Competition Law and Collaborative Forecasting Agreements and Competition

1. Introduction

Collaborative forecasting agreements arise when two or more businesses—particularly actual or potential competitors—jointly exchange, develop, or rely upon forecasts concerning future market conditions, such as:

• future prices;
• expected demand and sales;
• production volumes;
• capacity;
• inventory;
• costs;
• market growth;
• customer demand;
• supply shortages;
• future discounts or surcharges;
• investment plans;
• output expectations; or
• forecasts generated through a common algorithm, platform, trade association, or data intermediary.

Forecasting itself is not inherently unlawful. Businesses may legitimately collaborate on forecasts for joint ventures, production planning, R&D, logistics, sustainability projects, capacity planning, or supply-chain resilience. The competition-law problem arises where forecasting becomes a mechanism through which competitors obtain information about each other's future competitive behaviour.

The central question is therefore:

Does the forecasting arrangement merely improve legitimate market knowledge, or does it reduce strategic uncertainty between competitors and facilitate coordination of their competitive conduct?

Modern competition law treats the second situation as potentially serious, particularly where forecasts are individualised, forward-looking, commercially sensitive, reciprocal, frequent and shared in concentrated markets. The EU's current horizontal-cooperation framework expressly recognises that information can be exchanged directly or indirectly through third parties, platforms, algorithms, trade associations and market-research organisations.

2. Meaning of Collaborative Forecasting Agreements

A collaborative forecasting agreement can take several forms.

A. Direct competitor forecasting

Competitors exchange their respective forecasts:

Company A: "We expect demand to fall by 8% next quarter."

Company B: "We expect to reduce production by 10%."

Although apparently predictive, such information can allow competitors to anticipate each other's output decisions.

B. Joint forecasting platform

Several competitors submit confidential information to a common platform which produces forecasts.

The competition issue becomes particularly significant if the platform:

• identifies individual contributors;
• produces competitor-specific forecasts;
• recommends future prices;
• recommends output levels;
• distributes individualised forecasts;
• or enables competitors to observe one another's strategic plans.

C. Trade-association forecasting

An industry association may collect information from competitors and distribute a forecast.

A carefully designed aggregated and historical forecast may be relatively benign. A forecast showing individual companies' expected future prices or output can be much more problematic.

D. Algorithmic forecasting

Competitors may use the same AI or forecasting system trained on their confidential information.

This creates a modern hub-and-spoke problem where the software provider can become an intermediary through which commercially sensitive information is exchanged.

The CMA has specifically recognised that common algorithms or data hubs can facilitate indirect exchanges of competitively sensitive information, including pricing plans, future strategies, stock levels and spare capacity.

3. Applicable Competition-Law Framework

A. European Union

The principal provision is Article 101 TFEU.

A forecasting arrangement may fall within Article 101(1) where it constitutes:

1. an agreement;
2. a decision by an association of undertakings; or
3. a concerted practice,

which has the object or effect of restricting competition.

The European Commission's 2023 Horizontal Guidelines specifically address information exchanges. They distinguish legitimate information sharing from exchanges that can eliminate strategic uncertainty and facilitate coordination.

Most importantly, the Guidelines identify exchanges of individualised information concerning intended future prices or quantities as particularly serious.

B. United Kingdom

The equivalent framework is primarily the Chapter I prohibition of the Competition Act 1998.

UK competition law is concerned with whether information exchange substantially reduces uncertainty concerning the future commercial behaviour of competitors. Government guidance expressly recognises that commercially sensitive information can include information enabling competitors to understand future conduct.

C. India

In India, the principal provisions are:

• Section 3(1), Competition Act 2002 — prohibition of anti-competitive agreements;
• Section 3(3) — agreements between enterprises engaged in identical or similar trade which have specified anti-competitive objects/effects;
• Section 3(3)(a) — direct or indirect determination of purchase or selling prices;
• Section 3(3)(b) — limitation or control of production, supply or markets;
• Section 3(3)(c) — market sharing;
• Section 3(3)(d) — bid rigging/collusive bidding.

