. Intel Corp. v. European Commission, Case C-413/14 P

The Intel litigation is particularly important for economic analysis of exclusionary rebates.

The Court of Justice required greater attention to whether the conduct was capable of producing exclusionary effects and emphasized the relevance of economic analysis in appropriate circumstances.

Relevance to digital twins

A digital twin could simulate:

• effective prices;
• rival costs;
• customer switching;
• foreclosure;
• contestable demand.

The model could therefore help an authority investigate whether conduct actually has the capacity to exclude an equally efficient competitor.

Principle: Economic effects and competitive capability can be critical in analysing exclusionary conduct.

5. Post Danmark A/S v. Konkurrencerådet, Case C-209/10

The case concerned exclusionary pricing and the treatment of below-cost pricing by a dominant undertaking.

It is relevant to digital-twin enforcement because pricing conduct can be examined through counterfactual models.

A twin could simulate:

dominant firm price → rival response → customer switching → rival exit → post-exclusion price.

Relevance

The case supports the broader proposition that competition authorities may need to assess actual competitive effects rather than merely the formal existence of aggressive pricing.

Principle: The assessment of exclusionary pricing requires attention to the competitive context and effects.

6. United Brands Co. v. Commission, Case 27/76

United Brands is foundational to EU dominance law.

The case addressed:

• relevant market;
• dominance;
• barriers to entry;
• commercial behaviour;
• exclusionary effects.

Relevance

A digital twin can operationalize the type of market analysis that United Brands helped establish.

For example, it can simulate whether consumers can realistically switch to alternative products and whether competitors can constrain a dominant firm.

Principle: Market power must be assessed within the economic and competitive circumstances of the relevant market.

7. Bronner v. Mediaprint, Case C-7/97

Bronner is important for refusal-to-deal and essential-facility analysis.

The Court imposed stringent conditions before requiring a dominant undertaking to provide access to infrastructure.

Digital-twin relevance

A regulator could simulate:

No access → rival exclusion

against:

Mandatory access → competitive entry

and assess whether the infrastructure is genuinely indispensable.

Principle: Not every commercially important infrastructure qualifies as an essential facility; indispensability and competitive necessity matter.

8. Commission v. IMS Health, Case C-418/01

IMS Health concerned access to an intellectual-property protected structure that competitors needed for market participation.

It is relevant to digital twins because modern digital ecosystems frequently involve proprietary:

• data structures;
• APIs;
• interoperability standards;
• software architectures.

A simulation could examine whether denying access eliminates effective competition and whether an access remedy restores competitive conditions.

Principle: Under exceptional circumstances, control over indispensable infrastructure or intellectual property can generate competition-law obligations concerning access.

15. Indian Competition-Law Relevance

Digital-twin enforcement can also fit within India's competition-law framework, particularly the Competition Act, 2002.

Relevant provisions include:

• Section 3 — anti-competitive agreements;
• Section 4 — abuse of dominant position;
• Section 5 — combinations;
• Section 19 — inquiry into agreements and dominance;
• Section 26 — investigation procedure;
• Section 27 — orders after inquiry;
• Section 31 — orders relating to combinations.

The Competition Commission of India could potentially use simulation as an economic and investigative tool, subject to the evidentiary requirements applicable to the particular proceeding.

16. Digital Twins and Evidence

A major legal problem is:

Can a simulation itself constitute proof of an infringement?

Normally, it should not automatically be treated as conclusive proof.

A digital twin should be viewed as one evidentiary component alongside:

• internal documents;
• emails;
• contracts;
• transaction data;
• algorithmic logs;
• source-code evidence;
• witness evidence;
• consumer evidence;
• economic evidence.

The authority must establish that:

1. the underlying data are reliable;
2. the model is methodologically sound;
3. assumptions are transparent;
4. alternative specifications have been tested;
5. uncertainty is disclosed.

17. The Problem of Model Bias

Digital twins can create a new competition-enforcement risk:

regulatory model bias.

Suppose the authority's twin assumes that consumers switch rapidly between platforms.

If consumers actually face:

• high switching costs;
• data lock-in;
• behavioural inertia;

the simulation may underestimate market power.

Conversely, assuming excessive switching costs may overstate dominance.

Therefore, authorities should conduct sensitivity analysis.

Example

Model A:

Switching rate = 30%

Model B:

Switching rate = 15%

Model C:

Switching rate = 5%

If the enforcement conclusion changes dramatically between models, the authority should treat the result cautiously.

18. Explainability and Procedural Fairness

A digital-twin system creates an important procedural question:

Can the investigated undertaking understand and challenge the model used against it?

This is particularly important when enforcement decisions rely upon sophisticated algorithms.

A fair system should allow appropriate disclosure of:

• model architecture;
• relevant assumptions;
• variables;
• data sources;
• uncertainty;
• sensitivity analysis;
• error rates.

