Documentation

~ File No. CT-94/01 1llE COMPETITION TRIBUNAL IN 1llE MATIER OF an Application by the Director of Investigation and Research under section 79 o -·--w·-·-·~········--Competition Act, R.S.C. 1985 c. C-34 as amen ed;c,.,~:::.:~~·~,: '~.· ·!..-~~·~=-~~,--•~~.t'l'.: .. ,,~ •. ·.•.. ' . . ~:."~·.:jC~ AND IN 1llE MATIER OF certain practices 1>f ~ A.C. Nielsen Company of Canada Limited ~ OCT 2s 1994 AJ ~ Rt\.,,_, ... ,.1.,, .... , T BE1WEEN: 0T1A~V .. \, ONT. -r_.:.i9 .t3 9 TIIB DIRECTOR OF INVESTIGATION AND RESEARCH

Applicant COMPETITION TRIBUNAL TRIP' ·~AL OE LA CONCURRENCE • and • ~le No. (!,T -1f ft , t;_. du do11i11 JL. ~ 1 <f3 tl.d ~_) .lWM+l&L ~ ~ r=r==THE D&B COMPANIES OF CANADA LID. Exhibit No. f - 5 f (bj No. de la piM / J ~ Respondent 1 ; Filed o• rev r-' ', 14-; 11r.3r oipoa6e ~· -.J.,/..~ · and -Regittrar ....;-~-...::.-U---;.:;:;....p.-----Grettier INFORMATION RESOURCES, INC.

Intervenor

RESPONSE AFFIDAVIT I, DONAID N. THOMPSON, of the Oty of Toronto, in the Province of Ontario, make oath and say as follows:

1. In January of 1994 I was retained by the firm of Davies, Ward & Beck of Toronto, legal counsel to Information Resources, Inc. ("IRI"), the Intervenor in this i

proceeding, to advise on the competitive implications of certain marketing practices in Canada of the Respondent firm, The D&B Companies of Canada Ltd., one division of which is Nielsen Marketing Research, referred to hereafter as "Nielsen• referring to the Canadian company and •A.c. Nielsen• refe~g to the U.S. parent company. I was asked to assist legal counsel to the Intervenor in the analysis of economic and other information on this matter.

2. On September 20th, 1994, I swore an Affidavit in this matter, appended to which was my Confidential Affidavit Evidence, for filing with the Competition Tribunal.

3. On September 21, 1994, I was given the confidential Affidavit of Professor Frank Ma ­thewson ("Mathewson•), on behalf of the Respondent, and the confidential Affidavit of Margaret E. Guerin-Calvert ("Guerin-Calvert"), also on behalf of the

Respondent. I was asked by counsel to IRI to prepare a response to the arguments raised by Mathewson and Guerin-Calvert in such Affidavits. Attached as Exhibit "A" to this my Affidavit is a true copy of the report prepared pursuant to that request.

4. My report and opinions are based on my professional training as an economist, my reading of Mathewson and Gucrin-calvcrt, my review of documents provided by Nielsen, IRI and the Director in this proceeding, my reading of transcripts of examination for discovery of Mr. Stephen Churchill of Nielsen, and my review of relevant literature. My report and opinions arc also based on my independent research, data gathering and data analysis, including discussion of industry practices with officials of IRI, with a number of executives of consumer packaged goods companies in Canada

ii

and in the U.S., and with several executives of supermarket firms which provide scanner data to research firms.

SWORN before me at the City ) of Toronto in the Regional ) /)JJ Municipality of Metropolitan ) • .· , Toronto in the Province of ) · nV Ontario this 4th day of ) ·-"- ( ~(,:_~ ) commlo eror Taking Affidavits iii

File No. CT-94/01 THE COMPEi I llON TRIBUNAL IN TIIE MATIER OF an Application by the Director of Investigation and Research under section 79 of the Competition Act, R.S.C 1985 c. C-34 as amended;

