Competition Bureau

News & Resources > Public comments > Draft - Information Bulletin on the Abuse of Dominance Provisions as applied to the Telecommunications Industry > II. Market Definition

Draft - Information Bulletin on the Abuse of Dominance Provisions as applied to the Telecommunications Industry: Part 2 - Market Definition  

2.1 Role of Market Definition in Abuse of Dominance Cases

As noted in section 1.3 above, market definition is necessary to establish the first element of subsection 79(1), namely, that one or more persons substantially or completely control, throughout Canada or any area thereof, a class or species of business.

Defining the relevant product and geographic markets traditionally focuses on identifying competitors that are likely to constrain the ability of a firm to profitably raise price^[13] or otherwise restrict competition. Such competitors are identified by their provision of alternative products or geographic sources of supply to which buyers would be willing and able to substitute if the price for the product were to rise above competitive levels. As a practical matter, the way market definition principles are applied in abuse of dominance cases can differ from merger cases and forbearance analyses in that abuse cases are typically retrospective in nature while merger and forbearance analyses are typically prospective.

The boundaries of a market for competition analysis are delineated using the “hypothetical monopolist” framework to determine the smallest group of substitute products and the smallest region of production that a firm must control such that a profit-maximizing firm (the hypothetical monopolist) would have an incentive to implement a small, but significant and non-transitory increase in price (referred to as a “SSNIP”)^[14] above competitive levels. The alleged anti-competitive acts effectively provide the initial candidate product and geographic market in which the acts have the potential to maintain or enhance market power. These acts also identify the time period during which it is alleged that the firm exercised any such market power.

2.2 Determining Competitive Price Levels – Avoiding the Cellophane Fallacy

In allegations of abuse of dominance, a SSNIP refers to an increase in price above competitive levels rather than prevailing levels. Using the prevailing price^[15] could lead to overly broad definitions of product markets. This is because products that appear to be in the market based on the prevailing price may not be considered substitutes at competitive price levels. The same is true for geographic market definition. This problem was first identified in a case in the United States involving the producers of cellophane and is therefore referred to as the “cellophane fallacy”.^[16]

The potential for the cellophane fallacy necessitates an evaluation of whether, and if so the extent to which, the prevailing price differs from the competitive price. Market definition in allegations of abuse of dominance is therefore considerably more complicated than, for example, merger cases because it involves the often difficult assessment of the competitive price level. The Bureau ideally would begin this assessment by estimating the approximate price level for the product in the absence of the alleged practice of anti-competitive acts.^[17]

Regardless of the Bureau's ability to make such an estimate in a given case, the Bureau will look to the following types of evidence in an effort to address the cellophane fallacy:

·lack of functional interchangeability between the products of the alleged dominant firm and other potential competitors and differences in the technology and inputs used by different firms;^[18]

• differences in the effect of cost shocks on prices;^[19]
• differences in supply-side substitution across regions;
• supra-competitive profits being earned by the alleged dominant firm;^[20]
• lack of positive price correlation between the products of the alleged dominant firm and its rivals; and
• substantial differences in price levels.^[21]

There are several examples from past cases of evidence that avoids the cellophane fallacy. In Laidlaw, the Tribunal used observations of usual business practice in other geographic regions and the increase in costs if service was provided more than 50 kilometers away from a hub, to define relatively narrow geographic markets for commercial waste collection. In NutraSweet, the Tribunal defined the product market to be aspartame, in part because of the large price differences between Canada and the United States (where aspartame was still under patent protection), even though the set of similar sweeteners was essentially the same in both countries.^[22] In Canada Pipe, the Tribunal relied on evidence of lack of price correlation between obvious substitutes; high profit margins; price differences between regions attributable to differences in competition; and domestic prices above import prices to define relevant markets and assess market power.^[23]

In recently deregulated telecommunications markets, there may be added difficulty in assessing evidence and data.  For example, in an abuse of dominance analysis, the Bureau may consider evidence, such as business plans, strategic documents and data, during both the regulated and forborne time periods but interpreting this evidence in the context of both time periods may prove difficult.

2.3 Relevant Product Market Definition

As noted in section 2.1 above, product market definition involves identifying whether there are close substitutes for the product in question such that buyers would switch to these substitutes in the event of a SSNIP. If there are, these substitute products will be included in the relevant product market.

In assessing whether products are close substitutes for one another, what matters are the characteristics of the product and consumers' ability and willingness to switch from one product to another. ^[24] Direct evidence of switching behaviour in response to a SSNIP (i.e., price changes and concomitant quantity changes) would indicate substitutability.^[25] In the absence of direct evidence, a number of factors may assist in determining product substitutability. These include the views, strategies, behaviour and identity of buyers and sellers or other market participants; end-use or functional interchangeability of the products; physical and technical characteristics of the products; switching costs; and other impediments to trade. ^[26]

2.4 Relevant Product Market Definition in the Telecommunications Industry

A number of issues related to product substitutability are particularly likely to arise, or to be relevant, in the telecommunications industry.

