Competition Law and Governance of Self-Monitoring Market Structures

Introduction

A self-monitoring market structure is a market in which firms, platforms, algorithms, intermediaries, or industry participants continuously observe market conditions and adjust their conduct without requiring a human decision-maker for every transaction.

Examples include:

• algorithmic pricing and dynamic pricing;
• automated competitor-price monitoring;
• marketplace ranking and recommendation systems;
• automated inventory and supply decisions;
• self-executing contractual mechanisms;
• AI-based detection of deviations from commercial strategies;
• platforms monitoring sellers, buyers, competitors and consumer behaviour;
• third-party software that collects information from several competing firms and generates recommendations.

Self-monitoring itself is not unlawful. It can increase efficiency, improve matching, reduce search costs and allow firms to respond rapidly to market conditions. The competition-law concern arises when monitoring changes from legitimate observation into coordination, information exchange, exclusion, self-preferencing, price alignment or automated enforcement of an anticompetitive arrangement.

Modern competition authorities increasingly recognise that pricing algorithms and analytical tools can increase market observability and make coordination easier. The U.S. merger guidelines, for example, specifically identify pricing algorithms and surveillance tools that track competitors' prices or actions as mechanisms that can increase observability.

1. Meaning of Self-Monitoring Market Structures

A conventional market generally operates through independent decisions:

Competitor A → observes market → makes decision

Competitor B → observes market → independently makes decision

In a self-monitoring market:

Market data → algorithm → continuous observation → automated response → new market data → algorithmic response

This creates a feedback loop.

Basic structure

Data collection

↓

Market monitoring

↓

Prediction / comparison

↓

Automated decision

↓

Market response

↓

Further monitoring

↓

Further automated adjustment

The competition-law problem is that such systems can sometimes transform independent competition into automated interdependence.

2. Why Competition Law Is Concerned

Self-monitoring can affect competition through five principal mechanisms.

A. Increased market transparency

When competitors can observe each other's prices almost instantly, deviations from a common pricing pattern become easier to detect.

B. Faster retaliation

An algorithm can immediately respond when a rival reduces price.

C. Reduced uncertainty

Competition normally involves uncertainty about rivals' future conduct. Continuous monitoring can reduce that uncertainty.

D. Automated coordination

A firm may use software to implement an agreement without employees manually communicating every pricing decision.

E. Strategic self-preferencing

A vertically integrated platform can use information obtained from competitors to modify its own ranking, pricing, inventory or product decisions.

Thus, the important question is not simply:

"Does the firm monitor the market?"

but:

"What information is being monitored, who receives it, how is it processed, and what competitive response does the system generate?"

3. Legal Framework

A. Agreements and concerted practices

Self-monitoring can fall within cartel law when monitoring is used to implement or police an agreement.

The classical requirements remain relevant:

1. agreement or concerted practice;
2. coordination between independent undertakings;
3. restriction of competition;
4. effect or object prohibited by the applicable law.

The fact that a computer program executes the arrangement does not necessarily remove the underlying human or corporate responsibility.

B. Abuse of dominance

A dominant platform can use self-monitoring infrastructure to:

• obtain competitors' confidential information;
• discriminate between sellers;
• manipulate rankings;
• favour its own products;
• impose discriminatory access conditions;
• use rivals' data to enter adjacent markets;
• automatically exclude competitors.

Therefore, the same technological architecture may create both Article 101-type coordination concerns and Article 102-type exclusionary concerns in the EU context, or corresponding national-law issues.

C. Information exchange

Self-monitoring becomes particularly sensitive when information concerns:

• current prices;
• future prices;
• output;
• capacity;
• inventories;
• customer-specific information;
• discounts;
• margins;
• strategic plans.

A third-party algorithm can potentially become a hub through which competitors' commercially sensitive information is aggregated and transformed into pricing recommendations.

4. Major Competition-Law Risks

4.1 Algorithmic price coordination

Suppose five competing hotels submit their prices to one pricing platform.

The platform:

1. receives their prices;
2. analyses competitors' prices;
3. recommends a price to each hotel;
4. detects deviations;
5. continuously adjusts recommendations.

Even if the hotels do not directly communicate with each other, the intermediary can potentially create a mechanism through which competitive information is shared.