A forecasting exchange may become problematic when it is evidence of, or facilitates, coordination concerning price, output, supply, capacity or market allocation.

4. Why Future Forecasts Are Particularly Sensitive

Competition normally depends upon competitors making decisions under uncertainty.

For example:

Independent competition

A does not know what B will charge → A independently determines price → B responds independently.

Forecast-sharing

A knows B's expected price → B knows A's expected price → uncertainty falls → strategic responses become easier → competitive rivalry may weaken.

This is why future information is generally more sensitive than historical information.

The CJEU has recently reiterated that information concerning future prices, or factors determining future prices, is inherently problematic because of its capacity to reduce uncertainty regarding competitors' future behaviour.

5. Factors Used to Assess Collaborative Forecasting

5.1 Nature of information

The more commercially sensitive the information, the greater the competition concern.

Higher risk

• future prices;
• future discounts;
• future output;
• future capacity;
• future production cuts;
• planned investment;
• future customer allocations;
• future strategic responses.

Lower risk

• old historical data;
• publicly available information;
• highly aggregated industry statistics;
• information that cannot identify individual competitors.

5.2 Individualised versus aggregated data

This is one of the most important distinctions.

Individualised

Company A expects price = ₹100
Company B expects price = ₹105
Company C expects price = ₹102

This permits competitors to predict one another's behaviour.

Aggregated

Industry expected average price next year = ₹102.

The latter may substantially reduce the risk of coordination because individual competitive intentions cannot easily be identified.

The EU Guidelines specifically observe that aggregated information may achieve legitimate benchmarking objectives with less risk than individualised information.

6. Six Important Case Laws

1. T-Mobile Netherlands BV and Others v Raad van bestuur van de Nederlandse Mededingingsautoriteit

Case C-8/08, CJEU, 2009

This is one of the leading cases on information exchange and concerted practices.

Mobile-network competitors participated in a meeting where commercially sensitive information concerning pricing behaviour was discussed.

Principle

The Court held that a single meeting can be sufficient to constitute a concerted practice where the information exchanged has an anti-competitive object.

The law does not necessarily require:

• repeated meetings;
• a written cartel;
• a long-term agreement; or
• continuous communications.

The case is highly relevant to forecasting because a single exchange of strategically important future information can potentially affect competitors' subsequent market behaviour.

Relevance to forecasting

If competitors meet once to exchange:

• next-quarter price forecasts;
• expected production reductions; or
• planned capacity,

the absence of a continuing formal forecasting agreement does not automatically eliminate competition-law risk.

2. John Deere Ltd v Commission / UK Agricultural Tractor Registration Exchange

Case C-7/95 P and related proceedings

The agricultural tractor industry operated an information-exchange system involving sales and registration information.

The Commission found problems where competitors could obtain detailed information identifying competitors' sales and market positions. The General Court and CJEU proceedings established the importance of assessing the structure and characteristics of information exchanges.

The original Commission decision held that the exchange of information identifying individual competitors' sales and dealer/import information infringed the competition rules.

Principle

Information does not need to contain an explicit price agreement to create competition concerns.

The degree of transparency created among competitors can itself weaken competitive uncertainty.

Relevance to forecasting

A forecasting consortium that publishes:

• Company A's expected output;
• Company B's expected sales;
• Company C's anticipated market share;

may create a similar transparency problem.

3. ASNEF-EQUIFAX v Ausbanc

Case C-238/05, CJEU, 2006

This case concerned a system for exchanging creditworthiness information between financial institutions.

The Court recognised that information-sharing systems can have legitimate economic functions, including improving information available to lenders.

Principle

Information exchange is not automatically anti-competitive.

Its competitive assessment depends upon:

• the characteristics of the information;
• market structure;
• accessibility of information;
• effects on competitive uncertainty;
• and the overall economic context.

