Exclusive dealing in on-premise establishment sales of beer in Edmonton.
In recent years, there has been renewed research interest in the economics of exclusive dealing. While Howard Marvel's free riding explanation of exclusive dealing retains its prominence, other related free riding explanations have recently been offered, along with other efficiency based explanations. (1) On the other hand, economists have also continued to develop anticompetitive explanations for exclusive dealing, largely based on market foreclosure.
One industry in which free riding, efficiency, and anticompetitive explanations for exclusive dealing might be relevant is the beer industry. Bars and restaurants might wish to sell beers from multiple brewers in order to free ride on the investments made in brands and promotion by some of the brewers. One can easily imagine that brewers might achieve significant distribution cost savings or other efficiencies by dealing exclusively with their retail customers. At the same time, exclusive dealing in draft beer by large national brewers with bars and restaurants could test the viability of regional and microbrewers.
The brewing industry in some provinces in Canada also allows brewers to provide their customers with nonmonetary inducements in exchange for agreeing to purchase a certain amount of their products. If a sufficient quantity of the product is specified in the agreement, this could imply de facto exclusive dealing. (2) Given that retail customers may have an incentive to deal exclusively with a brewer anyway, the inducement might be used to determine the brewer from which the retailer will make its purchases.
The inducements given by brewers to bars and restaurants have been the subject of some government investigation and enforcement in Canada. In 2000, after an extensive investigation, the Alberta Gaming and Liquor Commission fined hundreds of bars and restaurants in Alberta for accepting inducements and reportedly fined Canada's two largest brewers for providing them. Apparently, exclusivity in draft beer was one of the objectives of at least some of the inducements. (3) There is also evidence that exclusive arrangements were sought by the major brewers in Quebec. (4)
While the incentive for exclusive dealing by brewers with their retail customers might exist, and some historical evidence has been cited as to its implementation in Alberta and Quebec, we do not know either the extent to which exclusive dealing by brewers in these jurisdictions exists today or the characteristics of the brewers or bars and restaurants where it does exist. And if there is some evidence of exclusive dealing, is it consistent with one or More of the received theories? This article serves three purposes. First, we examine existing rules regarding inducements in various Canadian provinces and describe the history of enforcement and legal cases regarding inducements in Canada. Second, using a unique self-collected data set on draft beer offerings at hundreds of bars and restaurants within a large city in Alberta, we establish the degree to which exclusivity in beer distribution through bars and restaurants exists and is important and describe the main characteristics of retailers that are prone to exclusivity. Finally, we consider whether the data are consistent with alternative explanations for exclusivity in beer distribution.
This article considers a number of possible explanations for exclusivity, including chance, foreclosure, efficiencies, dampening of competition, and incentive conflicts. To anticipate findings, basic observations about the degree and nature of exclusivity in draft beer offerings of bars and restaurants in Edmonton, Alberta, do not support the first four explanations for exclusivity. The data are broadly consistent with a free riding explanation for exclusive dealing.
In the next section of the article, we discuss the market setting for the empirical analysis we carried out. In section III, we review the regulations regarding inducements in various Canadian and American jurisdictions and the facts surrounding the Alberta and Quebec beer cases. Section W provides a review of the theory of exclusive dealing and of the related empirical work. Alternative explanations for exclusive dealing in beer distribution are discussed, along with predictions that can be examined with our data. In section V, the data used in the subsequent analysis are discussed. Section VI examines to what extent each of the alternative explanations for exclusive dealing by brewers with bars and restaurants is supported with data on establishment draft beer offerings in Edmonton. Section VII concludes.
II. THE MARKET SETTING
The beer sold through restaurants and bars in Alberta is produced by large international firms with breweries in Canada and elsewhere, smaller regional firms, breweries in other countries whose beer is imported, and domestic microbreweries. The two largest suppliers are Labatt and Molson; the combined share of beer sales of Labatt and Molson in Canada in 2005 was approximately 83%. (5)
Labatt today is owned by InBev, the world's largest brewer (with 200 brands and 14% of global brewing in 2005), based in Belgium. (6) Labatt was acquired by Interbrew in 1991. The latter company merged with Companhia de Bebidas das Americas (AmBev) in 2004 to form InBev. (7) Labatt Breweries in western Canada brews both national and regional brands, including Labatt Blue and Lite, Budweiser and Bud Light, Wildcat, and Lucky Lager. Labatt purchased Oland & Sons, brewer of Alexander Keiths, in 1971, and Columbia Brewery, brewer of Kokanee, in 1974. (8)
Molson products today are produced by Molson Coors Brewing Co., which was created in February 2005 by the merger of Molson Inc. and Adolph Coors Co. Prior to that date, Molson had acquired Foster's Brewing subsidiary Carling O'Keefe Breweries in 1988. It then formed a 50-50 joint venture with Foster's Brewing, but by 1993, Molson and Foster's held 40% each of Molson Breweries, with the other 20% held by Miller Brewing. Miller's stake was bought out by Molson and Foster's by 1997, and Molson bought Foster's stake in 1998. (9) Molson Coors Brewing Company brews its own products (e.g., Molson Canadian, Rickard's Red, Molson Export, Coors Light), and brews or distributes under license a variety of other brands (e.g., Corona, Miller Lite, Miller Genuine Draft, and Foster's). Molson is Canada's largest brewer by volume. (10)
Other major suppliers of beer in Alberta are Big Rock and Sleeman. Big Rock, with its brewery located in Calgary, commenced operations in 1985 and now produces eight different brands of beer. Big Rock considers itself a producer of "premium quality specialty beers, also known as craft beers." (11) Sleeman Breweries, which was Canada's third largest brewer, was acquired by Sapporo Breweries in 2006. (12) Sleeman, based in Guelph, Ontario, brews and distributes a variety of beers, including Sleeman Original Draught, Unibroue, and Stroh's.
There are also several microbreweries supplying beer in Alberta. Alley Kat Brewing Co., with its brewery in Edmonton, markets beer through liquor stores as well as through bars and restaurants. Other microbreweries, both within and outside Alberta, market beer in Edmonton.
Beer in Alberta is distributed to liquor stores, bars, and restaurants in two ways. First, Canadian domestic brewers may choose to have their products distributed through the central warehouse owned by the Alberta Government (the AGLC warehouse) but operated under contract by Connect Logistics. Alley Kat Brewing Co. distributes its products in this way. The AGLC warehouse also distributes all beer products that are imported into Alberta from outside Canada, regardless of whether these products are produced by a domestic brewer's foreign parent.
Second, Canadian domestic brewers may choose to operate their own distribution facilities. Beers brewed domestically by Labatt and Molson are distributed in Alberta by Brewers Distributor Ltd. (BDL), a private joint venture company owned by Labatt Breweries of Canada and Molson Breweries. BDL also distributes beer in British Columbia, Saskatchewan, Manitoba, the Northwest Territories, and the Yukon. (13) Big Rock has sales and distribution facilities in Edmonton and at its brewery in Calgary. (14) Sleeman Breweries distributes its products through a warehouse located in Calgary.
The four major brewers and Connect Logistics distribute beer products to all Class A liquor licensees in Alberta. (15) The geographic area that is the focus of this article is the city of Edmonton due to data collection constraints (discussed in the next section). The complete list of Class A licensees in Edmonton was obtained from the Alberta Gaming and Liquor Commission (AGLC) Web site in April, 2007. (16) This list provides the names and addresses of licensees, along with other information such as telephone and fax numbers. The list also indicates whether the licensee has a food primary (FP) license, a liquor primary (LP) license, or both. In total, there were 871 Class A licensees in Edmonton.