This is one of the principal issues raised in recent algorithmic-pricing litigation. The U.S. DOJ and FTC have stated that firms cannot use algorithms to accomplish conduct that would be unlawful if carried out by humans.

5. Case Laws

1. Trod Ltd / GB eye Ltd – Amazon Marketplace Pricing Case (UK, 2016)

This is one of the clearest examples of a self-monitoring pricing structure.

Two competing sellers on Amazon Marketplace agreed not to undercut each other. They then used automated repricing software to monitor prices and adjust their own prices accordingly.

The CMA found that the software helped implement the price-fixing arrangement. Trod was fined £163,371, while GB eye received immunity after reporting the cartel and cooperating with the CMA.

Principle

An automated system does not immunise an otherwise unlawful cartel.

The important distinction is:

Independent algorithmic competition → generally legitimate

versus

Agreement + algorithmic implementation → potentially cartel conduct

Significance for self-monitoring markets

The case demonstrates that software can act as an automated enforcement mechanism for an anticompetitive agreement.

6. Eturas UAB v Lietuvos Respublikos konkurencijos taryba, C-74/14 (CJEU, 2016)

Eturas concerned an online travel-booking system in Lithuania.

The administrator of the common electronic platform sent travel agencies a system message concerning a restriction on discounts that could be offered through the platform.

The CJEU examined when participation in a common electronic system can constitute evidence of participation in a concerted practice.

Principle

Digital infrastructure does not eliminate the possibility of a concerted practice.

The relevant question is whether participating undertakings:

• knew about the coordinated mechanism;
• accepted or failed to distance themselves from it in circumstances permitting an inference of participation;
• continued participating in the system.

Importance

Eturas is especially relevant to self-monitoring markets because a common digital infrastructure can itself facilitate coordinated conduct.

It demonstrates that competition law must examine the architecture through which firms interact, rather than merely looking for traditional meetings or written agreements.

7. Samir Agrawal v. Competition Commission of India & Others (NCLAT, 2020)

This Indian case involved allegations concerning algorithmic pricing by Ola and Uber.

The allegation was that algorithmic pricing restricted the ability of individual drivers to compete independently because the platforms determined fares through their algorithms.

The Competition Commission of India initially found no prima facie case of an agreement or arrangement between the relevant parties, and the NCLAT considered the appeal.

Principle

The existence of algorithmically determined prices does not, by itself, establish a cartel.

Competition law must still establish the legally relevant relationship or agreement necessary under the applicable statutory provision.

Importance

This case provides an important counterpoint to the algorithmic-collusion cases.

It shows that:

Algorithmic price similarity ≠ automatically unlawful coordination.

There must be sufficient evidence connecting the technological system to conduct prohibited by competition law.

8. Google Shopping – European Commission

The Google Shopping case illustrates another dimension of self-monitoring: algorithmic self-preferencing.

The European Commission found that Google systematically favoured its own comparison-shopping service in general search results.

The issue was not merely that Google monitored the market. Rather, the platform's ranking infrastructure could affect the visibility of competing services while favouring Google's own service.

The CMA has subsequently used Google Shopping as an important illustration of how ranking algorithms can facilitate self-preferencing.

Principle

A dominant platform's algorithmic decision-making can constitute an important competitive instrument.

Relevance

A self-monitoring platform may simultaneously be:

• infrastructure provider;
• data collector;
• market intermediary;
• competitor;
• ranking authority.

That combination creates a particularly important competition-law governance problem.

9. Amazon Marketplace – CMA Investigation and Commitments (UK, 2022–2023)

The CMA investigated Amazon's use of third-party seller data and its mechanisms for selecting the offer appearing in the Buy Box.

The CMA was concerned about:

• Amazon Retail's use of non-public seller data;
• the criteria for selecting the Buy Box offer;
• Prime eligibility;
• possible advantages for Amazon's own retail operation. 

Amazon subsequently offered commitments, including restrictions on using non-public third-party seller data and requirements for objectively verifiable and non-discriminatory criteria in selecting the Featured Offer.

An independent monitoring trustee was appointed to oversee compliance.