Otherwise, the undertaking could effectively be required to defend itself against a black-box regulatory conclusion.

19. Digital Twins and Regulatory Sandboxes

Competition authorities could establish controlled competition-enforcement simulation environments.

For example:

Stage 1

Create market model.

↓

Stage 2

Validate against historical data.

↓

Stage 3

Introduce hypothetical conduct.

↓

Stage 4

Run simulations.

↓

Stage 5

Compare alternative remedies.

↓

Stage 6

Conduct human legal assessment.

↓

Stage 7

Issue enforcement decision.

This separates machine-assisted analysis from legal judgment.

20. Digital Twin Enforcement Architecture

A sophisticated enforcement system could look like:

                    REAL MARKET                        │          ┌─────────────┼─────────────┐          ↓             ↓             ↓       Firms        Consumers      Platforms          │             │             │          └─────────────┼─────────────┘                        ↓                 MARKET DATA                        ↓                 DIGITAL TWIN                        │        ┌───────────────┼────────────────┐        ↓               ↓                ↓   Merger Test     Conduct Test      Remedy Test        │               │                │        ↓               ↓                ↓  Counterfactual   Foreclosure       Intervention        │               │                │        └───────────────┼────────────────┘                        ↓             ECONOMIC ANALYSIS                        ↓              LEGAL ASSESSMENT                        ↓               ENFORCEMENT DECISION

21. Advantages

1. Better prediction

Authorities can evaluate future competitive consequences.

2. Counterfactual testing

Multiple alternative market conditions can be compared.

3. Remedy optimisation

Different remedies can be tested before implementation.

4. Early detection

Suspicious algorithmic or market behaviour may be detected earlier.

5. Dynamic analysis

Innovation and technological change can be incorporated.

6. Complex-system analysis

Digital ecosystems involving many interacting firms can be studied simultaneously.

22. Limitations

Digital twins also present substantial risks.

A. Garbage-in, garbage-out

Poor data produce poor predictions.

B. Model uncertainty

A simulation is not reality.

C. Strategic manipulation

Firms may alter behaviour when they know the authority's model.

D. Transparency

Complex models may be difficult for courts and investigated undertakings to scrutinise.

E. False precision

A numerical prediction can appear more certain than it actually is.

F. Regulatory overreach

Authorities may become overly dependent on predictions rather than established legal evidence.

G. Privacy

Consumer-level digital twins may require extremely sensitive behavioural datasets.

23. Digital Twins and AI Competition Enforcement

The importance of digital twins becomes greater with autonomous AI agents.

Consider:

AI Agent A → price decision
AI Agent B → price decision
AI Agent C → inventory decision
AI Agent D → advertising decision

These agents interact continuously.

A digital twin could simulate thousands of iterations to determine whether market conditions encourage:

• coordination;
• exclusion;
• predatory strategies;
• market tipping;
• discriminatory access;
• algorithmic retaliation.

This could become an important investigative technique for future competition authorities.

24. Digital Twins and Institutional Accountability

There is also a constitutional and administrative-law dimension.

The ultimate decision should ordinarily remain attributable to the competition authority, not to the simulation.

A useful principle is:

Digital twin for prediction; human authority for adjudication.

The digital twin may recommend:

"Remedy B produces the strongest improvement in competitive conditions."

But the authority must still determine:

• whether the conduct violates competition law;
• whether the legal test is satisfied;
• whether evidence is sufficient;
• whether the remedy is proportionate.

25. Relationship Between Simulation and Competition-Law Standards

Digital twins should not replace legal standards such as:

• agreement or concerted practice;
• dominance;
• substantial foreclosure;
• appreciable restriction of competition;
• substantial lessening of competition;
• consumer harm;
• efficiencies;
• indispensability;
• causation;
• proportionality.

Instead:

Law defines the question → Digital twin tests economic scenarios → Evidence validates assumptions → Authority applies the law.

26. Key Case-Law Principles at a Glance

27. Conclusion

Digital Twins for Competition Enforcement Simulation represent a potential evolution from largely retrospective competition enforcement toward continuous, predictive and counterfactual competition analysis.

Their greatest value lies in complex markets where:

• algorithms interact dynamically;
• network effects reinforce dominance;
• data create feedback loops;
• consumers multi-home or face switching costs;
• market boundaries evolve rapidly;
• remedies have uncertain consequences.

However, a digital twin should not become a substitute for legal judgment. Its proper role is to provide a structured environment for testing economic hypotheses and counterfactuals.

The emerging enforcement model can therefore be expressed as:

Market Data → Digital Twin → Counterfactual Simulation → Economic Evidence → Legal Evaluation → Human Enforcement Decision

The central legal challenge will be ensuring that increasingly sophisticated simulation systems remain transparent, contestable, evidence-based, proportionate and subject to judicial review.