Finally, it is important to note that all liquor licensees in Alberta have access to the same wholesale price list. Suppliers are permitted to change their supply price every two weeks, and then a formula that includes the supply price, taxes, and the government's markup is used to calculate the wholesale price. Suppliers are not allowed to negotiate discounts on the wholesale price with licensees.
III. BEER DISTRIBUTION REGULATIONS AND INCENTIVES FOR EXCLUSIVITY
A. Canadian regulations regarding inducements
Many provinces have developed policies to address the possibility that the manufacturers and suppliers of alcoholic beverages may attempt to provide inducements to liquor stores, restaurants, bars, and other retailers to promote their products over those of their competitors or even to sell their products exclusively. (17) In Alberta, the relevant text of the Gaming and Liquor Act is in paragraphs 66(1) to 66(2): (18)
[paragraph] 66(1) Unless the regulations provide otherwise, no liquor supplier or liquor agency may enter into an agreement with a liquor licensee whose licence authorizes the sale of liquor to customers for their own consumption in which the licensee agrees to sell the liquor of the supplier or agency.
(2) Unless the regulations provide otherwise, no liquor licensee whose licence authorizes the sale of liquor to customers for their own consumption may enter into an agreement with a liquor supplier or liquor agency in which the licensee agrees to sell the liquor of the supplier or agency.
Guidelines issued by the provincial liquor agencies, which discuss their policy towards inducements, were obtained for Alberta, British Columbia, Saskatchewan, Ontario, and Newfoundland. The available guidelines suggest that the focus of regulations on inducements is the use of inducements to persuade a retailer to sell or promote the products of a particular supplier and not specifically inducements to achieve exclusivity. Alberta, British Columbia, Ontario, and Saskatchewan provide in their guidelines lists of items that suppliers are prohibited from giving to other licensees. Monetary payments and rebates are prohibited in all four provinces, as are gifts of additional product. (19)
In Alberta and British Columbia, current policy requires the use of "buy/sell agreements" when a manufacturer wishes to provide certain items to a retailer. According to the Alberta guidelines, a buy/sell agreement "allows the liquor supplier to give items to a licensee in return for the promotion of specific brands or types of liquor by the licensee." (20) In practice, under a buy/sell agreement a retailer can agree to purchase a certain amount of particular products from the manufacturer in exchange for permitted promotional items.
B. Inducement cases
The province of Alberta adopted the use of buy/sell agreements in 2000, after many licensed establishments were fined for violating the inducement provisions of the Gaming and Liquor Act. In April 2000, the AGLC concluded its investigation into marketing practices by Canada's two largest brewers, Labatt and Molson. The AGLC levied fines totaling $832,250 on 371 bars and restaurants for taking inducements, a violation of section 63(2) of Alberta's Gaming and Liquor Act. Another 89 bars and restaurants were sent warning letters regarding taking inducements, but were not fined. Molson and Labatt reportedly were fined $713,500 and $541,500, respectively. (21) There is no confirmation on the AGLC website, however, that the brewers' fines were paid. (22)
According to the Edmonton Journal, Molson and Labatt were giving bar and restaurant owners and managers free tickets to sporting events, sports equipment, and free trips in return for "preferential treatment" of their draft beer. (23) AGLC spokesman David Henning is quoted as saying, "In some cases it may have been exclusivity, where they may have been providing something to the licensee so that they would only sell their draft beer at an establishment." (24) In one article, it was suggested that "bar and restaurant owners were refusing to sell draft beer from smaller brewers because of the arrangements they had with the big breweries." (25) In another article, it is stated that "[o]fficials with Big Rock Breweries of Calgary, the province's largest independent, say they frequently complained about being squeezed out of bars and restaurants by these practices." (26)
The AGLC reports on its Web site violations of the Gaming and Liquor Act, (27) including the name of the licensee that was fined or warned, the city in which it is located, the amount of the fine, and the section of the Act that was violated. With respect to the inducement violations under section 63(2) in 2000, the list does not appear to be complete. Table 1 shows the numbers of Class A (i.e., bar and restaurant) violations of section 63(2) in 2000 for Calgary, Edmonton, and the Rest of Alberta that are reported on the AGLC website. The table breaks them down by chain and independent, (28) where a chain consists of outlets with the same name or that are members of the same corporate family of chains. The split between chain stores and independents is almost fifty-fifty. Almost two-thirds of the violations occur in Calgary and Edmonton. The chain outlets that are in violation in the Rest of Alberta are members of the same chains found to be in violation in Calgary and Edmonton.
The only other case in Canada where inducements may have been an issue involved a Competition Bureau inquiry into the Quebec beer market. This inquiry commenced in August 2000. It was in response to complaints that Labatt and Molson had negotiated agreements with licensees that contained "exclusivity clauses, shelf-space allocation clauses, clauses requiring establishments to sell certain brands at the same price as their competitors, clauses restricting advertising, and right of first refusal clauses." (29) The objective of the inquiry was to determine whether Labatt or Molson had violated the exclusive dealing or abuse of dominance provisions of the Competition Act. The Bureau defined the relevant product market as beer and the relevant geographic market as Quebec. The Bureau found that Labatt and Molson accounted for nearly ninety percent of beer sales in Quebec and that they had market power in the Quebec market. The Bureau found that Labatt and Molson had exclusive dealing agreements with some on-premise consumption licensees and that both draft and bottled beer might have been covered by the agreement. However, the Bureau was unable to conclude that the companies' practices led to a substantial lessening of competition.
IV. EXCLUSIVE DEALING: THEORY, EVIDENCE, AND EMPIRICAL PREDICTIONS
In a retailing context, exclusive dealing refers to a retailer signing a contract with a manufacturer or supplier of a product that it sells agreeing not to sell the products of other manufacturers or suppliers. Economic theory has focused on addressing why exclusive dealing is observed and its effect on welfare.
As discussed by Joseph Farrell, a standard Chicago analysis suggests that exclusive dealing should not be observed. (30) Suppose that an incumbent manufacturer and a retailer at first bargain over an exclusivity contract; if such an agreement is reached, wholesale and retail prices are chosen, while if an agreement is not reached, a new manufacturer enters to compete against the incumbent. Under certain assumptions, the joint surplus of the incumbent and retailer will not be greater under exclusive dealing than under common agency, and they will never agree to exclusivity. However, since exclusive dealing is observed in some settings, an important question for economic theory regards what extensions are required to yield exclusive dealing.