AND IN THE MATTER OF certain practices by AC. Nielsen Company of C-anada Limited

BElWEEN: THE DIRECTOR OF INVESTIGATION AND RESEARCH Applicant - and -THE D&B COMPANIES OF CANADA LTD. Respondent - and -INFORMATION RESOURCES, INC. Intervenor

RESPONSE AFFIDAVIT OF DONALD N. TIIOMPSON

1. INTRODUCTION 1.1 I believe that certain key assumptions in the Mathewson and Guerin-Calvert analyses are incorrect. Therefore, the principal conclusions they have reached are, in my opinion, without support. 1

1.2 I set out below what I consider to be the key assumptions in the Mathewson and . Guerin-Calvert reports, and the evidence and analysis which suggests that those assumptions are incorrect. The assumptions relate to the following areas: (a) the relevant product market; (b) the relevant geographic market; ( c) the "brokering" of retailer data; ( d) the contract negotiating process; and ( e) the price of Nielsen's syndicated market tracking ("SMT") services. 1.3 As a preliminary matter, I note that both Mathewson and Guerin-Calvert discuss the evolution of the contract conditions for scanner data supply from Canadian grocery retailers and, in particular, the origin of Nielsen's exclusive data access provisions. See Mathewson, 2.4.1 to 2.4.5, and Guerin-Calvert, paragraphs 13, 17, 18, 26 and elsewhere.

1.4 However, in considering allegations of an abuse of dominant position, how ·the dominant position was established and the history of whatever "abuses" are associated with that dominant position are both irrelevant. From an economic standpoint, what is relevant in discussing Nielsen's abuse of dominant position is: (a) whether at a point of time (in this case, when the Director's Application was filed), Nielsen was a dominant firm in providing what buyers consider as a distinct product in a distinct geographic market;

2

(b) whether Nielsen's market practices have been or will be anticompetitive in terms of deterring entry, and therefore constitute an abuse of Nielsen's dominant position; and (c) entry conditions to the market, with and without the market practices in question, at that point in time. 2. DISCUSSION OF KEY ASSUMPTIONS AND CONCLUSIONS BY MATHEWSON AND GUERIN-CALVERT

2.1 1be Relevant Product Market 2.1.1 Both Mathewson and Guerin-Calvert state that the relevant product market is "all" market tracking services, including those that use data collection methodologies other than scanner data. They argue that data sourced from store audits, warehouse shipment and other sources is equally acceptable (or substitutable) to buyers of SMT services based on scanning data. Mathewson goes further and states that market tracking services include both data on consumer purchases and analyses of these data. Mathewson says at 2.1.1: "'lbe products in this market [for market tracking services] include both data on consumer purchases and analyses of these

data."

Mathewson says at 2.3.1: "'lbe product market (market tracking services) is large and diverse ... there are alternative sources of marketing data and analysis... . The data may be sourced from store scanning, store audits,

warehouse shipments, factory shipments, consumer diaries, and consumer scanning."

Guerin-Calvert says at para. 31: ". .. there are actually a number of products that fall into the market tracking category, only some of which are scanner-based and only some of which arc national products. The information I have reviewed thus far indicates that there were alternatives to scanner-based products for consumer packaged goods manufacturers during the period 1986 to the present These products include those that are based on individual chain data or on regional data, as well as reports that arc based on data from store audit or warehouse withdrawal. In addition, there were products, such as those produced by ISL, which were based on consumer data. During the period 1986 to the present, there were products

3

to which consumer packaged goods manufacturers could have turned as substitutes for those provided by Nielsen."

Discussion: 2.1.2 A relevant product market is defined by inclusion of all products which are close substitutes in use. The Director's Application defines the product market in terms of an input, namely "scanner-based" SMT services. The relevant product market in this case can therefore be determined by asking whether there are SMT services using other inputs which could be considered a close substitute in use for scanner-based SMT services.