Relevance to forecasting

This is particularly important for collaborative forecasting.

A forecasting arrangement designed to address:

• information asymmetry;
• demand forecasting;
• supply-chain planning;
• credit risk;
• capacity utilisation;

may have legitimate efficiencies.

The legal question is whether the arrangement goes beyond what is necessary for those efficiencies.

4. Eturas UAB and Others v Lietuvos Respublikos konkurencijos taryba

Case C-74/14, CJEU, 2016

Travel agencies used a common computerised booking system. The system administrator sent a communication concerning a restriction on discounts available through the platform.

The case dealt with whether the common technological infrastructure and the communication could support an inference of a concerted practice.

Principle

Competition law can apply to indirect technological coordination.

A common digital platform does not become competition-law neutral merely because competitors do not communicate directly with one another.

Relevance to forecasting

Suppose competing retailers use a common AI forecasting platform.

If the platform:

1. receives their confidential forecasts;
2. processes the information;
3. produces competitor-sensitive predictions; and
4. distributes those predictions to the participants,

the platform may create a hub-and-spoke information-exchange risk.

5. Dole Food Company Inc. v European Commission

Case C-286/13 P, CJEU, 2015

The case concerned exchanges of information relating to the banana market, including information relevant to pricing.

The Court examined whether the communications could reduce uncertainty about competitors' future market conduct.

Principle

Information can be strategically sensitive even where it does not expressly state a final price.

Information may be problematic if it enables competitors to understand or predict:

• future pricing;
• pricing intentions;
• commercial strategy; or
• other competitive parameters.

The concept of strategic information therefore extends beyond a simple exchange of final prices.

This principle is particularly important for forecasting arrangements because a forecast may effectively communicate a company's future competitive strategy without stating a precise price.

6. Banco BPN/BIC Português SA and Others v Autoridade da Concorrência

Case C-298/22, CJEU, 29 July 2024

This is especially important for modern collaborative forecasting.

Several Portuguese credit institutions exchanged information concerning:

• current and future credit spreads;
• risk variables;
• individual production figures;
• commercial conditions.

The exchange was regular, organised, confidential, individualised and concerned current and future conduct. The six largest institutions represented a very substantial portion of the Portuguese banking sector.

The CJEU dealt directly with whether such an information exchange could constitute a restriction of competition by object.

The Court emphasised that information relating to future prices or factors determining those prices can be inherently problematic because it reduces uncertainty about future competitive behaviour.

Relevance to forecasting

This case is perhaps the closest modern judicial illustration of the forecasting problem.

A collaborative forecast containing:

• future prices;
• future margins;
• future risk variables;
• expected production;
• individual company data;

can move from legitimate information sharing toward an anti-competitive exchange.

7. Additional Important Authority: Container Shipping

Although not a judicial case, European Commission Case AT.39850 — Container Shipping is highly relevant.

Fourteen liner shipping companies regularly announced intended future price increases through public communications.

The Commission investigated whether these announcements could allow carriers to signal their future pricing intentions to competitors and ultimately accepted commitments addressing the concern.

Importance

The case demonstrates that the competition concern does not necessarily require a secret private meeting.

Even public communications can raise issues where they effectively transmit detailed future pricing signals to competitors.

8. Collaborative Forecasting Through a Third Party

A particularly important modern scenario is:

Competitor A
↓
Forecasting Platform / AI Provider
↓
Competitor B

The competitors may argue:

"We never communicated directly."

That is not necessarily sufficient.

EU guidance expressly recognises direct and indirect information exchange through:

• service providers;
• platforms;
• online tools;
• algorithms;
• common agencies;
• market-research organisations;
• suppliers; and
• customers. 

Thus, a third-party forecasting intermediary cannot automatically eliminate competition-law risk.

9. AI-Based Collaborative Forecasting

The problem becomes more sophisticated when AI is involved.

