Competition Act, 2002
Horizontal Agreements and Cartels
A horizontal agreement is one between enterprises at the same level of the production chain, which is to say between competitors. Section 3(3) treats four kinds of such agreement as presumptively harmful: price fixing, limiting production or supply, market sharing and bid rigging. Between them they describe what a cartel does. The law on this subject is mostly evidentiary, because a cartel leaves no contract and few documents, and the central question in almost every case is whether what the firms did in the market can only be explained by agreement or can equally be explained by the structure of the market itself.
1. Horizontal Agreements: Essentials
Four elements must be present before Section 3(3) applies. There must be an agreement within the wide definition in Section 2(b), which includes an arrangement, an understanding or an action in concert. The parties must be enterprises, associations of enterprises, persons or associations of persons, and after the amendment of 2023 an enterprise that actively participates in furthering the agreement is included even if it is not at the same level. The parties must be engaged in identical or similar trade of goods or provision of services. And the agreement must fall within one of the four clauses, or, if it does not, it may still be examined under Section 3(1) without the benefit of the presumption.
Section 3(3) applies expressly to a practice carried on, or a decision taken by, an association of enterprises or persons, which is why trade association conduct is dealt with under this sub-section and not separately.
2. Cartel: Meaning and Ingredients
Section 2(c) defines a cartel to include an association of producers, sellers, distributors, traders or service providers who, by agreement amongst themselves, limit, control or attempt to control the production, distribution, sale or price of, or trade in, goods or provision of services. Four ingredients follow: two or more independent enterprises; an agreement amongst themselves; the subject matter being production, distribution, sale, price or trade; and the object of limiting or controlling it, an attempt being sufficient.
The economic description is more useful than the statutory one. A cartel is an arrangement by which competitors coordinate their conduct so as to achieve between them the result a monopolist would achieve alone. It requires the members to agree on what to do, to be able to detect a member who departs from the agreement, and to be able to punish him; without detection and punishment the arrangement decays, because each member gains by secretly undercutting the others. That is why cartels are most stable in markets with few firms, homogeneous products, transparent prices, stable demand and high barriers to entry, and why those market features are examined before an inference is drawn.
3. The Four Kinds of Cartel Conduct
3.1 Price fixing
Section 3(3)(a) covers an agreement that directly or indirectly determines purchase or sale prices. The words directly or indirectly are wide enough to cover agreements on discounts, rebates, credit terms, freight equalisation, a common price list, a formula for calculating price, or a minimum price below which none will sell. The definition of price in Section 2(o), which includes every valuable consideration whether direct or indirect, prevents evasion through a bundled benefit. Purchase prices are covered as well as sale prices, so a buyers' cartel is within the clause.
3.2 Limiting production or supply
Section 3(3)(b) covers an agreement that limits or controls production, supply, markets, technical development, investment or provision of services. Output restriction is the arithmetical equivalent of price fixing, since reducing what reaches the market raises the price without any need to agree on the figure. The clause also covers agreements to restrict technical development and investment, which are the dynamic counterparts.
3.3 Market sharing
Section 3(3)(c) covers an agreement that shares the market or source of production or provision of services by way of allocation of the geographical area of the market, or type of goods or services, or number of customers in the market, or any other similar way. The three named techniques are geographic allocation, in which each member takes a territory; product allocation, in which each takes a type of goods; and customer allocation, in which each takes named customers and the others do not solicit them. The closing words any other similar way keep the clause open.
3.4 Bid rigging and collusive bidding
Section 3(3)(d) covers an agreement that directly or indirectly results in bid rigging or collusive bidding. The Explanation defines bid rigging as any agreement between enterprises engaged in identical or similar production or trading of goods or provision of services which has the effect of eliminating or reducing competition for bids, or adversely affecting or manipulating the process for bidding.
- Cover or complementary bidding. A member submits a bid that is deliberately too high, or carries conditions it knows are unacceptable, so that the designated winner appears to have won a competitive contest. This is the commonest form.
- Bid rotation. The members take turns to win, the others covering each time, so that over a series of tenders each obtains a share.
- Bid suppression. A member agrees not to bid at all, or to withdraw a bid already submitted.
- Subcontracting arrangements. The losers are compensated by subcontracts from the winner, which is how the gain is divided and is often the best evidence of the arrangement.