2.1.3 It may be that, in theory, a national SMT service could be developed using data collection methodologies other than scanner data. However, based on the information that I have seen and collected in conjunction with this matter and, in particular, on my interviews with industry participants, it is my opinion that any such service would not be a close substitute for a scanner-based SMT service. A scanner-based SMT service has different characteristics (for example, frequency, resolution, accuracy, completeness, inclusion of price data) and far broader and more flexible applications (for example, in relating to causal factors) than does a warehouse withdrawal SMT service, an audit-based SMT service, or a diary-based service. Other types of data inputs and services are not good substitutes for scanner-based SMT services. 1 1 The fact that a few manufacturers purchase warehouse withdrawal SMT services rather than scanner based SMT is likely a function of the two products• relative price levels. Warehouse withdrawal SMT data, because of its characteristics, is not competitive with scanner SMT data. But a monopolist seller can raise scanner-based SMT prices to a point where a much less expensive warehouse withdrawal SMT service becomes a substitute. This phenomenon of a product with fewer desirable attnbutes becoming part of a competitive set only after a monopolist inacascs price is dcscnbcd in economic literature as the •cellophane fallacy". 4

2.1.4 The superiority of a scanner-based SMT service over other forms of SMT services is reflected in the fact that, in the United States, where there are no exclusive data access contract provisions for scanner data, IRI entered the market with great success, relying only on scanner data for its SMT service. Lack of market demand caused AC. Nielsen to discontinue offering its SMT services that were based on in-store audits (its NFI/NDI Index) and on warehouse withdrawal measurement (NWSS) and to rely completely on scanner-based data collection for its SMT services.

2.1.5 Mathewson and Guerin-Calvert state that IRI could have used a mixture of audit, warehouse withdrawal and/or diary panels to offer an SMT service in competition with Nielsen. However, as I described above, these methods produce inferior tracking information. Based on my discussions with industry participants, I believe that few, if any, Canadian customers would buy an SMT service based on these inputs, rather than Nielsen's SMT service which is at least partially based on scanner data.

2.1.6 Furthermore, IR.I could not have utilized these inputs to develop an SMT service as suggested by Mathewson and Guerin-Calvert.

s

2.1.7 A final point relates to Mathewson's contention that market tracking services include both data on consumer purchases and analyses of these data "Data" and "analyses" are not one product but complementary products. The fact that there may be competitors offering analyses is also irrelevant; if there is a monopoly in the supply of processed scanner data, all monopoly profit can be taken at that level, and the price of data plus the complementary product of analyses is irrelevant.

2.2 The Relevant Geographic Market 2.2.1 Mathewson and Guerin-Calvert state that the market for SMT services may be local, regional, or provincial as well as national. They state that a new entrant would not need a national scanner data sample or access to all major retailers; it could build an SMT service limited to individual retailers or regions. Mathewson says at 2.3.1: "The data required may be local, regional, provincial or national."

Guerin-Calvert says at footnote 4: "Yet, there are a variety of entry strategies that are based on obtaining data from less than "all retailers", and include building up to a national sample. An entrant does not have to wait to make sales in this industry until it has developed a national sample:

6

Guerin-Calvert says at para. 24: "Winning this [one] contract would have enabled IRI to begin building the set of retailers that it needed to sell one or more types of marketing services".

Guerin-Calvert says at para. 32: "While the Director's Application focuses primarily, if not exclusively, on a national grocery market tracking report based on scanning, there are actually a wide array of other services produced by Nielsen in this area. These include key accounts and sectional reports, among others... Similarly, city and sectional or regional reports do not require scanner data from the entire country... Production of these services can be undertaken on a far smaller scale than by development of samples based on the entire channel of distribution. This can be an effective entry strategy as well as a marketing niche approach for firms in this industry." n·i scuss1.0 n: 2.2.2 It may be that, in theory, a new entrant could enter the market tracking services business by developing a local, regional, provincial or key account SMT service. However, based on the information that I have seen and collected in conjunction with this matter, I do not believe that a strategy would be successful. Even if it was successful, it is my opinion that a local, regional, provincial or key account service would not be a close substitute for a national SMT service and, accordingly, it is my opinion that the relevant geographic market is Canada.