A recent overview of the economic theory of exclusive dealing is provided by Tim Sass, who divides the theoretical literature into groups based on the underlying reason for exclusive dealing: (i) incentive conflicts, (ii) foreclosure, and (iii) dampening of competition. (31)
1. INCENTIVE CONFLICTS One set of explanations compares the incentives of the retailer under common agency and exclusive dealing. One argument, put forward by Howard Marvel, regards free riding on an investment made by one of the manufacturers. (32) Suppose that a manufacturer invests in advertising which increases demand for its product but also for rival products. A retailer selling the products of this manufacturer but also of a competitor may wish to encourage consumers to purchase the product of the other manufacturer with a lower wholesale price, on which product the retailer can earn a higher markup. Thus, exclusive dealing may prevent the rival manufacturer from free riding on advertising investments, in effect protecting the investing manufacturer's property rights. Benjamin Klein and Andres Lerner argue that retailer free riding can occur without investments such as advertising, since in general the amount of promotion of a manufacturer's product chosen by a retailer will be less than that desired by the manufacturer. (33)
Other sources of incentive conflict are moral hazard problems and adverse selection problems. B. Douglas Bernheim and Michael Whinston provide a model with moral hazard, in which undesirable incentives arise from differences in the degree of risk aversion of manufacturers and the retailer. (34) In David Martimort's article, incentive conflicts are created when the retailer has private information on demand or retailing costs. (35) Bernheim and Whinston also note that Marvel's argument can be formally modeled as a double moral hazard problem. (36)
2. FORECLOSURE A standard reference regarding foreclosure is the paper by G. Frank Mathewson and Ralph Winter. (37) In their model, there are two manufacturers and one retailer. One manufacturer (firm A) is assumed to have lower costs than the other, firm B. The two manufacturers offer contracts to the retailer and can offer either exclusive dealing contracts or common agency contracts. The only variable in the contract is the wholesale price; there are no fixed fees, and wholesale prices are constant over output. If the manufacturers choose to offer exclusive dealing contracts, they then compete with each other through the wholesale price. Firm A will be chosen, but must offer a wholesale price no more than the wholesale price that makes the retailer indifferent between choosing firm A's exclusive dealing contract and an exclusive dealing contract with firm B with a wholesale price equal to firm B's marginal cost. Whether exclusive dealing will be chosen will depend on how firm A's profits under this contract compare to the profits under the equilibrium common agency contract.
Other models of foreclosure have been offered. Bernheim and Whinston propose a model in which a manufacturer may have an incentive to sign an exclusive contract with a retailer in one market to prevent another manufacturer from entering a future market. (38) Essentially, because of scale effects, successful operation of the new entrant in the future market requires that it also operate in the first market. A similar foreclosure story based on scale effects is presented by Eric Rasmusen, Mark Ramseyer, and John Wiley, whereby exclusive dealing arises from coordination problems among buyers. (39) Philippe Aghion and Patrick Bolton consider a situation in which exclusive contracts award a penalty to the incumbent if the buyer chooses to deal with a new entrant and find that these contracts can lead to a lack of entry when such entry would be efficient. (40)
3. DAMPENING OF COMPETITION Another explanation for exclusive dealing, proposed in papers by Y. Joseph Lin and Margaret Slade, is that exclusive dealing softens competition between upstream manufacturers. (41) Slade argues that if a single bar or restaurant sells competing beers, then the perceived own-price demand elasticities of these products will be greater than if the beers were sold in separate establishments. Similarly, according to Lin, if two brewers sell through different retailers, then a reduction of wholesale price by one will be followed by a reduction in its retailer's price, which will in turn lead to a reduction in the competing retail price. Alternatively, if both brewers sell through the same retailer, then a reduction in one wholesale price will lead to less reduction in the competing retail price; hence perceived own-price demand elasticities for the brewers will be greater under a single retailer. Choosing exclusive dealing effectively differentiates the products of the two brewers.
A similar argument is made in the spatial model of partial foreclosure given by David Besanko and Martin Perry. (42) In this model, there are two manufacturers. Under common agency, the products of both manufacturers are available at all retail stores, while under exclusive dealing, a particular retailer sells the product of only one manufacturer. Under exclusive dealing, because neighboring retailers no longer sell the same line of products, they set higher prices. As well, consumers face greater transportation costs, since the product of a particular manufacturer is not available at every store. However, if fixed costs of entry for a retailer are higher under common agency, cost savings from exclusivity may offset increased retail prices and transportation costs.
B. Empirical work on exclusive dealing
Much of the empirical literature on the incidence of exclusive dealing has been focused on the insurance industry. (43) Marvel argues that if exclusive dealing is used to prevent free riding on advertising investments, one should see exclusive dealing used for types of insurance that require promotion and advertising at the company level and independent agents used for insurance that relies on direct contact from salespeople. (44) Evidence of greater reliance on exclusive dealing in selling personal insurance than commercial insurance, and evidence that companies that spend more on advertising are more likely to use exclusive dealing, are offered to support Marvel's hypothesis. (45)
Tim Sass and Micha Gisser argue that only large insurance companies will use exclusive agents, since small companies cannot generate enough business to compensate agents for being exclusive. (46) They find that company size is a significant determinant of whether an insurance company uses exclusive agents, as is population density. Other empirical studies of exclusive dealing in the insurance industry include one by Won-Joong Kim, David Mayers, and Clifford Smith, as well as one by Laureen Regan and Sharon Tennyson. (47)
Jan Heide, Shantanu Dutta, and Mark Bergen analyze survey data for 147 manufacturing firms. (48) The authors find that firms that believe rivals may be able to free ride off their promotional expenditures or support for their distributors are more likely to use exclusive dealing. As well, they find that exclusive dealing is less likely when managers believe that customers have a preference for distributors who offer products from multiple manufacturers. The authors find no relationship between firm size and exclusive dealing and no relationship between the perceived likelihood of new entry and exclusive dealing.
A few studies examine exclusive dealing in the beer industry. Sass studies the use of exclusive dealers by brewers in the United States in 1996. (49) He finds that distributors with larger sales territories, and whose main suppliers have greater state-level market shares, are more likely to be exclusive to a single supplier. Sass also finds that exclusive dealing is less likely in states that restrict outdoor advertising. Finally, exclusive dealing is less likely the longer the distributor has been owned by the same family.
Asker studies exclusive dealing between brewers and distributors in the Chicago area in 1994 and finds that exclusive dealing contracts yield lower costs for brewers. (50) In a second paper, Asker constructs tests of whether exclusive dealing between brewers and distributors results in foreclosure and rejects the foreclosure hypothesis. (51) Slade studies restrictions on tied houses in the U.K. and the effects of the subsequent divestiture of many public houses. (52) Slade finds that the forced divestitures lead to higher retail prices, but that brewer profits fell or did not change. It is argued that exclusive dealing will be more likely when the products of the competing manufacturers are closer substitutes.
C. Alternative explanations for exclusivity in beer retailing and empirical predictions
The theoretical and empirical exclusive dealing literature offers a number of alternative explanations for exclusive dealing and variables that might be helpful in distinguishing them. Unfortunately, a number of the suggested variables are either inappropriate for examining exclusive dealing in beer or unavailable for our market. The focus of this section is to consider potential explanations for why a retailer would be observed selling only the draft beer of a single supplier and to discuss predictions that can be taken to the data.
1. EFFICIENCY Under an efficiency hypothesis, a bar or restaurant might choose to offer the brands of a single brewer because it is less costly for it to deal exclusively. Exclusivity and geographic concentration of customers might also allow the brewer to reduce transportation costs by distributing larger product volumes to a particular area in the city.
2. CHANCE Under a chance hypothesis, exclusivity is not the result of inducements; some outlets would be observed as exclusive to one brewer simply on the basis of random choice of brands. For example, suppose that two major brewers each sell eleven different draft beers, that a restaurant has two taps, and that the restaurant chooses its beers randomly from the twenty-two major brand beers available. (53) Then, the probability it will choose two beers from the same brewer is twenty-four percent.
Under this hypothesis, small establishments will be more likely to be exclusive than large establishments. Continuing the previous example, if a restaurant wished to choose four beers instead of two, the probability of being exclusive to a single brewer would fall to five percent, while if it were choosing six beers, the probability would fall to less than one percent. Hence, the more beers that an establishment offers, the less likely it is that it will be exclusive to a single brewer under the chance hypothesis.
