Algorithmic Pricing and Anti-Competitive Conduct

1. Introduction

Algorithmic pricing refers to the use of computer programmes, artificial intelligence and machine-learning systems to determine or adjust prices automatically. Businesses may use algorithms to:

monitor competitors’ prices;

forecast demand;

calculate discounts;

manage inventory;

personalise offers;

respond to changes in costs; and

maximise revenue.

Algorithmic pricing can produce legitimate efficiencies. However, it may also facilitate anti-competitive conduct by enabling businesses to coordinate prices, exchange sensitive information, punish discounting, discriminate against customers or exclude competitors.

The central competition-law issue is whether the algorithm is merely an independent business tool or whether it implements, facilitates or creates an unlawful restriction of competition.

2. Forms of Algorithmic Pricing

A. Rule-based pricing

A business establishes fixed rules, such as:

“Set the price 5% below the lowest competitor.”

This may be lawful if independently adopted. However, it may become problematic if competitors use a common system or agree to follow the same rule.

B. Dynamic pricing

Prices change according to:

demand;

supply;

time;

location;

inventory;

consumer behaviour; and

competitor prices.

Dynamic pricing is not inherently unlawful. It becomes problematic where it is used to coordinate prices or exploit market power.

C. Personalised pricing

Algorithms may charge different customers different prices based on:

purchasing history;

location;

browsing activity;

willingness to pay;

device type; or

urgency.

Personalisation may improve efficiency, but it can raise concerns where it involves discrimination, exploitation or exclusion.

D. Predictive pricing

AI systems may predict how competitors will react to price changes. If several firms use similar systems, the algorithms may learn to avoid price competition.

E. Platform-based pricing

An online platform may recommend or impose prices for independent sellers, hotels, drivers, restaurants or retailers. The platform may thereby become a central coordinator of market prices.

3. Anti-Competitive Effects

A. Price fixing

Competitors may use algorithms to implement an agreement to maintain or increase prices.

Example

Two online sellers agree that their algorithms will never price below a specified amount. The software implements the agreement automatically.

The conduct remains price fixing even though no employee manually changes each price.

B. Tacit coordination

Algorithms can rapidly observe competitors and respond to price changes. They may learn that avoiding price cuts is more profitable than competing aggressively.

This can result in:

stable high prices;

reduced discounting;

reduced output; and

weaker competition.

However, similar prices alone do not necessarily prove unlawful coordination.

C. Price signalling

A business may use public algorithmic announcements to communicate future pricing intentions. Competitors can then respond without direct private communication.

Repeated publication of future prices may reduce strategic uncertainty and facilitate coordination.

D. Hub-and-spoke coordination

A common platform or software provider may transmit information or pricing recommendations to competing businesses.

The platform may act as the “hub,” while the participating businesses operate as the “spokes.”

E. Predatory pricing

A dominant business may use an algorithm to set prices below cost in order to eliminate competitors and subsequently raise prices.

Algorithmic losses may be difficult to detect because the system can target particular geographic areas, products or competitors.

F. Margin squeezing

A vertically integrated platform may charge high wholesale or access prices while using its downstream algorithm to offer artificially low retail prices.

This may make it impossible for independent competitors to compete effectively.

G. Discriminatory pricing

A dominant platform may charge different prices or provide different terms to similarly situated customers or business partners.

Discrimination may be unlawful where it:

exploits consumers;

disadvantages competitors;

favours affiliated businesses; or

is imposed by a dominant enterprise without objective justification.

H. Self-preferencing

A platform may use algorithmic rankings to favour its own products or services over competing suppliers.

Examples include:

ranking the platform’s own products first;

giving affiliated sellers better visibility;

restricting competitors’ access to data; or

using rivals’ information to compete against them.

4. Indian Legal Framework

A. Section 3 of the Competition Act, 2002

Section 3 prohibits agreements which cause or are likely to cause an appreciable adverse effect on competition.

Algorithmic pricing may fall within Section 3 where it facilitates:

price fixing;

output limitation;

market allocation;

customer allocation;

bid rigging;

exchange of commercially sensitive information; or

coordinated refusal to supply.

Section 3 applies to agreements that may be written, oral, formal, informal or inferred from conduct.

B. Section 3(3)

Section 3(3) is especially relevant to horizontal arrangements between competitors. It addresses agreements involving:

directly or indirectly determining purchase or sale prices;

limiting or controlling production, supply or markets;

sharing markets or customers; and

bid rigging or collusive bidding.

The use of software does not change the legal character of the underlying agreement.

C. Section 4: Abuse of dominant position

A dominant enterprise may breach Section 4 by using algorithmic pricing to:

impose unfair or discriminatory prices;

engage in predatory pricing;

deny market access;

restrict production or technical development;

tie products or services;

favour affiliated businesses; or

impose discriminatory conditions.

D. Sections 5 and 6: Combinations

Acquisitions involving pricing platforms, data providers, artificial-intelligence companies or online marketplaces may raise combination concerns.