2.2.3 As a preliminary matter, I believe that it is most unlikely that an entry strategy which involves starting with a local, regional, provincial or key account SMT service and then building to a national SMT service would be successful. There are several reasons for this conclusion. First, based on my interviews with consumer packaged goods manufacturers in Canada and in the United States, I believe that there would be little or no demand for a scanner-based SMT service which covered only one region or metropolitan market. A customer would not buy both an SMT service for one city or 7

for one region from one supplier such as IRI and a national SMT service from another supplier such as Nielsen. Nielsen and IRI have incompatible market measures, and the· customer would have to make elaborate and expensive programming changes to use two incompatible sets of data. Accordingly, a customer that purchases a national SMT service would most likely buy a local, regional, provincial or key account SMT service from the supplier of its national service.

2.2.4 Therefore, the market for a local, provincial, regional or key account service would be limited to manufacturers that did not also purchase a national ·S MT service.

8

2.2.6 These Nielsen documents and figures indicate that the size of the market that would be available to a new entrant offering only a local, provincial, regional or key account service would be insufficient to support such entry.

2.2. 7 Furthermore, I believe that a new entrant would be unable to acquire the rights to the scanner data that would enable that entrant to offer a local, provincial, regional or key account SMT service -- this point is discussed in further detail under the heading "'The 'Brokering' of Retailer Data" below.

2.2.8 In any event, if a new entrant were able to offer a local, provincial, regional or key account SMT service, such a service, by definition, would not be a good substitute for a national SMT service.

2.2.9 The behaviour of buyers of SMT services in Canada suggests that the relevant geographic market is Canada (excluding Newfoundland, which Nielsen apparently does not cover). Mathewson at 2.2.4 lists the top 10 Nielsen customers for SMT services. Mathewson states that these top 10 customers account for of Nielsen's gross revenues from its major customers for its MarketTrack Grocery service. 8 9

2.3 'lbe 'Brokering' of Retailer Data 2.3.1 Mathewson states that retailer scanner data could somehow be "brokered" as an alternative to sale. Mathewson says at 2.4.6: ". .. intermediaries and sellers of marketing services such as Nielsen could act as brokers, collecting and processing the data for retailers and manufacturers ... with those manufacturers desiring to purchase key account data paying the retailer directly. A market with two major. retailers might have two intermediaries assembling, processing and marketing scanning da~ on an exclusive basis for each... This illustrates another possibility for organizing market tracking services."

Mathewson says at 2.4.15: "One option open to any retailer is to sell its scanning data to an intermediary who would use the data to provide either a regional market tracking service or to provide chain-specific key account data to manufacturers."

Discussion: 2.3.2 The "brokering" of scanning data argument is a variant of Mathewson's earlier point that there is a market for a scanner-based SMT service which relates to only one retailer or one region. There is no evidence that this is true.

10

2.3.3 Consider the scenario that Mathewson uses as an example, one where chooses IRI as its exclusive agent to market its scanner data, aud is IRl's first Canadian data supplier. (Thus, IRI has access to scanner data only.) would be paid for its data, but it would be unable to obtain from IRI research output on its share or performance in its own markets, because IRI has no other data to provide a market "denominator". would receive no data on its competitors, nor would it have access to comparative data and to the same performance data that its suppli~rs have. requires these types of data to compete effectively. I believe that would not sell its scanner data to any broker under these conditions. 2.3.4 Furthermore, it is likely that IRI would find it uneconomic to pay the amount sought by for the exclusive rights to its scanner data. First, as I have noted above, there is little or no market for the sale of an SMT service that utilizes scanner data alone, certainly not enough demand to justify the costs of data acquisition, processing and marketing. Second, it is unlikely that Nielsen would swap data from other supermarkets with IRI for IRI's ,.. data. Nielsen would have no incentive to do so, as it could use warehouse withdrawal or audit methods to estimate the "missing" sales data. 2.3.S Mathewson could in fact have argued a less restrictive example, that could sell data access to Nielsen and also to IRI on a non-exclusive basis. In this example, · would also get access to Nielsen market reports. But even this 11

example is implausible, because Nielsen will always, given its monopoly position, offer more money for a data access exclusive than would receive for two non-exclusives.