For example, in considering an allegation of abuse of dominance by a firm in its provision of local residential telephone service, the Bureau would first assess the willingness and ability of consumers to substitute to other local residential telephone services provided by different technologies (e.g., circuit switched, IP, and wireless technologies) to determine whether the services provided by these technologies are in the same product market. As part of this exercise, the Bureau would consider whether there are characteristics of a particular local residential telephone service (e.g., reliability and clarity) that sufficiently differentiate it from other local residential telephone services such that customers are unlikely to switch in response to a SSNIP. The Bureau would also consider whether there are costs involved in switching, which would make switching a less likely response to a SSNIP. Examples of such switching costs include penalties associated with terminating an existing contract before it expires; service charges associated with migrating to another service provider; and new equipment that is required to use a similar service offered by a competitor (e.g., a wireless phone handset).

Bundling is a market definition issue that may be relevant in the telecommunications industry. A bundle is an offering of a package of services at a price benefit compared to the price of the individual services in the package taken on a stand-alone basis. A bundle could include any number of communications services, such as video; high-speed Internet access; wireless, long distance andor local telephony services; and, optional local services. Generally, bundling of multiple communications services gives rise to the possibility that the relevant product market will be defined as the bundle. Whether bundles define the market may depend on whether there is a significant enough difference in cost to consumers between buying a bundle and buying each service independently.  Bundles maybe a separate product market if consumers would not turn to separate services if the bundled price increases by a SSNIP.  If that is the case, relevant markets could be defined around the bundles.

However, a few key services may differentiate one service provider's bundle from that of another's and accordingly may significantly affect the willingness of consumers to substitute between bundles.  In such cases, if not all competitors can offer a comparable bundle, aggregation may not be possible.  In the alternative, to the extent that any key service of the bundle is simply a function of the intelligence of the network/technology, and other providers can access the appropriate software on similar terms to provide that service as part of their bundle, it is unlikely that such services will be sufficient to differentiate one supplier's bundle from another.  In this case, aggregation of the product market as described above may be possible.^[27]

2.5 Relevant Geographic Market Definition

For the purpose of geographic market definition, what matters is the ability and willingness of consumers to switch from suppliers at one location to suppliers in another location in response to a SSNIP. A relevant geographic market consists of the smallest region within which a “hypothetical monopolist” of all sources of supply that are regarded as close substitutes by buyers, could impose a SSNIP.^[28]

In the absence of direct evidence of switching behaviour in response to a SSNIP, a number of factors may assist in ascertaining consumers' willingness to switch between suppliers in different locations. These generally include the views, strategies, behaviour and identity of buyers; trade views, strategies and behaviour; switching costs; transportation costs; price relationships and relative price levels; shipment patterns; and foreign competition.^[29]

2.6 Relevant Geographic Market Definition in the Telecommunications Industry

For many telecommunications services, the provision of the service is tied to a location and the number of competitive alternatives that are available to consumers can differ depending on where they live or carry on business.^[30] In such cases, geographic market definition typically involves starting with the location(s) where telecommunications service is supplied and considering whether customers would switch to suppliers at a different location in response to a SSNIP.^[31] For example, with respect to telecommunications services, the Bureau would define the relevant geographic market based on a specific location if subscribers to services provided at that location were not willing to substitute to services supplied at a different location. ^[32] In essence, a household or place of business theoretically could be defined as a relevant geographic market. 

Where appropriate, the Bureau would aggregate all locations that have the same competitive alternatives (within the product market) for the relevant telecommunications services into a single geographic market. In some cases, such aggregation may follow directly from an examination of the geographic location of networks and consumers. Where there are differences in the geographic coverage of competing networks (i.e., “holes”), it may be necessary to also look at the factors listed in section 2.5 in order to ascertain whether a hypothetical monopolist would impose a SSNIP with respect to such customers.

After such aggregation, a geographic market can be defined around the network of a dominant firm based on its overlapping footprint with competing networks that provide the relevant telecommunications services (i.e., in assessing which locations have the same competitive alternatives, the Bureau includes potential competitors that can easily provide service to that location). Aggregation of geographic markets identified by competitive alternatives (including potential entrants) is generally sufficient to delineate geographic markets that are appropriate for the identification of market power.^[33]  

For telecommunications services that are not tied to the location of a customer's residence or place of business (e.g., mobile telecommunications, terminal equipment), the analysis of geographic market would proceed immediately to the factors described in section 2.5.

---

[13] Unless otherwise indicated, price means price, output, quality, variety, service, advertising, innovation and other dimensions of competition. Therefore, raising price may also mean reducing these other non-monetary dimensions of competition.  