The Competition Commission of India may consider whether the transaction will give the acquirer control over:

pricing infrastructure;

consumer data;

competitor information;

digital advertising;

cloud computing; or

distribution channels.

E. Sections 19 and 20

The Commission may investigate algorithmic pricing by examining:

pricing records;

software contracts;

model instructions;

internal communications;

source-code documentation;

data-sharing arrangements;

market outcomes; and

evidence of price coordination.

5. Important Case Laws

1. United States v Topkins, 2015

Facts

Online sellers of posters agreed to use pricing algorithms to maintain prices and avoid undercutting each other. The software automatically implemented the price-fixing arrangement.

Decision

The defendant pleaded guilty to criminal price-fixing charges.

Principle

An algorithm is not a defence to cartel liability. Where businesses agree to fix prices, the agreement remains unlawful even if software performs the actual pricing.

Relevance

The case establishes that competition authorities should investigate:

who designed the pricing rule;

whether competitors agreed to use it;

whether the system prevented price reductions; and

whether employees knew the algorithm was implementing coordination.

2. Eturas v Lietuvos Respublikos konkurencijos taryba, Case C-74/14, 2016

Facts

Travel agencies used a common online booking platform. The platform administrator sent an electronic message limiting the discounts that agencies could provide.

The platform technically prevented discounts beyond the authorised level.

Decision

The Court of Justice held that participation in the system could support an inference of concerted practice where the agencies knew or should have known about the anti-competitive message and continued using the platform.

Principle

Electronic systems can facilitate coordination between competitors. A business may be responsible for participating in a digital arrangement even where the restriction is implemented technically by a platform.

Relevance

The case is directly relevant to:

hotel pricing software;

airline reservation systems;

travel platforms;

online marketplaces; and

common discount-management tools.

3. United States v Airline Tariff Publishing Co., 1994

Facts

Airlines used a computerised fare-publication system to announce and monitor fare changes. The system enabled airlines to observe competitors’ proposed prices and respond to them.

The authorities alleged that the system facilitated coordination of airline fares.

Resolution

The matter was resolved through an antitrust settlement.

Principle

Public price announcements may become anti-competitive when they communicate future pricing intentions and reduce uncertainty between competitors.

Relevance

Modern algorithmic pricing systems can make such coordination faster and more effective by:

monitoring fares in real time;

automatically matching price increases;

detecting discounting;

signalling future prices; and

discouraging competitive price reductions.

4. T-Mobile Netherlands BV v Raad van bestuur van de Nederlandse Mededingingsautoriteit, Case C-8/08, 2009

Facts

Mobile telecommunications operators exchanged commercially sensitive information during a meeting concerning dealer remuneration and other competitive conditions.

Decision

The Court of Justice held that a single meeting could constitute a concerted practice where it was capable of reducing strategic uncertainty.

Principle

Competition law protects the competitive uncertainty that exists between independent firms. An unlawful concerted practice does not always require a detailed written price-fixing agreement.

Relevance

Algorithmic pricing may reduce uncertainty through the exchange of:

future price intentions;

inventory data;

demand forecasts;

capacity plans;

customer information; and

expected reactions to competitors’ strategies.

5. United States v Apple Inc., 791 F.3d 290 (2d Cir. 2015)

Facts

Apple was accused of facilitating coordination among major book publishers to increase electronic-book prices. Contractual arrangements changed the competitive structure of the market and enabled coordinated price increases.

Decision

The Second Circuit upheld the finding of liability against Apple.

Principle

A technology company may be liable where it knowingly facilitates coordination among suppliers or competitors, even if it does not directly set every price.

Relevance

The case applies to digital pricing platforms where a technology provider:

designs the pricing architecture;

encourages uniform prices;

shares competitors’ information;

monitors compliance; or

structures contracts to reduce price competition.

6. Competition Commission v Steel Authority of India Ltd., (2010) 10 SCC 744

Facts

The Supreme Court examined the powers and procedure of the Competition Commission of India under the Competition Act, 2002.

Principle

The Commission possesses statutory investigative powers, subject to the legal safeguards governing competition proceedings.

Relevance

In algorithmic-pricing cases, the Commission may need to examine:

pricing data;

algorithmic instructions;

internal emails;

platform agreements;

communications with software providers;

model-training records; and

the relationship between human managers and automated systems.

The case is relevant to the procedural foundation for investigating technologically complex conduct.

7. Excel Crop Care Ltd v Competition Commission of India, (2017) 8 SCC 47

Facts

The case concerned cartel conduct in the market for aluminium phosphide tablets. The Supreme Court examined cartel liability and the calculation of penalties.

Principle

Competition-law penalties should be connected to the relevant anti-competitive conduct and the affected market.

Relevance

Where an enterprise uses algorithms across several markets, a penalty for algorithmic pricing conduct should consider:

the relevant product;

the affected geographic market;

the duration of the infringement;

the enterprise’s role; and

the relevant turnover.

The case also emphasises the importance of economic evidence in competition proceedings.