2.3. 7 Even if IRI was willin2 to pay IRI would find it uneconomic to offer any sort of service with t! ccanner data, particularly in circumstances where Nielsen has access to the same data and also has exclusive access to scanner data from other grocery retailers.

2.3.8 The most compelling argument is that, if brokering of data were a viable alternative strategy or form of market organization to compete against a firm like AC. Nielsen or IRI, it would be reasonable to expect that someone in the U.S. would have tried it over the past twenty years. No one has. In fact, in the United States, there have been no new entrants into the market for national scanner-based SMT services since IRI started to carry on business in 1979.

12

2.4 The Contract Negotiating Process 2.4.1 There are several statements regarding the "contract negotiating process" with both retailers and with manufacturers found in Guerin-Calvert and Mathewson. Guerin-Calvert says at para. 19: "During this contracting process, each major grocery chain independently considered and ultimately signed contracts with Nielsen. ... Nielsen does not have market power over the retailers such that it could force exclusives on each of them."

Mathewson says at 2.2.5: ". .. manufacturing firms ... are sophisticated and knowledgeable and have significant bargaining power in their relationships with their suppliers to secure inputs at minimum prices and on reasonable terms ... including those marketing services such as marketing data. .. ".

Mathewson says at 2.2.6: "Such large firms are unlikely to be the recipient of any alleged anticompetitive behaviour on the part of any supplier of inputs ... these manufacturers have bargaining power ... could cut back on their demand for Nielsen's services ... could purchase services from an alternative source of market tracking information such as ISL".

Mathewson says at 3.5: "Since 1986, the contract negotiation process between the major retail grocery chains and Nielsen has been independent

and sequential. This means that a firm such as IRI ... could have chosen to bid on these contracts on any basis it saw fit, in accordance with each

retailer's request."

Discussion: 2.4.2 It is certainly true, as Guerin-Calvert suggests, that Nielsen does not force exclusives on major grocery chains. Nielsen does not have to.

13

2.4.3 Assume that each retailer individually decides, for its own reasons, that it would be better off with two firms, IRI and Nielsen, competing for its data. What would happen?

2.4.4 Under the existing structure of Nielsen's staggered contracts with retailers, those contracts come up for bidding or renewal at different times. Assume that the contract is up for renewal. Because has concluded that it wants to have two firms competing for its data, asks both Nielsen and IRI to submit offers to purchase its scanne~ data on a non-exclusive basis, and both companies do so. However, in order to foreclose IRl's entry into Canada, Nielsen will also submit a bid for the exclusive rights to scanner data. Based on the terms of Nielsen's offers made to retailers in similar circumstances, it is likely that the amount offered by Nielsen for exclusivity will be more, perhaps substantially more, than the amount offered for non-exclusivity. Notwithstanding the retailer's initial objective of having two firms to compete for its data, will accept Nielsen's exclusivity offer. 2.4.S Furthermore, the incentive for to "hedge" by having a short-term contract or a contract with a termination clause is also negated because knows that other retailers have signed longer term (frequently 5 year) contracts. There are few advantages to · in holding out for a shorter term contract, because knows that it alone does not have enough market importance to attract a data purchase offer from IRI.

14

2.4.6 So while there may be a collective advantage to retailers of having non-exclusive data supply or of short-term flexibility in contracts, no single retailer has any incentive to hold out for these, and thus the collectivity of retailers cannot achieve them. Does Nielsen "force exclusives" on each retailer? Not really. Data supply exclusives are the inevitable result of Nielsen's incentive to pay for exclusivity, and the length and staggering of existing contracts.

2.4.7 In the United States, where Nielsen and IRI compete for data supply, there are no exclusive data supply agreements. Every industry participant I interviewed has argued that no U.S. retailer that thought through the consequences would ever grant a data supply exclusive. The argument is that if retailer "A" granted an exclusive to Nielsen, IRI would immediately retaliate by paying whatever was necessary to purchase an equivalent data exclusive from retailer "B" in the same market.