Finally, one might observe exclusivity under the chance hypothesis if a bar or restaurant's patrons have preferences for a particular type of beer. For example, suppose that a bar or restaurant serves a clientele with strong preferences for European imports, and suppose that it chooses two European imports from the twenty-two brands offered by the two major brewers. Suppose one brewer has five European imports, while the other has two. The probability that the bar or restaurant will be exclusive to the first brewer under random choice is forty-eight percent, while the probability that it will be exclusive to the second brewer will only be five percent.
3. STRONG FORECLOSURE Under a strong foreclosure hypothesis, the major brewers that dominate the market monopolize the taps in bars and restaurants with their own brands, regional brands, and partner brands. Regional brewers, microbrewers, or imports, if they are observed on tap at all, would be observed in either bars and restaurants with a small number (e.g., one or two) of taps, assuming a small number of taps reflects a relatively low demand for draft beer, or in specialty establishments (e.g., an Irish pub) that mainly serve imports or nonmainstream products.
4. DAMPENING OF COMPETITION Under a dampening of competition hypothesis, each major brewer offers its particular brands to bars and restaurants that are spatially differentiated. By concentrating sales outlets for their brands spatially, each major brewer might be able to charge higher wholesale prices for draft beer, and bars and restaurants would charge higher retail prices. Since a restaurant or bar may ultimately choose whatever beers it wishes, one would not expect to see every retailer in a region selling exactly the same beers. However, if brewers concentrate selling efforts by sending sales agents into selected areas, one would expect to see spatial patterns in brand presence. (54) In addition, if dampening of competition is sufficiently important to the major brewers as to produce significant spatial outcomes, it might also lead the major brewers to avoid competing with each other for business within particular establishments (whereas competing with the regional or microbrewers might be of second order importance).
5. INCENTIVE CONFLICTS Under a free riding hypothesis, the major brewers that dominate the market would establish exclusive arrangements with bars and restaurants to prevent them from free riding on the brewers' demand enhancement and brand promotion investments. In other words, the major brewers' brands and promotional investments might attract customers to a bar or restaurant, and in the absence of exclusive dealing, a nonmajor brand with a larger gross margin could receive more aggressive marketing by the retailer or even be substituted for one of the major brewer's brands. (55)
The two major Canadian brewers, Labatt and Molson, engage in extensive television advertising and other promotional activity. Labatt, for example, sponsors both professional and amateur sports, including baseball, football, and hockey. It supports sporting organizations at the neighborhood, municipal, and provincial levels, as well as local charities. (56) Molson engages in similar activities, sponsoring all six Canadian NHL teams, the NBA and the Toronto Raptors, and the CFL and five of its teams. Molson states that it "gives to the community through a variety of initiatives that help build consumer loyalty and establish the Molson presence in large and small communities across Canada." (57) These promotional activities are clearly costly for both brewers; they can enhance demand for both beer and the brewers' brands, and are subject to free riding.
While the brewers might attempt to induce bars and restaurants to accept an exclusive dealing arrangement, the gains from being nonexclusive might outweigh what the brewers are willing to pay for the arrangement. For example, the gains from free riding could be larger for bars with multiple taps than for restaurants, given that a larger proportion of their revenues would be from beverage alcohol than from food. In addition, nonexclusivity might be observed in bars, given that bars exist as places where people go to drink, and they may be expected to offer greater product variety. (58) The sales gains from offering variety might outweigh the benefits of being exclusive to a particular brewer. Then too, if restaurant patrons have no strong preference for particular brands of beer, a restaurant owner might find it easier to free ride by offering substitutes for the major brands. The brewers of these brands might have a greater incentive to prevent free riding in restaurants through the use of exclusive arrangements.
More exclusivity might be observed for chain restaurants than for independent restaurants or bars for double moral hazard reasons. A chain can negotiate an exclusive arrangement with a brewer and in return receive inducements as well as commitments with respect to promotion of its brands and service for the account. If the brewer shirks his duties under the agreement, the costs could be the loss of the exclusive arrangement for the entire chain. The brewer, on the other hand, might be prepared to offer substantial inducements and other benefits for exclusivity by a successful chain and threaten termination of the benefits if the chain shirks. A brewer will be less concerned about a chain than an independent reneging on an agreement because (1) it is less costly to monitor chain adherence to an agreement than an equivalent number of independent outlets, (2) punishment costs for shirking can be higher, and (3) the chain may have strong incentives to maintain its reputation for not shirking its commitments.
These are the major explanations for exclusivity that we will entertain. In what follows, we will attempt to find whether the data on draft beer offerings in Edmonton, Alberta, are consistent with one or more of them.
V. DATA DESCRIPTION
At an early stage of the data collection, it was determined that it would be feasible only to collect data on which draft beers were available in each establishment. Data on an establishment's beer offerings had to be obtained largely by visiting the establishment. (59) While the draft beer selection could be observed, the bottle selection frequently could not.
It should also be noted that in the early data collection where bottle brands were observed, exclusivity did not appear to be an issue. (60) There are reasons to expect the brewers to view exclusive arrangements for draft beer as more important than for bottles: (1) bars and restaurants may more easily substitute one draft beer for another, either on purpose or by accident, (61) and (2) by not imposing exclusive arrangements in bottled beer, the brewers can claim that they are not engaged in exclusive dealing. (62)
Due to the time required to visit a large number of bars and restaurants, data were not collected for the entire city. Rather, data were collected for most establishments in eleven postal code forward sorting areas (FSAs). These areas are identified in bold type face on the map of Edmonton in the figure. These FSAs were chosen to cover all geographic regions of the city and also to represent regions with different demographics. (63) In total, these 11 regions contain 288 (33%) of all of the Class A licensees in Edmonton. (64)
Of the 288 licensees listed by the AGLC, data regarding whether draft beer was offered were obtained for 252 (88%). (65) Of these 252 establishments, 159 (63%) were found to offer draft beer. The breakdown of these 159 establishments by FSA is given in table 2.
Of the licensees in the sample area, 44% were identified as members of retail chains, compared to only 33% of the licensees for the city as a whole. Of the 159 locations in the sample that serve draft beer, 59% are members of retail chains. An establishment is considered to be exclusive to a particular brewery if it sells on tap only the beers of that firm, excluding any house brands. In total, 42% of the establishments in our sample with beer on tap were found to be exclusive to either Molson or Labatt; an additional 8% of the sample establishments are exclusive to either Big Rock or Alley Kat. Clearly, it is not the case that most establishments are exclusive.
Licensees that have been identified as violating the inducement provisions of the Gaming and Liquor Act may have been demonstrating a taste for exclusivity, assuming that the violation was for accepting inducements in return for selling the draft beer of one of the two major brewers. It is of interest to see to what extent bars and restaurants that were in violation in 2000 and that are in our sample are exclusive today, given that buy/sell agreements might be used to achieve de facto exclusivity. We have identified all licensees who were listed by name as having been in violation in April/May 2000 or who are part of a chain that had at least one member in Edmonton found in violation.
Table 3 reports the number and percentage of establishments found to be exclusive to either Molson or Labatt, according to whether they were previously found in violation. For example, twenty-four of the establishments previously found in violation were exclusive to Molson or Labatt in our data; this represents 51% of all establishments in our sample that were previously found in violation. In contrast, only 38% of establishments not previously found in violation are currently exclusive to either Molson or Labatt. (66) Likewise, if we look only at retail chain establishments, 50% of chain establishments previously found in violation are currently exclusive, compared to 44% of chain establishments not previously found in violation. Hence, only about half of the restaurants or bars that were previously found to have accepted inducements are currently exclusive to one of the two major brewers.