๐ Excel Crop Care Ltd. v. Competition Commission of India, (2017) 8 SCC 47 Held: Four manufacturers of aluminium phosphide tablets had quoted identical rates in successive tenders of a public buyer and had boycotted an earlier tender. The Court upheld the finding of bid rigging under Section 3(3)(d), holding that identical bids by parties with different cost structures, together with the boycott and the surrounding circumstances, established the agreement. On penalty it held that where an enterprise is multi-product, the penalty under Section 27(b) must be computed on the relevant turnover, that is the turnover of the product to which the contravention relates, and not on total turnover, since the penalty must be proportionate. It also held that the Commission's jurisdiction extends to conduct continuing after the Act came into force, even if it began earlier. Significance: The leading Indian decision on bid rigging and, until the amendment of 2023 redefined turnover, on the base for penalties. The proportionality principle it states now operates through the Commission's penalty guidelines. |
โ Bid rigging and legitimate joint bidding A consortium bid is not bid rigging. Firms that could not individually meet the tender requirements, whether of capacity, technical qualification or financial standing, may bid together, and such a bid adds a competitor to the contest rather than removing one. Three features distinguish the legitimate case: the parties could not realistically have bid alone; the arrangement is disclosed to the procuring authority in the bid itself; and it is confined to the tender in question rather than extending to the parties' conduct generally. Where firms each capable of bidding alone agree to bid jointly, or where a joint bid is used to exchange information about future pricing, the arrangement falls within Section 3(3)(d). |
4. Proving a Cartel
Cartel proof proceeds along two routes, and the second does most of the work.
- Direct evidence. Documents or testimony showing the agreement itself: minutes of a meeting, an exchange of messages, a price circular, a leniency applicant's disclosure, or material seized in a search under Section 41 read with the powers of the Director General. Leniency under Section 46 is the principal source, because a cartel member who comes forward brings the documents with him.
- Circumstantial evidence. Where there is no direct proof, the agreement is inferred from conduct. The standard is preponderance of probabilities, not proof beyond reasonable doubt, since the proceeding is civil; but the inference must be the only reasonable one on the material, and conduct equally consistent with independent behaviour will not support it.
4.1 Parallel conduct and conscious parallelism
In a market with few sellers and a homogeneous product, firms watch each other and match each other's prices. That behaviour is called conscious parallelism, and it is not unlawful: each firm is entitled to adapt intelligently to what its competitors do, and in an oligopoly rational independent conduct produces the same outward pattern as collusion. The law therefore requires something beyond parallelism before an agreement is inferred.
4.2 Plus factors
- Conduct against the individual self-interest of each firm unless the others behave identically.
- Opportunities to collude: meetings, association gatherings, common agents or shared personnel close in time to the conduct.
- Identical bids or prices to the last rupee, or simultaneous changes without any common cost event.
- Abrupt and unexplained changes in a long-standing pattern of competition.
- Artificial or unexplained differences in conduct as between customers or regions.
- The exchange of information that a firm acting alone would have no reason to give a rival.
- A high degree of market concentration with stable shares over time.
๐ Rajasthan Cylinders and Containers Ltd. v. Union of India, (2020) 16 SCC 615 Held: Identical bids for cylinders in tenders floated by a single public sector buyer did not establish a cartel. The Court examined the market: there was one dominant buyer with countervailing power, a limited number of suppliers, capacity allocated by the buyer, and a reserve price known to the industry. In those conditions the parallel conduct was explicable without collusion, and the presumption under Section 3(3) stood rebutted. The Commission must consider the structure of the market before drawing an inference from parallel behaviour. Significance: The counterweight to Excel Crop Care. Read together, the two decisions show that identical bidding is a starting point and not a conclusion, and that the answer depends on whether the market itself explains the pattern. |
5. Information Exchange and Trade Associations
Exchanging commercially sensitive information between competitors is the commonest facilitating practice, because it removes the uncertainty on which competition depends. Information about future prices, planned output, capacity utilisation, costs, customers and bidding intentions is sensitive; historical, aggregated and anonymised data published with a lag is generally not. The exchange may itself be an agreement within Section 3(3) where it enables the parties to align their conduct, and it is powerful circumstantial evidence in any event.
Trade associations occupy a difficult position. They exist to represent an industry, which requires their members to meet and to share concerns, and the same meetings provide the opportunity to coordinate. Section 3(3) applies expressly to a practice carried on or a decision taken by an association, so a resolution fixing a price, allocating quotas, prescribing a uniform discount or directing members not to deal with a person is within the prohibition, and the association is liable along with the members who implemented it.
๐ Competition Commission of India v. Coordination Committee of Artists and Technicians of W.B. Film and Television, (2017) 5 SCC 17 Held: A coordination committee of associations of artists and technicians had directed television channels not to telecast a dubbed serial, and had procured compliance by threatening non-cooperation. The Supreme Court held that the committee was an association of enterprises acting in the economic sphere, that its decision was an agreement within Section 3(3)(b) limiting the supply of goods and services in the relevant market, and that the relevant market was the market for the broadcast of television serials in the language and region concerned. The character of the members as a trade union did not remove the conduct from the Act, because what was done was economic in nature. Significance: The leading decision on decisions of associations and on collective boycott, and a useful statement on how the relevant market is identified in Section 3 cases. |
A boycott agreement, by which competitors collectively refuse to deal with a supplier, a customer or a new entrant, is treated under Section 3(3)(b) as limiting supply, and is one of the most serious forms of horizontal conduct, because its object is usually to discipline the person boycotted into accepting the group's terms.
