Competition Law and Synthetic Market Testing Platforms and Antitrust

1. Introduction

Synthetic Market Testing Platforms (SMTPs) are digital or AI-enabled systems that simulate market conditions before a product, price, promotion, allocation rule, algorithm, or business strategy is deployed in the real market.

A synthetic market may combine:

• artificial consumers and demand models;
• historical transaction data;
• competitor-price datasets;
• simulated supply and demand;
• AI-generated consumer behaviour;
• pricing and elasticity models;
• digital twins of markets;
• algorithmic experimentation;
• simulated auctions;
• virtual market-entry testing; and
• scenario analysis for mergers or strategic decisions.

From a competition-law perspective, these platforms can be pro-competitive or anti-competitive depending upon their design and use. A platform that independently predicts demand can improve innovation and reduce experimentation costs. Conversely, a platform that allows competitors to test, coordinate, optimise, or signal future prices may facilitate concerted practices, information exchange, collusion, algorithmic coordination, or exclusionary conduct.

A central difficulty is that the market being tested is partly artificial. Competition authorities therefore have to determine whether conduct occurring inside a simulated environment has sufficiently strong connections with actual or future competition to attract antitrust liability.

2. Meaning of Synthetic Market Testing

A synthetic market-testing platform can be understood as:

A technological environment in which market conditions, consumer behaviour, competitor responses, prices, quantities or other competitive variables are modelled or simulated to predict the effects of commercial strategies.

For example, ten competing retailers could independently use an AI system to simulate:

“What happens if all major retailers increase prices by 8% next month?”

That activity is not automatically unlawful.

The competition-law concern becomes greater if the platform allows those retailers to:

1. observe each other's confidential strategies;
2. communicate intended future prices;
3. receive recommendations based on competitors' confidential information;
4. coordinate algorithmic responses;
5. test which common price maximises industry profits;
6. punish firms that deviate from the simulated strategy; or
7. create a common pricing mechanism.

Thus, the distinction between independent market intelligence and facilitated coordination is fundamental.

3. Competition-Law Framework

Synthetic market-testing platforms can potentially implicate several areas of competition law.

A. Agreements and concerted practices

The principal concern is whether competing firms use the platform as a mechanism for coordination.

Relevant conduct may include:

• exchange of future pricing intentions;
• sharing of production plans;
• exchange of strategic capacity information;
• coordinated market-entry decisions;
• common algorithmic parameters;
• restrictions on discounting; and
• agreements to follow platform-generated recommendations.

In many jurisdictions, an express written agreement is unnecessary where the circumstances demonstrate a concerted practice.

B. Information exchange

Synthetic testing can transform commercially sensitive information into a powerful coordination mechanism.

Examples include:

• future prices;
• margins;
• inventory;
• production capacity;
• customer segmentation;
• promotional calendars;
• bidding strategies;
• expected demand; and
• strategic investment plans.

The competition problem can arise even where the platform never explicitly says:

“Competitors should collude.”

If the architecture facilitates the transmission of competitively sensitive information, competition authorities may examine the overall mechanism.

4. Algorithmic Collusion

Synthetic market testing becomes particularly important in algorithmic markets.

Suppose competing firms use a platform that repeatedly simulates:

1. Firm A raises price.
2. Firm B responds.
3. Firm C responds.
4. Demand is redistributed.
5. Algorithms identify the equilibrium producing the highest collective profit.

The platform could effectively become a coordination laboratory.

The firms might then implement the simulated equilibrium in the real market.

The important legal question is not merely:

“Did the computer communicate?”

Instead, authorities may examine:

Did businesses knowingly use the technological system in a manner that reduced independent competitive decision-making?

5. Hub-and-Spoke Risks

A synthetic market-testing platform can also operate as a potential hub connecting competing firms.

The structure may resemble:

Competitor A
↓
Synthetic Market Platform
↑
Competitor B

and

Competitor C
↓
Synthetic Market Platform

If the platform receives strategic information from multiple competitors and uses that information to influence their commercial behaviour, it can potentially become the centre of a hub-and-spoke arrangement.

The risk becomes greater where participants know that their competitors are participating and understand that information or recommendations are being shared across the system.