Principle

Where a platform continuously monitors participants in its ecosystem, governance mechanisms may be required to prevent the platform from converting informational advantages into competitive advantages.

Significance

This is particularly important for self-monitoring market structures because the platform may know:

• what competitors are selling;
• how much they sell;
• their prices;
• their inventory;
• consumer demand;
• their performance.

The platform can potentially use that information to compete against the very businesses dependent on its infrastructure.

10. RealPage Algorithmic Pricing Litigation – United States

The RealPage litigation represents one of the most significant modern applications of competition law to algorithmic self-monitoring.

The U.S. DOJ alleged that landlords provided competitively sensitive information to RealPage, whose software generated rental-pricing recommendations.

The 2024 complaint alleged violations of Sections 1 and 2 of the Sherman Act and focused on the use of competitors' non-public rental information in algorithmic pricing.

Subsequent settlements and proposed settlements involving landlords have included restrictions on:

• use of competitors' competitively sensitive information;
• anticompetitive pricing algorithms;
• participation in competitor meetings;
• use of uncertified third-party pricing algorithms.

Some proposed decrees have also contemplated independent monitoring.

Principle

A common algorithmic intermediary can create competition concerns where it aggregates competitors' sensitive information and translates that information into pricing recommendations.

Governance lesson

Competition compliance may need to extend beyond the participating companies to the architecture of the pricing technology itself.

11. Cornish-Adebiyi v. Caesars Entertainment – Hotel Algorithmic Pricing Litigation

This litigation concerns allegations of algorithmic coordination in hotel-room pricing.

In 2024, the DOJ and FTC filed a joint statement of interest explaining that competitors cannot use an algorithm to achieve conduct that would be unlawful if achieved by direct human coordination. They also emphasised that an unlawful arrangement need not necessarily involve direct competitor-to-competitor communications where an algorithm provider is alleged to act in concert with competitors.

Principle

The legal analysis must examine the economic and organisational function of the algorithm, rather than simply asking whether humans communicated directly.

Relevance

This is especially important for self-monitoring markets because the algorithm may effectively become the mechanism through which market participants observe and react to each other.

12. Flipkart Internet Pvt. Ltd. v. Competition Commission of India (Karnataka High Court, 2021)

The Flipkart litigation concerned allegations involving preferential treatment, preferential listing and other practices on the online marketplace.

The allegations included claims that preferred sellers received advantages and that their products received more favourable placement in search results.

The Karnataka High Court's 2021 proceedings are relevant to the broader question of algorithmic and platform-mediated competition, particularly because digital platforms determine visibility through technological systems.

The CCI's records identify the 23 July 2021 Karnataka decision concerning Flipkart Internet v. CCI.

Principle

Digital marketplace governance can influence competition not only through price but also through:

• ranking;
• visibility;
• preferred status;
• search results;
• access to consumers.

Relevance

In self-monitoring markets, control over the information environment can itself become a source of market power.

13. Self-Monitoring vs. Traditional Market Monitoring

The efficiency benefits are substantial, but the competitive risks can also be amplified.

14. The Four Models of Algorithmic Coordination

Competition scholarship and enforcement increasingly distinguish different mechanisms.

Model 1 – Explicit collusion implemented by algorithms

Competitors agree → software implements agreement

This is essentially traditional cartel conduct using technology.

Trod/GB eye is the classic illustration.

Model 2 – Hub-and-spoke coordination

Competitors → common algorithm/provider → coordinated recommendations

RealPage-related enforcement illustrates the concern with this model.

Model 3 – Algorithmic monitoring without explicit agreement

Each firm independently deploys software that continuously observes rivals.

The algorithms may react predictably to competitor conduct.

This creates the more difficult question of whether independent algorithmic adaptation can generate coordination without an identifiable agreement.

Model 4 – Autonomous AI coordination

More advanced systems may:

• learn from market behaviour;
• predict competitors;
• modify strategies;
• test prices;
• respond to deviations;
• continuously optimise.

This creates difficult questions about attribution, explainability and proof.

The CMA has recently identified AI-enabled algorithmic collusion as an emerging competition issue and has emphasised the need for businesses to manage the competition-law risks associated with AI systems.

15. Governance Architecture for Self-Monitoring Markets

Competition compliance should increasingly become technological as well as legal.