In terms of overall violation of the inducement provisions, 120 (13%) of the 871 current licensees were found in violation in 2000 or are members of chains found in violation. In our sample area, this count becomes 48 (17%), suggesting that our sample is representative of the city as a whole in this regard. Of the 159 locations in our sample with draft beer, 47 (30%) were found in violation in 2000 or are members of chains found in violation.
VI. EXCLUSIVE DEALING BY BARS AND RESTAURANTS IN EDMONTON
In this section, we consider to what extent our sample of bars and restaurants in Edmonton exhibits exclusivity with brewers and whether the pattern of exclusivity is consistent with one or more of the explanations tendered in section IV.
In Alberta, the efficiency explanation seems without merit. As noted in section II, domestically brewed beer from the major brewers (Molson and Labatt) is distributed through a joint venture of the two firms, Brewers Distributor Ltd. Therefore, a retailer deals with only one distributor, regardless of whether it sells the beer of one or both major brewers. In addition, all imported beers and some beers from smaller domestic breweries are handled through Connect Logistics, which also distributes all wines and spirits. Hence, restaurants and bars are already doing business with Connect Logistics and have no cost-based reason to avoid brands that it distributes.
Under the chance hypothesis, the degree of exclusivity should fall off sharply as the size of an establishment, here measured by the number of taps, increases. It is indeed the case that less exclusivity is observed among establishments with more taps: while 45% of establishments with two to six taps are exclusive to either Molson or Labatt, exclusivity is observed for only 20% of establishments with more than 6 taps. (67) Looking only at restaurants, these numbers become 55% and 30%, respectively. However, it is nonetheless the case that the degree of exclusivity is much higher than would be expected if retailers were simply choosing a number of products at random from a list of all products offered. This higher degree of exclusivity could be the result of preferences or inducements.
The chance theory also suggests that if one of the major brands offers more high-end and expensive beers, while the other offers more mid-range and low-end products, and bars and restaurants have a preference for carrying either low- or high-end products, then one would be less inclined to observe exclusivity based on chance. To consider whether such a pattern exists, we ranked beers observed to be on tap at establishments in the sample, according to price, from lowest to highest. Because wholesale price data could not be obtained for Alberta, retail prices were obtained in May 2007 for both British Columbia and Ontario (68); the rankings for each province were then averaged. Draft beer prices could not be obtained and as a result prices for cans or bottles were used
It was found that the lowest-priced beers are evenly distributed among the two major brewers, with Molson accounting for five of the ten lowest priced beers, and Labatt accounting for four. Hence, it would be roughly equally likely for establishments offering low-end products to choose Molson as Labatt. On the other hand, Labatt offers six of the ten most expensive beers, compared to two for Molson. Therefore, while one might expect high-end establishments to offer more Labatt products than Molson products and to be more likely to be exclusive to Labatt, preferences for high-end over low-end beers seem unable to explain the establishments that are exclusive to Molson.
C. Strong foreclosure
Regional breweries and microbreweries do have a presence on tap in Edmonton, with 30% of establishments offering Big Rock products, 14% offering Sleeman, and 4% offering Alley Kat. Further it is not the case that these products are being offered only in very small establishments. Of the forty-seven bars and restaurants observed to carry Big Rock products on tap, eighteen (38%) had two to six taps, twenty-five (53%) had more than six taps, and only four (9%) had a single tap. Likewise, 45% of establishments observed to carry Sleeman products had two to six taps, with the remaining having more than six taps. Alley Kat, as well, is observed in establishments with tap counts ranging from one to over six. Hence, there is no evidence to suggest that these brands have been foreclosed from all establishments except for very small bars and restaurants.
It is also the case that the products of Big Rock and Sleeman tend to appear in bars alongside draft beers from the major brewers and are not confined to specialized establishments. Eighty-one percent of all establishments selling Big Rock draft beer also sell draft beer from either Molson or Labatt. For Sleeman, this number is 95%. For Alley Kat, only 29% of establishments also sell Molson or Labatt. However, because so few locations sell Alley Kat, this number is not very meaningful.
D. Dampening of competition
Under the dampening of competition hypothesis, the two major brewers seek to serve different areas in the city as a method of avoiding a bidding war for exclusivity and to support a higher wholesale price for their products. Tables 4 and 5 present, for seven different regions in the city, the number and percentage of establishments offering Molson and Labatt products, and the number and percentage offering Molson but not Labatt products, and vice versa. These regions are either FSAs or aggregations of adjacent FSAs. Table 4 considers all establishments, while table 5 looks only at independents. Table 4 does suggest some geographic patterns in terms of where the brewers' brands are found. The percentage of establishments offering Molson but not Labatt varies from 13% to 52%, while the percentage offering Labatt but not Molson varies from 12% to 54%. However, some of this variation may be explained by differences in chain presence and demographics across regions. As well, the small number of independents present in many of these regions prevents drawing any strong conclusions from the numbers reported in table 5. Hence, these tables do not provide convincing evidence of geographic patterns being generated by an attempt to dampen competition.
E. Incentive conflicts
The first basic prediction of a free riding theory of exclusivity, in which brewers wish to induce retailers to engage in more promotion for their products, is that exclusivity should be carried out primarily by the largest brewers. (69) This turns out to be the case. Eighty-four percent of establishments in our sample that are exclusive to some brewery are exclusive to either Molson or Labatt. (70) Further, a higher fraction of the establishments that carry Molson and Labatt are exclusive than of the establishments carrying Big Rock and Sleeman. Forty-seven percent of establishments that sell Molson or Labatt are exclusive to one of these two brands. In comparison, only eight percent of places selling Big Rock are exclusive to it; there are no exclusive Sleeman outlets.
As discussed earlier, free riding theories seem to make no strong predictions regarding whether exclusivity will be more prevalent in restaurants than in bars. However, as indicated in table 6, at least for establishments with more than two taps, restaurants are more than three times as likely to be exclusive than bars. This would suggest that, if free riding is the incentive for exclusivity, then for bars it is outweighed by the desire for variety, or the attractiveness of free riding to bars is so great that it cannot be compensated for through inducements.
Under a free riding explanation for exclusivity, one would expect more exclusivity among chains than independents. Table 6 illustrates that this holds for outlets with more than two taps; 39% of chain outlets with more than two taps are exclusive, compared to only 16% of independent outlets with more than two taps. (71)
To examine whether exclusivity depends on whether an outlet is a bar or restaurant or whether it is part of a regional or national chain, the following probability unit model has been estimated:
Prob([I.sub.i]=l) = [phi]([x.sub.i][beta]),
where [I.sub.i] is a variable that equals one if establishment i is exclusive, [x.sub.i] contains observations for establishment i for relevant variables, [beta] is the vector of coefficients, and [Phi] is the cumulative density function for the standard normal distribution.
In the model, the probability that an establishment is exclusive is a function of the following variables:
[fp.sub.i] = 1 if establishment i has a food-primary license, and 0 otherwise;
[fplp.sub.i] = 1 if establishment i has a food-primary/liquor-primary license, and 0 otherwise;
[national.sub.i] = 1 if establishment i is part of a national chain, and 0 otherwise;
[over6.sub.i] = 1 if establishment i has more than 6 taps, and 0 otherwise;
[chainsize.sub.i] = the number of Canadian outlets in the establishment's chain, if it is part of a chain, and 0 otherwise.