[14] Generally, a 5 percent real price increase above competitive levels lasting for one year is considered to be a SSNIP. Where price refers to a non-monetary dimension of competition (e.g., quality, service, etc.), a SSNIP may refer to a decrease below competitive levels.  With respect to the specific tests used under the SSNIP, it is recognized that market characteristics may sometimes necessitate using a different price increase or time period.

[15] The prevailing price is the price at which the product is currently offered to customers in the market.

[16] For an example of how the cellophane fallacy has been recognized in Canada, see Canada (Director of Investigation and Research) v. Laidlaw Waste Systems Ltd., [1992] 40 C.P.R. (3d) 289 (Comp. Trib.) [Laidlaw].

[17] A number of observations relating to this estimate should be kept in mind. First, this method assumes that the competitive price can be accurately determined. If that is the case, the answer is immediately apparent without resorting to the market definition step. In such instances, the analysis is, in effect, defining the market and assessing competitive effects simultaneously and there is no need (except to complete the record) to actually delineate the market. Second, it is noteworthy that, since abuse cases typically proceed some time after the alleged anti-competitive act has taken place, it may be possible to observe prices and the number of competitors in the market both before the conduct was engaged in, and after. Where quantitative evidence is available, the competitive price may be estimated (recognizing that other factors that affect price may have also changed, which would make the estimate inaccurate in some cases). Where quantitative evidence is not available, the pattern of consumer switching behaviour over this period may provide some insight into which product and geographic areas are not close substitutes at the competitive price level.

[18] Evidence that costs, and consequently competitive prices, are different indicates that the products are likely in different relevant markets.

[19] Evidence that costs shocks affect the alleged dominant firm's product (i.e., cause its price to change in response) but not those of its alleged rivals (i.e., do not cause their prices to change) suggests that the rivals' products are not in the same market.

[20] Evidence of accounting profits is usually very difficult to interpret.

[21] Substantial differences in price levels indicate that it is unlikely that the higher priced product constrains the market power of the lower priced product when it is priced at its competitive price. Accordingly, the higher priced product should be excluded from the market. This holds even where the two products are functionally interchangeable.

[22] supra, note 9.

[23] supra note 10.

[24] Competition Bureau, Merger Enforcement Guidelines, (Ottawa, Industry Canada, 2004), online: Merger Enforcement Guidelines. The Bureau's Merger Enforcement Guidelines clarify that market definition looks only at demand responses (i.e., the incentive and ability of consumers to change their consumption patterns and turn to existing competitors). Supply responses (e.g., the likely responses of “potential” competitors) are important when assessing the potential for the exercise of market power, but are not examined when defining relevant markets. Whether this is done at both the market definition and assessment of market power stages of the analysis or solely in the latter, does not substantively affect the analysis since both existing and potential competitors must be considered before making a conclusion on the market power question. However, since evidence of demand responses is usually distinct from, and usually comes from different sources than, evidence of supply substitution, the Bureau has found as an operational matter that separating the analysis into two steps can be helpful, particularly when dealing with cases in complex industries such as telecommunications.

[25] Canada (Director of Investigation and Research) v. The D&B Companies of Canada Ltd., [1995] 64 C.P.R. (3d) 216 (Comp. Trib.) [Nielsen].

[26] These factors are described in greater detail in the Enforcement Guidelines on the Abuse of Dominance Provisions, supra, note 3 at 11-12.

[27] A discussion of the potential for bundling to constitute a practice of anti-competitive acts is set out in Section 4.4.

[28] When geographic price discrimination is present (and buyers and third parties are unable to arbitrage between low and high price areas), geographic markets are defined around the location of each targeted group of buyers.

[29] These factors are described in greater detail in the Enforcement Guidelines on the Abuse of Dominance Provisions, supra note 3 at 12-13.

[30] Examples include local residential telecommunications services and many business services.

[31] Whether consumers are able and willing to buy service from suppliers at other locations in general depends on the extent of transaction costs.  If transaction costs are sufficiently high, an analysis of demand responses indicates that each location will be a relevant geographic market holding the product market definition constant.

[32] A key issue involves the extent of transaction costs incurred by a subscriber at a given location to substitute to an alternative location for access.  Prima facie it seems very unlikely that a subscriber at one location would cancel service there and substitute to access at another location in the face of a SSNIP.  If this is true, then every location is a relevant geographic market and only suppliers that can provide access to that location are in the market.

[33] For example, in its review of the Rogers/Microcell merger, the Bureau found the geographic market to be less than national in scope. A national market was rejected because there were some differences in the number of competitors, product offerings and prices across provinces, and there was no persuasive explanation showing that a SSNIP could not have been imposed on a provincial basis. The Bureau did not pronounce on whether the geographic market could have been defined more narrowly (e.g., individual cities) since it was determined that the findings would have been the same whether markets were defined as provinces or more locally. It was determined that an aggregate description of the geographic market (i.e., provinces) was appropriate.