A. Data governance

Companies should identify:

• what competitor data is collected;
• whether the data is public or confidential;
• whether it is current or historical;
• who can access it;
• whether the data enters an algorithm.

B. Algorithm governance

Companies should maintain:

• documented algorithmic objectives;
• version histories;
• testing protocols;
• approval procedures;
• audit trails;
• change-management systems.

C. Competitor-information firewalls

Sensitive information should be separated from commercial decision-making where appropriate.

For example:

Competitor-sensitive data

↓

Restricted database

↓

Compliance screening

↓

Permitted analytical use

rather than:

Competitor data → automatic pricing engine

16. Human Oversight

Human oversight is particularly important where algorithms:

• determine prices;
• rank competitors;
• allocate customers;
• determine discounts;
• recommend output;
• select suppliers;
• identify competitors for acquisition;
• monitor compliance with pricing policies.

However, merely inserting a human approval step does not automatically cure an antitrust problem.

The question is whether the human genuinely exercises independent commercial judgment.

17. Monitoring Trustee Model

The Amazon investigation demonstrates the usefulness of an independent monitoring trustee.

The CMA accepted commitments requiring Amazon to appoint an independent trustee to monitor compliance.

A monitoring trustee can:

1. inspect algorithmic systems;
2. test compliance;
3. review data access;
4. examine ranking criteria;
5. receive complaints;
6. report breaches to the authority.

This model can be particularly useful where ordinary periodic regulatory supervision cannot capture continuously changing digital systems.

18. Competition-by-Design

A major emerging principle is competition-by-design.

Instead of waiting until an algorithm produces an anticompetitive outcome, firms can incorporate competition safeguards into the system from the beginning.

Example

Before deployment:

Algorithm design

↓

Competition-law risk assessment

↓

Data classification

↓

Sensitive-information controls

↓

Simulation

↓

Independent validation

↓

Deployment

↓

Continuous audit

This approach is more appropriate for self-monitoring systems than traditional once-a-year compliance reviews.

19. Algorithmic Audit

An algorithmic competition audit should ask:

Input

• What information enters the system?
• Does it include competitor information?

Processing

• Does the system compare competitors?
• Does it predict competitor responses?
• Does it optimise against competitors?

Output

• Does it recommend prices?
• Does it recommend exclusion?
• Does it favour particular sellers?

Feedback

• Does the system learn from competitors' responses?
• Does it automatically modify future decisions?

Governance

• Who approved the system?
• Who can modify it?
• Are changes logged?
• Is there independent review?

20. Self-Monitoring and Market Transparency

Transparency has a dual character.

Beneficial transparency

Consumers can:

• compare prices;
• identify cheaper alternatives;
• compare quality;
• search efficiently.

Anticompetitive transparency

Competitors can:

• observe prices instantly;
• detect deviations;
• retaliate quickly;
• converge on common strategies.

Therefore:

More transparency does not necessarily mean more competition.

The competitive effect depends on who receives the information, its granularity, its frequency and how it is used.

The U.S. DOJ's merger guidelines expressly recognise that increased observability can make coordination easier, including through pricing algorithms and surveillance tools.

21. Self-Monitoring and Dominant Platforms

The problem becomes more complex when the monitoring entity is itself dominant.

Consider:

Platform

→ monitors sellers
→ controls ranking
→ controls consumer access
→ collects seller data
→ operates competing products
→ determines platform rules

This produces a potential multi-role conflict.

The platform is simultaneously:

1. regulator of its ecosystem;
2. infrastructure provider;
3. data collector;
4. market participant;
5. competitor.

Amazon's UK Marketplace investigation illustrates why this combination can attract competition-law scrutiny.

22. Remedies

Competition authorities may use several remedies.

Structural remedies

• separation of platform and competing business;
• divestiture;
• data separation.

Behavioural remedies

• prohibition on use of competitor data;
• non-discriminatory ranking;
• data-access restrictions;
• algorithmic transparency;
• independent monitoring.

Technological remedies

• algorithm certification;
• audit logs;
• controlled data inputs;
• independent testing;
• automated compliance alerts.