In addition, dummy variables were included for the first six regions, using the region T6K+T6L as the control group.
Variables [fp.sub.i] and [fplp.sub.i] are dummy variables that control for whether an outlet is a restaurant or a restaurant that has a lounge. Given the results in table 6, one might expect the coefficient on [fp.sub.i] to be positive.
The variable [national.sub.i] is a dummy variable that controls for whether the bar or restaurant is a national chain, where a chain is defined as national if it has outlets in Alberta and at least one other province. [Chainsize.sub.i] measures the size of the chain in Canada. A positive sign is expected on the coefficients of both variables, given the potential gains to brewers from having an exclusivity arrangement with national chains and large chains. [Over6.sub.i] is another dummy variable, indicating whether an outlet has more than six taps. If it does, one would expect less exclusivity. Having that many taps could indicate a taste for offering variety that might not be satisfied by one brewer's products. (72)
Results of estimation are presented in table 7. Two specifications are estimated. In the first specification, all observations are used. The coefficients are jointly significant at the 1% level. In the first specification, the coefficients on national and over6 are both significant at the one percent level with the expected sign and jointly significant as well. The coefficients on the region dummy variables are not jointly significant at reasonable significance levels, suggesting that the degree of exclusivity does not vary by FSA, once one controls for basic features of the establishment. (73) The coefficients on fp and fplp are not significantly different from zero at reasonable levels of significance, indicating that when other variables are controlled for, bars are not less likely to be exclusive than establishments with food primary licenses. Hence, the association noted earlier between exclusivity and whether the establishment is a restaurant was likely picking up the relationship between exclusivity and whether the establishment is part of a national chain.
Unexpectedly, the coefficient chainsize is negative and significant at the ten percent level. However, further inspection suggests that this is the result of an outlier. The Boston Pizza chain, which represents 9 establishments in our sample, has by far the largest number of Canadian outlets: 247 compared to a sample average of 53. (The next largest chain in our sample is Swiss Chalet with 172 Canadian outlets.) Boston Pizza establishments are all nonexclusive, carrying both Molson and Labatt products. When the nine Boston Pizza establishments are dropped from the sample, the coefficient on chainsize becomes positive and insignificant, while the other results are unchanged; these results are reported in specification 2.
To aid in the interpretation of the coefficients on the national and over6 variables, we computed the effect on the probability of being exclusive of switching these variables from zero to one. For this exercise, consider the hypothetical case in which all of the regional dummies are set to zero (so that the establishment is in the T6K+T6L region) and in which the establishment has an food-primary license. The first specification was used for these calculations. It turns out that the effect of the national and over6 variables is quite large. Holding all other variables constant, switching an establishment to a national chain increases the probability that it is exclusive by 0.30, while switching an establishment from no more than six taps to over six taps decreases the probability of being exclusive by 0.43.
In general, our results suggest that whether an establishment is exclusive is associated with whether it is part of a national chain and the number of taps in the establishment. These results are consistent with the free riding explanation for exclusive dealing and with the payoff from offering a large variety being greater than the payoff from being an exclusive dealer for establishments whose demand supports a large number of taps. There is no evidence that restaurants are more likely to be exclusive once we control for other variables. Likewise, there is no indication that whether an establishment is exclusive varies spatially.
A further possibility is that, instead of offering inducements to obtain complete exclusivity, brewers might use inducements to seek a certain number or proportion of available taps. An agreement that ensures a supplier a specific proportion of a retailer's purchases has been labeled market share exclusion by Mikko Packalen, who demonstrates that such contracts can be profitable for the supplier and anticompetitive. (74)
To consider this possibility, we redefined a restaurant or bar as exclusive to a specific brewer if at least seventy-five percent of the taps offer the beer of that brewer. Under this new definition, the fraction of licensees in the sample found to be exclusive increases from forty-three percent to fifty-two percent. The qualitative results of the analysis regarding variables associated with exclusivity still hold. (75)
VII. SUMMARY AND CONCLUSIONS
It is well known that there are alternative explanations for exclusive dealing. In the case of draft beer served by bars and restaurants, there could well be an efficiency explanation for the exclusivity arrangement between a bar or restaurant and a particular brewer. It could be a mistake to view such an arrangement as anticompetitive or exclusionary or to create regulations that inhibit its existence. Nevertheless, provinces in Canada and the U.S. government have had restrictions on a brewer's use of inducements either to achieve exclusivity or to encourage the sale of its products. In light of the uncertainty surrounding the reasons for exclusivity in beer distribution, it is important to collect data from a jurisdiction, such as the province of Alberta, which allows inducements (through buy/sell agreements) that could result in exclusivity, and to examine whether the data are consistent with one or more of the proffered explanations for exclusive arrangements. A city in the province of Alberta is chosen for this purpose because (1) Alberta allows inducements (through buy/sell agreements) that could result in exclusivity, and (2) prior to the legalization of buy/sell agreements in 2000, hundreds of on-premise consumption establishments were fined for taking inducements from brewers to sell their products, which might reveal a taste for dealing exclusively with brewers. Rejection of the anticompetitive explanations could signal a need to revise policy regarding exclusive dealing in beer distribution. (76)
Five alternative explanations for observations of exclusive dealing by bars and restaurants with a particular brewer are considered: (1) efficiency, (2) chance, (3) strong foreclosure, (4) dampening of competition, and (5) incentive conflicts. To examine these explanations, data on draft beer offerings from a sample of on-premise consumption establishments in the city of Edmonton, Alberta, were collected.
An efficiency explanation is not supported given the way that beer is distributed in Alberta. Observing exclusivity because of random choices of beers by bars and restaurants is rejected by the data. The degree of exclusivity is much higher than would be expected on the basis of random choice. The chance explanation is also inconsistent with the amount of exclusivity observed for the brewers in their lower-end products, where there are multiple offerings from both of the major brewers. With respect to the strong foreclosure hypothesis, it is clearly not the case that the products of the principal rivals of the major brewers appear only in bars and restaurants that do not carry the products of the two major brewers. Nor is there any convincing evidence of geographic patterns being generated by an attempt by the major brewers to dampen competition.
The incentive conflict explanation for exclusive dealing is the one that is broadly consistent with the data. As predicted, exclusivity is primarily carried out by the largest brewers. National chains are more likely to be exclusive, while establishments that have more than six taps are not. When other variables are controlled for, bars are not less likely to be exclusive than establishments with food primary licenses.
Achieving and maintaining an exclusivity relationship between a bar/restaurant and brewer in Alberta is not easy. Buy/sell agreements might be written to achieve de facto exclusion, but contracts cannot be written that would exclude a rival brewer. Even where gains to brewers from having an exclusive relationship would seem to be large (with establishments whose demand can support a large number of taps), the establishment's business model might require a variety that cannot be provided by a particular brewer's products. The costs of changing brands in a bar or restaurant to become nonexclusive, other than some menu costs and possible loss of inducements, are also likely to be low. That we observe as much exclusivity as we do in Edmonton's bars and restaurants attests to the weight given to reducing possibilities for free riding, at least with respect to some market participants.
BY ANDREW ECKERT AND DOUGLAS S. WEST, Department of Economics, University of Alberta.