6. Dominance and Platform Power

A synthetic market-testing platform may itself become dominant.

This can occur where one platform controls:

• the largest consumer dataset;
• the most accurate demand model;
• essential simulation infrastructure;
• industry-wide benchmarking data;
• proprietary AI models;
• APIs required for market testing; or
• access to important market simulations.

A dominant platform could potentially engage in:

• discriminatory access;
• exclusion of competing simulation providers;
• tying;
• refusal to provide interoperability;
• self-preferencing;
• discriminatory data access;
• excessive licensing charges; or
• restrictions preventing customers from using competing platforms.

Thus, SMTP antitrust analysis is not limited to collusion between platform users.

It may also involve abuse of platform dominance.

7. Merger-Control Implications

Synthetic market-testing platforms can also affect merger analysis.

A platform might enable a proposed merger to simulate:

• unilateral price effects;
• diversion ratios;
• customer switching;
• capacity reductions;
• coordinated effects;
• entry conditions;
• innovation effects; and
• efficiencies.

However, the reliability of these simulations must be examined carefully.

A synthetic market is based upon assumptions.

If the model assumes:

low consumer switching,

when consumers actually switch extensively between suppliers, the predicted competitive effect may be substantially distorted.

Therefore, competition authorities may scrutinise:

• model assumptions;
• datasets;
• calibration;
• sensitivity analysis;
• omitted competitors;
• endogenous responses;
• model architecture; and
• potential manipulation of simulations.

8. Six Important Case Laws

The following cases do not all concern modern “synthetic market-testing platforms” by name. Rather, they establish principles that are highly relevant to the antitrust treatment of information exchange, algorithmic coordination, market simulation, pricing systems and technology-assisted collusion.

Case 1: FTC v. Cement Institute

333 U.S. 683 (1948), United States

Facts

The Federal Trade Commission challenged a system involving cement manufacturers in which extensive information concerning prices and sales was exchanged.

The information system contributed to a pricing structure that reduced independent competitive behaviour.

Principle

The Supreme Court recognised that systematic exchange of competitively significant information can have substantial effects on competition.

Relevance to synthetic market testing

An SMTP can modernise the same basic phenomenon.

Instead of competitors physically exchanging:

• price lists;
• sales figures; and
• customer information,

the platform could collect and process the information automatically.

The technological sophistication of the system does not remove the competition concern.

Lesson

Data-driven coordination can raise antitrust concerns even when the mechanism is technologically sophisticated.

Case 2: FTC v. Indiana Federation of Dentists

476 U.S. 447 (1986), United States

Facts

The Federal Trade Commission challenged conduct by dentists involving restrictions on the provision of information to insurers.

The case concerned the competitive significance of information and the ability of market participants to make informed purchasing decisions.

Principle

The Supreme Court treated restrictions on information in a competitive market as capable of producing significant anticompetitive effects.

Relevance

Synthetic market platforms frequently depend upon information:

• price;
• quality;
• demand;
• consumer preferences;
• product characteristics; and
• transaction histories.

The platform's control over information can therefore affect the competitive process itself.

Lesson

The antitrust analysis must examine how information affects competitive decision-making, rather than treating data as competitively neutral.

Case 3: United States v. Airline Tariff Publishing Co.

1994, U.S. Department of Justice

Facts

The case concerned airline fare information and computerised systems through which airlines communicated pricing information.

The government alleged that the system facilitated coordination by allowing airlines to communicate and monitor future pricing behaviour.

Principle

Computerised publication systems can facilitate unlawful coordination even where firms do not communicate through conventional private meetings.

Relevance to SMTPs

This is especially important for synthetic market platforms.

Imagine competing airlines using a common simulation platform to test:

• future fares;
• capacity;
• route withdrawals;
• discount levels; and
• responses to competitors.

If the platform enables competitors to observe or communicate strategic intentions, the system can potentially perform a role analogous to a sophisticated electronic pricing communication mechanism.

Lesson

Technology can become the infrastructure through which coordination occurs.

Case 4: Eturas UAB v Lietuvos Respublikos konkurencijos taryba

C-74/14, Court of Justice of the European Union (2016)

Facts

Eturas operated an online travel-booking system used by multiple travel agencies.

A system-wide technical restriction concerning discounts was introduced through the platform.