6. Hub and Spoke Arrangements
In a hub and spoke arrangement the competitors, who are the spokes, do not communicate with each other. They each deal with a common party at a different level of the chain, the hub, which may be a supplier, a distributor, a platform or an association, and the hub passes information between them and coordinates their conduct. The rim of the wheel, meaning the understanding among the spokes themselves, is what converts a series of vertical arrangements into a horizontal cartel.
Before 2023 the difficulty was textual. Section 3(3) applied to enterprises engaged in identical or similar trade, and a hub is by definition not in the same trade as the spokes, so it could argue that the presumption did not touch it. The amendment of 2023 added a provision to the effect that an enterprise or person which is not engaged in identical or similar trade shall nevertheless be presumed to be part of such an agreement if it actively participates in the furtherance of it. The hub is therefore now within the presumption where its participation is active rather than passive.
๐ Samir Agrawal v. Competition Commission of India, (2021) 3 SCC 136 Held: The allegation was that cab aggregator platforms, by setting fares through an algorithm applied to all drivers on the platform, had enabled a hub and spoke cartel among the drivers. The Supreme Court affirmed the concurrent findings that no such agreement was made out on the facts: the drivers did not coordinate among themselves, and the platform's algorithmic pricing did not establish a rim between them. On locus standi the Court held that any person may provide information under Section 19(1)(a), and an informant need not be personally aggrieved, because the proceedings are in rem. Significance: The decision is authority both for the width of the right to inform and for the proposition that an algorithm applied by a platform is not, without more, a horizontal agreement among its users. It was decided before the amendment of 2023, and the position of a platform that actively coordinates is now governed by the added provision. |
โ Algorithmic and tacit collusion Algorithms create three distinct problems. The first is an algorithm used to implement an agreement already reached, which raises no new question of law: the agreement is proved as usual and the algorithm is the instrument. The second is the hub and spoke case, where competitors use the same pricing algorithm or the same provider, which the 2023 provision now addresses. The third is genuinely unresolved: pricing algorithms observing each other may learn to sustain supra-competitive prices without any communication between the firms, which is tacit collusion by machine. Indian law, like most systems, requires an agreement, and conduct with no communication and no meeting of minds falls outside Section 3 altogether. Whether such conduct should be reached, and by what provision, is the subject of the current debate on digital competition regulation. |
7. Cartels Distinguished
7.1 Cartel and conscious parallelism
Conscious parallelism is independent conduct in which each firm, knowing that its rivals watch it, chooses to match rather than to undercut. There is no communication and no commitment, and each firm remains free to depart. A cartel involves a shared understanding that constrains the members. The practical test is the one the plus factors express: is the parallel conduct explicable as rational independent behaviour in this market structure, or does it require an agreement to make sense?
7.2 Cartel and joint venture
A joint venture is an arrangement by which enterprises combine resources to do something neither would do alone, such as research, production or bidding for a project beyond their individual capacity. A cartel does not create anything; it coordinates conduct the members would otherwise carry on in competition with each other. The proviso to Section 3(3) reflects the distinction by excluding from the presumption an agreement entered into by way of a joint venture if it increases efficiency in production, supply, distribution, storage, acquisition or control of goods or provision of services.
- The exemption is from the presumption only. The agreement remains subject to Section 3(1), so a joint venture that causes an appreciable adverse effect is still a contravention; what it loses is the automatic assumption of harm.
- Efficiency must be shown. The proviso is conditional on the agreement increasing efficiency, and the increase must be demonstrated rather than asserted.
- Ancillary restraints are examined separately. A restriction that goes beyond what the venture requires, such as an agreement on the parties' prices in markets outside the venture, is not protected by the proviso.
- A sham venture is a cartel. Where the arrangement does not produce integration of function or resources and exists only to coordinate conduct, it is treated as what it is.
8. Related Topics and Provisions
Topic or provision | Connection |
|---|---|
Anti-Competitive Agreements: Section 3 | The structure of the section and the definition of agreement |
Per Se Approach and the Rule of Reason | What the presumption does |
Anti-Competitive Harm and Pro-Competitive Benefits | The factors used in rebuttal |
Basic Competition Economics | Market conditions favourable to collusion, and plus factors |
Sections 2(b), 2(c), 3(3), 19(3), 27 and 46, Competition Act, 2002 | Agreement, cartel, the presumption, factors, orders and leniency |