AUTHORS' NOTE: We wish to thank the editors, an anonymous referee, and seminar participants at the Competition Bureau and the Alberta Industrial Organization Conference for their helpful comments, McKay White and Andriy Maslo for research assistance, and the Social Sciences and Humanities Research Council for financial support.
(1) Howard Marvel, Exclusive Dealing, 25 J.L. & ECON. 1 (1982).
(2) Jonathan Lave, The Law and Economics of de Facto Exclusive Dealing, 50 ANTITRUST BULL. 143 (2005), discusses the use of economic inducements by a monopolist to achieve de facto exclusive dealing. It should be noted that provincial regulations might prohibit agreements that imply de facto exclusive dealing, although enforcement of such restrictions could be both costly and difficult.
(3) See Brewers Fined $1.2M for Illegal Inducements, CALGARY HERALD, May 7, 2000, at A5.
(4) See Competition Bureau, Inquiry into the Quebec Beer Market, available at http://competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/00786.html.
(5) See Molson Coors Brewing Co., Annual Report (Form 10-K), at 8 (Mar. 10, 2006).
(6) See InBev: 2006 Company Profile Edition 2, JUST-DRINKS, Nov. 2006, at 1.
(7) Id. at 11.
(8) See Labatt, http://www.labatt.com/english/lbc-company/lbc _history_3.htm (last visited May 29, 2009).
(9) See Molson Coors 2006 Company Profile Edition 2, JUST-DRINKS, Nov. 2006, at 12-13.
(10) Molson Coors Brewing Co., supra note 5, at 4-6. According to Molson Coors's annual report, two products, Molson Canadian and Coors Light, represent approximately 45% of Molson's Canadian sales volume.
(11) See Big Rock Brewery Income Trust, Annual Report 2006, at 8, available at http://media.integratir.com/T.BR.UN/financials/2006%20Annual %20Report.pdf.
(12) See Sapporo Acquisition of Sleeman on Tap, CBC NEWS, Aug. 11, 2006, available at http://www.cbc.ca/money/story/2006/OS/11/sapporo-sleeman.html.
(13) See Brewers Distributor Ltd., http://www.bdl.ca/welcome.htm.
(14) Big Rock Brewery Investment Trust, supra note 11.
(15) There are five classes of liquor licenses issued in Alberta. Class A licenses permit the sale of alcoholic beverages for onsite consumption at premises open to the public, including restaurants and bars. Class B licenses pertain to the onsite consumption of alcoholic beverages at places where people pay an entrance fee or purchase a ticket to use the facility; such establishments include sports stadiums and theaters. Class C licenses are required for private clubs, while Class D licenses are required by liquor stores for the sale of alcoholic beverages for consumption off-premise. A Class E license is required to manufacture liquor.
(16) Alberta Gaming & Liquor Commission, Liquor Licensee Service Providers Database, available at http://www.aglc.gov.ab.ca/aglc_public /aglc_site/liquor/licensee_list/lsearch.jsp.
(17) There does not seem to be a common definition of an "inducement" in this context. Nova Scotia, in its Liquor Control Act, R.S.N.S., ch. 260, [section] 1, defines an inducement as "any act by a liquor manufacturer or liquor representative to persuade a licensee or retailer through any benefit including, but not limited to, cash or liquor products, to buy more of a particular product than would be bought under normal circumstances, to the detriment of other manufacturers."
(18) In the Gaming and Liquor Act, R.S.A. 1996, ch. G-0.5, these paragraphs were numbered 63(1) and 63(2).
(19) In the United States, exclusionary arrangements in the distribution of beverage alcohol are dealt with under the Federal Alcohol Administration Act, 27 U.S.C.A. [section] 205(b), as well as in the liquor retailing regulations of various states. Exclusivity arrangements that a brewer might have with its wholesale distributors have given rise to a number of U.S. antitrust cases. Exclusive arrangements in wholesale distribution have not been an issue in Canada. In Canada, brewers distribute their products through the AGLC warehouse, through their own warehouses, or through wholesale distribution facilities that are jointly owned by two or more brewers.
(20) See AGLC, Policy Guidelines--Product Promotions in Licensed Premises, 3 (May 16, 2005), appended to the RETAIL LIQUOR STORES OPERATING GUIDELINES, available at http://www.aglc.gov.ab.ca/pdf/handbooks/retail_liquor_stores.pdf.
(21) See Fines $832,000 in Beer Probe, EDMONTON J., May 12, 2000, at A7.
(22) In an article by Larry Johnsrude, Molson, Labatt Together Fined $1.2M Over Inducements: Largest Liquor Infraction Penalties Ever, Gov't Says, EDMONTON J., May 7, 2000, at A6, AGLC spokesman David Hennig is reported to have said that both breweries have paid their fines.
(23) Fines $832,000, supra note 21. See also Larry Johnsrude, Clear Beer Rules Urged: Brewers Want Specific Guidelines on Promotion, EDMONTON J., May 9, 2000, at A7.
(24) Brewers Fined, supra note 5, at A5.
(25) Fines $832,000, supra note 21.
(26) Johnsrude, supra note 22.
(27) See Alberta Liquore & Gaming Commision, Disciplinary Board Decisions, http://www.aglc.gov.ab.ca/aglc_public/aglc_site/da/dasearch.jsp.
(28) Calgary and Edmonton are Alberta's two largest cities, with 2006 census populations of 988,193 and 730,372, respectively. Alberta had a 2006 population of 3,290,350. There were also some Class B (recreational facility), Class C (clubs or institutions), and Class D (liquor store) licensees found in violation, but only seven Class B, seventeen Class C, and two class D licensees are listed.
(29) See Competition Bureau, supra note 4.
(30) Joseph Farrell, Deconstructing Chicago on Exclusive Dealing, 50 ANTITRUST BULL. 465 (2005).
(31) Tim Sass, The Competitive Effects of Exclusive Dealing: Evidence from the U.S. Beer Industry, 23 INT'L J. INDUS. ORG. 203 (2005).
(32) Marvel, supra note 1.
(33) Benjamin Klein & Andres Lerner, Procompetitive Justifications for Exclusive Dealing (2006) (unpublished manuscript, available at http://www .usdoj.gov/atr/public/hearings/single_firm/docs/219980.pdf).
(34) B. Douglas Bernheim & Michael D. Whinston, Exclusive Dealing, 106 J. POL. ECON. 64 (1998).
(35) David Martimort, Exclusive Dealing, Common Agency, and Multiprincipals Incentive Theory, 27 RAND J. ECON. 1 (1996).
(36) See Bernheim & Whinston, supra note 34.
(37) G. Frank Mathewson & Ralph A. Winter, The Competitize Effects of Vertical Agreement: Comment, 77 AM. ECON. REV. 1057 (1987).
(38) Bernheim & Whinston, supra note 34.
(39) Eric B. Rasmusen, J. Mark Ramseyer, & John S. Wiley Jr., Naked Exclusion, 81 AM. ECON. REV. 1137 (1991).
(40) Philippe Aghion & Patrick Bolton, Contracts as a Barrier to Entry, 77 AM. ECON. REV. 338 (1987).
(41) See Y. Joseph Lin, The Dampening-of-Competition Effect of Exclusive Dealing, 39 J. INDUS. ECON. 209 (1990), and Margaret Slade, Beer and the Tie: Did Divestiture of Brewer-Owned Public Houses Lead to Higher Beer Prices?, 108 ECON. J. 565 (1998).