Principle

The CJEU examined whether participants could be considered involved in a concerted practice where a common electronic platform communicated a commercially significant restriction.

The Court emphasised the importance of knowledge and participation in assessing whether undertakings could be held responsible for coordinated conduct.

Relevance to synthetic market platforms

This case is highly relevant to platform-mediated competition.

Suppose an SMTP tells participating retailers:

“The optimal simulated market outcome occurs when discounts remain below 5%.”

If participating firms know that competing firms are receiving the same recommendation and subsequently implement it, the platform may create significant concerted-practice risks.

Lesson

A digital platform can be the mechanism through which coordinated commercial behaviour emerges.

Case 5: Trod Ltd / GB Eye Ltd

Competition and Markets Authority, UK, 2016

Facts

Online sellers agreed not to compete against one another on Amazon by maintaining agreed pricing arrangements.

The sellers used automated repricing software to implement their pricing strategies.

Principle

The UK competition authorities treated the use of automated technology as relevant to implementing an anticompetitive agreement.

Relevance to SMTPs

The case demonstrates that an algorithm does not create an antitrust safe harbour.

An SMTP could similarly:

1. identify a simulated common pricing strategy;
2. recommend that strategy;
3. connect competing firms;
4. automate implementation; and
5. monitor deviations.

The presence of AI or automated software would not necessarily eliminate the underlying competition-law problem.

Lesson

Automation may strengthen the practical effectiveness of an anticompetitive arrangement rather than neutralise it.

Case 6: United States v. Topkins

Northern District of California, 2015

Facts

The case concerned an agreement among online sellers to coordinate prices for posters and other goods.

The defendants used pricing algorithms to implement the arrangement.

One defendant pleaded guilty to criminal antitrust charges.

Principle

The case demonstrated that algorithmic pricing systems can be used as instruments for implementing a conventional price-fixing arrangement.

Relevance to synthetic market testing

An SMTP may create an additional stage before actual implementation.

The process could become:

Simulation → identification of common strategy → algorithmic recommendation → real-world implementation → automated monitoring

The fact that the initial coordination occurs through a simulated environment does not necessarily separate it from the real-world anticompetitive conduct.

Lesson

AI and algorithms can be evidence or instruments of coordination rather than independent defences to antitrust liability.

9. Comparative Significance of the Cases

10. Synthetic Market Testing and Tacit Coordination

One of the most difficult issues is tacit coordination.

Consider four competing firms using an independent market simulator.

The simulator predicts:

“If all four firms maintain prices within a narrow range, industry profits increase substantially.”

The firms do not communicate directly.

Nevertheless, each firm's algorithm independently reaches the same conclusion.

This creates a difficult distinction between:

Legitimate parallel conduct

Each company independently develops its own commercial strategy.

and

Coordinated conduct

The companies knowingly use a common mechanism that facilitates alignment.

Competition law generally does not prohibit mere parallel conduct simply because firms independently reach similar commercial decisions. The additional evidence concerning communication, knowledge, participation, information exchange or facilitating mechanisms can therefore become crucial.

11. Synthetic Consumers and Competition Law

Synthetic consumers are artificial agents designed to reproduce consumer behaviour.

They may be programmed to simulate:

• price sensitivity;
• brand loyalty;
• switching;
• income;
• geographic preferences;
• search behaviour;
• product substitution;
• purchasing frequency; and
• response to advertising.

This creates a new competition-law question:

Can an artificial representation of consumers distort the competitive analysis of a real market?

Potential problems include:

A. Biased consumer models

If the platform systematically underestimates switching, it could overstate market power.

B. Historical-data bias

Historical prices may reflect previous market power and therefore reproduce existing anticompetitive conditions.

C. Feedback loops

AI models trained on historical behaviour may repeatedly reinforce existing market structures.

D. Artificial coordination

Multiple competitors could use the same synthetic consumer environment and arrive at highly similar commercial strategies.

12. Market Definition Problems

Synthetic market testing can complicate traditional market definition.

Traditional competition analysis frequently asks:

Would consumers switch to another product following a small but significant price increase?

A synthetic platform may instead calculate:

• simulated diversion ratios;
• cross-price elasticities;
• hypothetical substitution;
• consumer utility;
• network effects; and
• predicted switching.