Imagine five competing airlines provide an AI system with:

• expected passenger demand;
• future seat capacity;
• intended fares;
• fuel-cost assumptions;
• planned capacity reductions.

The AI generates:

"Expected market equilibrium price: ₹8,500."

If every competitor receives information that reflects the other participants' confidential inputs, the system could reduce strategic uncertainty.

Competition risks include:

1. common algorithmic coordination;
2. exchange of future pricing information;
3. competitor-specific recommendations;
4. automated monitoring;
5. punishment of deviations;
6. hub-and-spoke coordination;
7. collective optimisation of prices or output.

The CMA has specifically identified common algorithms and data hubs as potential mechanisms for indirect exchange of sensitive information.

10. Legitimate Forecasting vs Anti-Competitive Forecasting

11. When Forecasting May Produce Efficiencies

Competition law should not be interpreted as prohibiting all collaborative forecasting.

Forecasting can generate legitimate efficiencies by:

A. Reducing supply-chain waste

Companies can predict:

• demand;
• transportation requirements;
• inventory;
• production requirements.

B. Supporting innovation

Joint forecasting can allow firms to determine whether a new technology will have sufficient market demand.

C. Improving capacity utilisation

Industries with high fixed costs may use aggregated forecasts to avoid excessive capacity.

D. Sustainability

Companies may jointly forecast:

• renewable-energy demand;
• carbon intensity;
• recycling requirements;
• electric-vehicle charging demand.

The EU Guidelines recognise that data sharing can generate legitimate efficiencies and facilitate new products, services and technologies.

12. Article 101(3) / Efficiency Defence

Even where an information exchange restricts competition, an undertaking may potentially seek to establish an efficiency justification under Article 101(3) TFEU, subject to its conditions.

The key requirements include:

1. efficiency gains;
2. consumer benefit;
3. indispensability; and
4. no elimination of competition.

The EU Guidelines stress that information exchanged for benchmarking should generally use the least restrictive form of data, and aggregated information may often achieve the objective with less competitive risk.

13. Competition-Law Safe Design of Forecasting Agreements

A compliant forecasting system should generally consider the following safeguards.

1. Use historical information where possible

Prefer:

Previous year's aggregate industry demand

over:

Competitor B's expected demand next month.

2. Aggregate data

Use:

Industry-wide forecast

rather than:

Company-specific forecast.

3. Anonymise contributors

Participants should not be able to identify which competitor supplied particular information.

4. Use an independent administrator

A genuinely independent data intermediary can reduce direct competitor-to-competitor contact.

5. Avoid future individualised prices

This is particularly important.

6. Establish information firewalls

Commercial teams should not automatically receive competitor-sensitive information.

7. Restrict access

Only personnel genuinely required for the legitimate forecasting project should receive the outputs.

8. Prohibit competitor-specific recommendations

The system should not tell competitors:

"Company B plans to reduce output by 15%."

9. Conduct competition-law audits

The forecasting model should periodically be reviewed for:

• data inputs;
• outputs;
• access rights;
• algorithmic functionality;
• communications;
• retention periods.

10. Document the legitimate objective

The agreement should identify precisely why forecasting is required and why the information exchanged is necessary.

14. Special Risks in Different Industries

Banking

High risk because forecasts can involve:

• lending rates;
• spreads;
• risk pricing;
• loan volumes.

The Banco BPN/BIC case demonstrates the sensitivity of current and future lending conditions.

Energy

Potentially sensitive information includes:

• future generation;
• capacity;
• outages;
• wholesale prices;
• supply expectations.

Airlines

Forecasting can concern:

• future fares;
• seat capacity;
• passenger demand;
• route reductions.

Shipping

Future freight rates and capacity forecasts can facilitate strategic coordination.

Retail

Competitors may exchange:

• expected promotions;
• inventory forecasts;
• future prices;
• demand forecasts.