Recent U.S. algorithmic-pricing settlements demonstrate the increasing use of highly specific technological and monitoring obligations rather than relying solely on traditional prohibitions.

23. Evidentiary Problems

Self-monitoring markets create difficult evidentiary questions.

Traditional evidence

• emails;
• meetings;
• contracts;
• telephone calls;
• written agreements.

Algorithmic evidence

• source code;
• model architecture;
• training data;
• API logs;
• system prompts;
• pricing histories;
• version histories;
• machine-generated recommendations;
• audit logs;
• internal model documentation.

The competition authority may therefore need technical forensic expertise.

24. The Indian Competition-Law Perspective

Under the Competition Act, 2002, self-monitoring systems can potentially implicate:

Section 3

Agreements, arrangements or concerted practices that cause or are likely to cause appreciable adverse effect on competition.

Section 4

Abuse of dominant position, including potentially discriminatory or exclusionary conduct.

Section 19

Investigation and assessment of alleged anti-competitive conduct.

Section 26

Investigation procedure where the CCI forms the necessary opinion regarding prima facie conduct.

The Samir Agrawal/Ola/Uber litigation demonstrates the importance of establishing the statutory ingredients of coordination rather than treating algorithmic pricing itself as sufficient proof of infringement.

25. Compliance Framework

A firm operating a self-monitoring system should adopt the following framework:

Stage 1 – Identify

Identify every algorithm that:

• observes competitors;
• sets prices;
• recommends prices;
• ranks market participants;
• allocates customers.

Stage 2 – Classify

Classify data as:

• public;
• aggregated;
• historical;
• commercially sensitive;
• competitively sensitive.

Stage 3 – Test

Test whether the algorithm could:

• coordinate prices;
• retaliate against competitors;
• discriminate between sellers;
• facilitate information exchange;
• self-preference.

Stage 4 – Control

Introduce:

• access restrictions;
• data firewalls;
• approval requirements;
• audit logs;
• independent reviews.

Stage 5 – Monitor

Continuously monitor:

• algorithmic outputs;
• price convergence;
• ranking changes;
• competitor-data usage;
• system modifications.

Stage 6 – Escalate

Potentially problematic outcomes should trigger:

algorithmic alert → legal review → technical investigation → corrective action

26. Key Legal Distinction

The central distinction can be represented as follows:

Lawful self-monitoring

Observe market → independently decide → compete

Potentially unlawful self-monitoring

Observe competitor → coordinate → automatically implement → monitor compliance

Potential dominance concern

Collect rival data → algorithmically process → disadvantage rival → favour own business

This distinction is more useful than treating all algorithmic monitoring as either inherently lawful or inherently unlawful.

27. Six Core Principles Emerging from the Case Law

28. Overall Legal Framework

The governance of self-monitoring market structures therefore requires integration of competition law + data governance + algorithm governance + corporate compliance + technological auditing.

The regulatory model can be expressed as:

Market monitoring

↓

Data collection

↓

Algorithmic processing

↓

Competitive decision

↓

Market response

↓

Feedback

↓

Continuous monitoring

At every stage, competition law should ask:

Does the system preserve independent competitive decision-making, or does it facilitate coordination or exclusion?

Conclusion

Self-monitoring market structures represent an important transformation in modern competition.

The technology itself is generally competition-neutral. Automated observation can create substantial efficiencies, but the same infrastructure can also increase market observability, facilitate coordination, transmit commercially sensitive information, or enable a dominant platform to exploit informational advantages.

The principal lessons from Trod/GB eye, Eturas, Samir Agrawal, Google Shopping, Amazon Marketplace, RealPage and Cornish-Adebiyi are that competition law increasingly looks beyond traditional human-to-human communication and examines the architecture, data flows and practical operation of automated decision systems.

The appropriate regulatory objective is therefore not to prohibit self-monitoring markets as such, but to ensure that:

independent decision-making + controlled data + auditable algorithms + effective oversight = competitive self-monitoring

Where those safeguards disappear, an apparently autonomous market structure can become a mechanism for algorithmic coordination, information exchange or exclusionary conduct. Recent enforcement developments in the United States and United Kingdom show that competition authorities are increasingly combining substantive antitrust rules with monitoring, algorithmic controls and compliance mechanisms to address these risks.