These outputs can be valuable but should not automatically replace empirical evidence.

A competition authority may need to compare:

Modelled behaviour + actual market evidence + historical evidence + consumer evidence.

13. Risks of Common Synthetic Market Models

A particularly significant risk arises when competing businesses use the same model.

Suppose:

• Retailer A uses Platform X.
• Retailer B uses Platform X.
• Retailer C uses Platform X.

The platform continuously learns from all participating users.

If the platform's optimisation objective is:

“Maximise expected industry profitability,”

rather than:

“Optimise each customer's independent competitive position,”

the platform could create substantial antitrust concerns.

The architecture therefore matters.

14. Data Governance and Antitrust

A synthetic market platform may process several categories of data.

Low-risk examples

• publicly available historical data;
• aggregated industry statistics;
• anonymised datasets;
• general economic indicators.

Higher-risk examples

• individual competitor's future prices;
• planned output;
• unpublished discounts;
• confidential customer information;
• strategic capacity;
• future investment plans.

The closer the information comes to future, firm-specific competitive strategy, the greater the competition-law concern.

15. Platform Design as a Competition-Law Issue

Competition compliance should therefore begin at the architecture stage.

A platform should consider:

Data separation

Competitor-specific information should not unnecessarily flow between participants.

Aggregation

Data can be aggregated so that individual competitors cannot be identified.

Anonymisation

Identifiable strategic information should be removed where feasible.

Independent optimisation

Each firm's model should optimise its own business rather than industry-wide profits.

Access controls

Participants should not be able to inspect competitors' confidential strategies.

Audit trails

The platform should preserve records showing:

• who supplied data;
• what data were used;
• what recommendations were generated;
• who received them; and
• whether recommendations were implemented.

16. Competition Risks Across the Platform Life Cycle

17. Synthetic Market Testing and Abuse of Dominance

A dominant SMTP provider could potentially exploit its position by:

Refusal to provide access

Competitors may depend upon the platform's unique market data or simulation infrastructure.

Discriminatory access

The platform may provide better data or model functionality to affiliated businesses.

Self-preferencing

A platform operating its own competing business could use synthetic-market insights to favour its own products.

Bundling

Access to the simulation engine could be tied to unrelated products.

Data foreclosure

The platform could prevent customers from transferring their historical data to competing simulation providers.

These issues may engage abuse-of-dominance principles depending upon the applicable jurisdiction and market conditions.

18. Efficiency and Pro-Competitive Benefits

Synthetic market testing is not inherently anticompetitive.

It can generate substantial benefits.

Innovation

Companies can test products without costly real-world experiments.

Consumer welfare

Businesses can better understand consumer preferences.

Reduced waste

Demand forecasting can prevent overproduction.

Improved entry

Small firms can simulate market-entry strategies.

Better regulatory analysis

Authorities can use simulations to examine potential effects of mergers and conduct.

Sustainability

Synthetic markets can test:

• carbon pricing;
• renewable-energy adoption;
• electric-vehicle demand;
• sustainable products; and
• circular-economy models.

The competition-law analysis therefore requires examination of both competitive risks and legitimate economic functions.

19. Compliance Framework

A synthetic market-testing platform should ideally implement a competition-compliance framework consisting of:

Step 1 — Classify data

Identify whether information is:

• public;
• aggregated;
• anonymised;
• commercially sensitive; or
• future strategic information.

Step 2 — Identify users

Determine whether users are:

• competitors;
• suppliers;
• customers;
• regulators;
• consultants; or
• internal business units.

Step 3 — Separate competitor environments

Prevent unnecessary cross-user visibility.

Step 4 — Audit recommendations

Determine whether the algorithm recommends:

• independent strategies; or
• market-wide alignment.

Step 5 — Prevent common pricing rules

Avoid functionality that automatically establishes common prices among competing users.

Step 6 — Preserve audit logs

Maintain evidence demonstrating independent decision-making.

Step 7 — Competition review

High-risk functionality should undergo antitrust review before deployment.