Digital platforms

Forecasting may be conducted through a shared:

• API;
• cloud platform;
• AI model;
• data marketplace;
• pricing engine.

This introduces additional hub-and-spoke and algorithmic risks.

15. India-Specific Analysis

Under Section 3 of the Competition Act 2002, the central concern is whether collaborative forecasting becomes a means of coordinating competitive parameters.

For example:

Legitimate

Five manufacturers provide anonymised historical production data to an independent consultant, who publishes:

"Total industry demand is expected to grow by 5%."

Higher risk

The same five manufacturers receive:

Manufacturer A intends to produce 80,000 units next quarter.
Manufacturer B intends to produce 75,000 units.
Manufacturer C intends to increase prices by 7%.

The second arrangement can potentially facilitate coordination concerning price, production and market conditions.

Indian competition analysis also recognises the broader concept of concerted behaviour, including situations where parallel conduct is accompanied by evidence of coordination. The Supreme Court has discussed the relevance of the EU Dyestuffs principles in Indian competition jurisprudence.

16. Relationship with Cartels

Collaborative forecasting can exist on a spectrum:

Legitimate forecasting

→ Aggregated market research

→ Confidential benchmarking

→ Competitor information exchange

→ Forward-looking strategic information exchange

→ Price signalling

→ Coordinated output/pricing

→ Cartel

The important point is that forecasting does not become unlawful merely because competitors participate.

The legal concern becomes much stronger when the forecasting mechanism is used to reduce uncertainty about future competitive decisions.

17. Key Legal Tests

When analysing a collaborative forecasting agreement, ask:

Test 1 — Who are the participants?

Are they:

• actual competitors?
• potential competitors?
• vertically related businesses?
• independent businesses participating in a joint project?

Test 2 — What information is exchanged?

Is it:

• historical?
• current?
• future?
• public?
• confidential?
• strategic?

Test 3 — Is it individualised?

Can Company A determine what Company B intends to do?

Test 4 — How frequently is information exchanged?

A one-off exchange may sometimes be sufficient, as T-Mobile demonstrates.

Test 5 — How concentrated is the market?

The fewer the major competitors, the greater the potential importance of strategic information.

Test 6 — Is there a legitimate efficiency?

Can the forecasting objective be achieved without exchanging individualised future information?

Test 7 — Is the exchange indispensable?

Could the parties achieve the same purpose through:

• anonymisation;
• aggregation;
• historical data;
• independent statistical analysis?

Test 8 — What does the forecasting system output?

This is crucial.

A system producing:

"Industry demand will grow 5%"

is fundamentally different from:

"Competitor X will increase price 8% next month."

18. Case-Law Principles at a Glance

19. Conclusion

Collaborative forecasting agreements occupy an important boundary between legitimate commercial cooperation and prohibited competitor coordination.

The decisive issue is not simply whether companies "share forecasts." Competition authorities and courts are more likely to examine:

• what is forecast;
• whether the information is future-oriented;
• whether it is individualised;
• whether competitors can identify one another's intentions;
• how frequently information is exchanged;
• whether the market is concentrated;
• whether an intermediary or algorithm is involved;
• whether the exchange reduces strategic uncertainty; and
• whether legitimate efficiencies can be achieved through less restrictive means.

The strongest competition-law danger arises where competitors use a collaborative forecasting mechanism to obtain confidential, individualised information about each other's future prices, output, capacity or strategic decisions. The 2024 Banco BPN/BIC judgment is particularly significant because the CJEU directly considered regular exchanges involving current and future commercial conditions and individualised production information.

Conversely, independent administration, aggregation, anonymisation, historical data, restricted access and demonstrable efficiency objectives can materially reduce the competition risk.

Thus, the governing principle can be summarised as:

Competition law generally permits forecasting that improves legitimate market efficiency, but becomes increasingly concerned when collaborative forecasting transforms competitors' uncertainty about future conduct into shared strategic knowledge.