8. Google Shopping, European Commission Decision, 2017

Facts

Google was found to have favoured its own comparison-shopping service in search results and to have demoted competing services.

Principle

A dominant digital platform may abuse its position by using an important algorithmic system to favour its own services and disadvantage rivals.

Relevance

Although the case was not principally about price fixing, it is highly relevant to algorithmic anti-competitive conduct. It shows how algorithms may be used to:

favour affiliated products;

restrict visibility of rivals;

exploit data obtained from competitors; and

control access to customers.

6. Difference Between Lawful Algorithmic Pricing and Illegal Conduct

|
Lawful algorithmic pricing

|

Potentially anti-competitive algorithmic pricing

|
| --- | --- |
|

Prices respond to genuine demand and supply

|

Prices are coordinated between competitors

|
|

Algorithm is independently developed

|

Competitors use a common anti-competitive rule

|
|

Competitors’ public prices are observed normally

|

Future pricing intentions are exchanged

|
|

Discounts remain independently determined

|

Algorithms punish competitors for discounting

|
|

Prices vary according to costs and inventory

|

Prices remain artificially high despite changing conditions

|
|

Platform provides neutral technical services

|

Platform imposes or facilitates uniform prices

|

The essential issue is not whether prices are similar, but why they are similar and how the similarity was achieved.

7. Evidence Required to Establish Liability

Relevant evidence may include:

Communications between competitors.

Agreements with common software providers.

Source-code instructions.

Pricing rules and algorithmic objectives.

Records of price changes.

Evidence of automatic retaliation against discounting.

Data-sharing arrangements.

Internal compliance warnings.

Evidence of abnormal price stability.

Changes in margins and output.

Instructions given to software developers.

Proof that management knew or should have known the algorithm’s effects.

Economic evidence may examine:

price correlation;

price dispersion;

margins;

output levels;

demand elasticity;

market entry;

frequency of price changes; and

whether prices respond normally to costs.

8. Liability of Different Participants

A. The business using the algorithm

A business may be liable where it:

approved the pricing objective;

entered into an agreement;

knowingly used a common pricing system;

ignored warning signs; or

benefited from unlawful coordination.

B. The software provider

A software provider may face liability where it knowingly designs or operates a system to facilitate cartel conduct.

However, providing ordinary neutral software without knowledge of anti-competitive use should not automatically create liability.

C. The platform operator

A platform may be responsible where it:

controls pricing rules;

communicates competitors’ information;

monitors compliance;

imposes uniform prices; or

uses its position to exclude rival sellers.

D. Human managers

Human involvement may be relevant at the design, approval, implementation or monitoring stage. A business cannot necessarily avoid responsibility by claiming that the final decision was made autonomously by an algorithm.

9. Defences

A business may argue that:

the algorithm was independently developed;

price similarities resulted from common costs;

no confidential information was exchanged;

the software provider acted independently;

the conduct generated consumer benefits;

the business did not know of the algorithm’s anti-competitive effects; or

competitors remained free to set their own prices.

These defences must be supported by documents, audit trails and technical evidence.

10. Compliance Measures

Businesses should adopt the following safeguards:

Conduct competition-law assessments before deploying pricing algorithms.

Prohibit exchange of future pricing intentions.

Avoid common pricing rules among competitors.

Maintain records of algorithm design and modifications.

Separate competitors’ confidential data.

Test whether the algorithm rewards price matching or market stability.

Monitor automatic responses to competitors’ discounts.

Require human review of high-risk pricing decisions.

Audit contracts with third-party pricing providers.

Train legal, commercial and technical staff.

Preserve logs and model documentation.

Establish procedures for reporting suspected coordination.

11. Critical Evaluation

Algorithmic pricing presents a difficult balance.

Benefits

better inventory management;

lower search costs;

efficient allocation of goods;

personalised discounts;

faster response to demand;

reduced waste; and

improved consumer convenience.

Risks

rapid cartel implementation;

increased price transparency for competitors;

automated retaliation;

discriminatory prices;

exclusion of small businesses;

predatory pricing;

self-preferencing; and

difficulty in identifying responsibility.

Regulators should not presume that every use of AI or every instance of price matching is unlawful. The investigation must establish the relevant market, the actual operation of the algorithm, the presence of coordination or dominance, and the effect on competition.

Conclusion

Algorithmic pricing is not inherently anti-competitive. It may promote efficiency, innovation and consumer welfare. Nevertheless, it becomes legally problematic where it implements price fixing, facilitates information exchange, supports tacit coordination, enables predatory pricing or allows a dominant platform to discriminate against competitors.

The decisions in United States v Topkins, Eturas, Airline Tariff Publishing, T-Mobile Netherlands, United States v Apple, SAIL, Excel Crop Care and Google Shopping demonstrate that competition law focuses on the economic substance of conduct rather than whether a human or a computer made the final pricing decision.

The appropriate legal approach is therefore technologically neutral but evidence-based. Businesses must maintain effective algorithmic compliance systems, while competition authorities must distinguish genuine independent pricing from coordinated or exclusionary conduct.