20. Key Legal Questions for Courts and Competition Authorities

Future cases involving SMTPs may ask:

1. Who designed the simulation?
2. Who supplied the data?
3. Were competitors aware of one another's participation?
4. Could users access competitor-specific information?
5. Did the platform communicate future commercial intentions?
6. Did participants knowingly rely upon common recommendations?
7. Was the algorithm optimising individual or collective profitability?
8. Did firms independently determine their final prices?
9. Did the platform monitor deviations?
10. Were deviations penalised?
11. Did the system facilitate market-wide alignment?
12. Did the platform possess substantial market power?
13. Were competitors denied access to essential data?
14. Were efficiencies objectively verifiable?
15. Did the conduct produce or threaten actual competitive harm?

21. Emerging Concept: Synthetic Market Manipulation

A particularly novel issue is synthetic market manipulation.

This occurs where a business intentionally manipulates the assumptions or datasets of a synthetic market in order to obtain a desired competitive outcome.

For example:

A dominant firm deliberately feeds exaggerated competitor-response data into a market simulator so that the model predicts that a rival will be unable to compete.

That output could then be used to justify:

• exclusionary pricing;
• acquisition of the rival;
• discriminatory access;
• refusal to interoperate; or
• strategic capacity decisions.

The competition authority may therefore need to examine not only what the simulation predicts, but how the prediction was produced.

22. Evidentiary Importance

SMTPs may become significant sources of evidence in antitrust investigations.

Authorities may examine:

• source code;
• model documentation;
• prompts;
• training datasets;
• system logs;
• API records;
• model outputs;
• internal emails;
• user instructions;
• recommendation histories;
• changes to algorithms; and
• implementation records.

An important principle is that the final price alone may not reveal the entire competitive problem.

The architecture and decision-making process behind the price may be equally important.

23. Hypothetical Example

Assume three competing online retailers use the same synthetic-market platform.

The platform receives:

• Retailer A's proposed prices;
• Retailer B's proposed prices;
• Retailer C's proposed prices.

It simulates 1 million synthetic consumers.

The platform discovers:

A common 10% price increase would maximise the combined expected profits of the three retailers.

It then recommends the same price strategy to all three.

The retailers implement the recommendation.

Potential concerns

The authority could examine:

• exchange of confidential pricing information;
• common algorithmic coordination;
• knowledge of competitor participation;
• communication through the platform;
• implementation of a common strategy;
• monitoring of deviations; and
• whether the platform functioned as a facilitating mechanism.

The legal conclusion would depend upon the evidence and the applicable jurisdiction.

24. Relationship Between Synthetic Markets and Traditional Antitrust

The development can be represented as:

Traditional Competition Law

↓

Information Exchange

↓

Electronic Information Exchange

↓

Algorithmic Pricing

↓

Common Data Platforms

↓

Synthetic Market Simulation

↓

AI-Based Market Prediction

↓

Autonomous Market Optimisation

The fundamental competition-law principles remain relevant, but the technological mechanisms become increasingly complex.

25. Conclusion

Synthetic Market Testing Platforms represent an emerging intersection between competition law, artificial intelligence, data governance, algorithmic pricing and economic simulation.

Their legitimate function is to allow businesses and regulators to understand markets before making decisions. However, the same technology can create competition risks where it:

• aggregates competitors' sensitive information;
• communicates future strategies;
• facilitates price alignment;
• produces common recommendations;
• automates coordinated conduct;
• monitors competitive deviations; or
• enables a dominant platform to exclude rivals.

The cases of Cement Institute, Indiana Federation of Dentists, Airline Tariff Publishing, Eturas, Trod/GB Eye and Topkins demonstrate that competition law has repeatedly focused on the economic substance and competitive effect of information and coordination mechanisms, rather than allowing technological form to determine legality.

Accordingly, the central principle for synthetic market-testing platforms is:

Simulation itself is generally a tool; the competition-law issue arises from the information architecture, purpose, participant conduct, algorithmic recommendations and real-world competitive consequences associated with the simulation.

Short exam formulation

Synthetic Market Testing Platforms should therefore be analysed through five principal antitrust questions:

Data → Coordination → Algorithm → Market Power → Competitive Effects

This framework enables competition authorities to distinguish legitimate market experimentation and forecasting from platform-enabled information exchange, algorithmic coordination and exclusionary conduct.