(42) David Besanko & Martin K. Perry, Exclusive Dealing in a Spatial Model of Retail Competition, 12 INT'L J. INDUS. ORG. 297 (1994).
(43) Another review of the empirical literature on exclusive dealing is provided by Sass, supra note 31.
(44) See Marvel, supra note 1.
(45) In this subsection, the focus is on studies that consider the circumstances under which exclusive dealing will be employed. In some cases, the studies discussed will also address other questions such as the price effects of exclusive dealing; for the most part, these results will not be discussed.
(46) Tim Sass & Micha Gisser, Agency Cost, Firm Size, and Exclusive Dealing, 32 J.L. & ECON. 381 (1989).
(47) See Won-Joong Kim, David Mayers, & Clifford W. Smith Jr., On the Choice of Insurance Distribution Systems, 63 J. RISK & INS. 207 (1996), and Laureen Regan & Sharon Tennyson, Agent Discretion and the Choice of Insurance Marketing System, 36 J.L. & ECON. 637 (1996).
(48) Jan B. Heide, Shantanu Durra, & Mark Bergen, Exclusive Dealing and Business Efficiency: Evidence from Industry Practice, 41 J.L. & ECON. 387 (1998).
(49) Sass, supra note 31.
(50) John Asker, Measuring Advantages from Exclusive Dealing (2004) (unpublished manuscript, available at http://pages.stern.nyu.edu/~jasker /Paper2_ExclusiveAdvantages.pdf).
(51) John Asker, Diagnosing Foreclosure Due to Exclusive Dealing (2004) (unpublished manuscript, available at http://pages.stern.nyu.edu/~jasker /Paper1_ExclusiveForeclosure.pdf).
(52) Slade, supra note 41.
(53) This assumption can be relaxed without changing conclusions.
(54) Different brand preferences across space could also support a finding that is consistent with dampening of competition. If different brewers appeal to people with different income levels or educational background, for example, then one might observe certain brewers to be more prominent in certain geographic areas. For example, lower income areas might favor lower end and lower cost beers.
(55) This possibility was raised by Cecil Chance Scott in testimony in Adolph Coors Co. v FTC, 497 F.2d 1178 (10th Cir. 1974). Mr. Scott testified that "Well, sir, when a waitress takes an order and two kids order a Coke and three others order a beer and somebody orders something else and one type of beer and she gets it all on this tray and turns it a half around nobody knows what they got. We just like to have our beer served in our glass, in a good clean glass.... I only have objections when they start serving somebody's else's beer in my glass or serving my beer in their glass telling them it is their beer."
(56) See Labatt, In the Community, http://www.labatt.com/english/lbc _community/lbc_sports.htm (last visited May 29, 2009).
(57) See Molson, http://www.molson.com/en/community.aspx (last visited May 29, 2009).
(58) Given that consumers choose restaurants on the basis of the food they serve, they may not expect (or demand) that restaurants offer a large variety of draft beers.
(59) Some establishments were telephoned for the information, while others post the information on their Web sites. In addition, by checking multiple locations of restaurants of the same chain, it became clear that, with certain exceptions, establishments within a chain offered draft beer from the same breweries. Therefore, in some cases, an establishment's draft beer offerings were assumed based on observations at establishments in the same chain.
(60) In addition, the Alberta inducement case referred to exclusive arrangements in draft beer, not bottles.
(61) While Cecil Chance Scott suggested the possibility of draft beer substitution in the Coors case, Adolph Coors Co. v. FTC, 497 F.2d 1178 (10th Cir. 1974), it is intuitive that substituting a cheaper bottled beer for a branded bottled beer that is ordered would be more difficult. This same point was made by Howard Marvel, Book Review, 8 REV. INDUS. ORG. 127 (1993). The possibility of brand substitution in draft beer was used as a justification for exclusive dealing by brewers in 1902 in submissions with respect to a 1902 bill. See Tim Mulcare, Exclusive Dealing in New Zealand Alcoholic Beverage Markets (unpublished undated manuscript, available at http://www.iscr.org.nz/f260,5167/ 5167_exclusve_dealing_nz_alcohol_011298.pdf).
(62) In an article by Johnsrude, supra note 23, Scott Ellis, Vice-President of Molson, is reported to have said that it was unclear what the prohibition of "product exclusivity" meant. "'When you walk into a bar, you can purchase products on tap or by the bottle,' he said. 'There was never a situation where the only product available was by one manufacturer.'"
(63) These regions should not be interpreted as individual markets.
(64) In addition, other locations were visited. Thirty-three observations (34% of outlets) from-downtown FSA TSJ were obtained, as were thirty locations (27% of outlets) from FSA T6E. Finally, at least one location was visited of every restaurant or bar chain that could be identified in Edmonton. In total, observations were obtained for 45% of the Class A licensees in Edmonton. In this section, however, only observations from the identified eleven postal codes are discussed.
(65) The remaining thirty-six establishments were a combination of those that were permanently closed or closed for renovations, did not sell liquor products when visited, or were not open for business when visited. A small number were not visited, since they were known to be closed during data collection sessions.
(66) Using a Pearson [chi square] test of association, the null hypothesis of no statistical association between whether an establishment is exclusive to Molson or Labatt and whether it was previously found in violation cannot be rejected at the 5% level of significance. See ROBERT WINKLER & WILLIAM HAYS, STATISTICS: PROBABILITY, INFERENCE, AND DECISION, 825-33 (2d ed. 1975) for details of the test.
(67) Independence between whether an establishment is exclusive and whether it has two to six taps or more than six is rejected using a Pearson test of association.
(68) In British Columbia and Ontario, retail prices are set by the government using a formula that takes the supply price given to it by the supplier.
(69) See supra text accompanying notes 46-48.
(70) It should be noted that this finding could also result from the fact that outlets that want to offer a range of products would be better able to do so by offering the products of a larger brewer, given its more extensive product line.
(71) The association between whether an establishment is exclusive and whether it is a chain, and the association between whether an establishment is exclusive and whether it is a restaurant, are both statistically significant at the 5% level, using a Pearson test of association, for establishments with more than two taps.
(72) Over6, a dichotomous measure of the number of taps in an establishment, is used because the precise number of taps was either not collected or deemed inaccurate for certain establishments, particularly those with multiple islands and many taps. However, re-estimating the model for the 116 establishments (83% of the total sample) with precise tap count observations, and using the number of taps as opposed to Over6, yields the same basic conclusions: the only statistically significant coefficients are those on national and the tap count variable.
(73) The model was re-estimated without the regional dummy variables, with no important changes to the results for other variables.
(74) Mikko Packalen, Market Share Exclusion (2007) (unpublished paper, available at http://law.bepress.com/cgi/viewcontent.cgi?article=2061 &context=alea).
(75) Pearson tests of association yield similar results, with the exception that the association between whether an establishment is exclusive and whether it is part of a chain is significant at the five percent level using all establishments, but for establishments with more than two taps, this association is significant at the ten percent level.
(76) Economists have had a longstanding interest in distribution arrangements in the beer industry. See, e.g., J.C.H. Jones, Mergers and Competition: The Brewing Case, 33 CAN. J. ECON. 551 (1967), and the beer industry study by Kenneth Elzinga, Beer, in THE STRUCTURE OF AMERICAN INDUSTRY 72 (Walter Adams & James Brock eds., 2005).
Table 1 Inducement Violations in Alberta, 2000 Area Chain Independent Total Calgary 62 55 117 Edmonton 45 50 95 Rest of Alberta 51 65 116 Total 158 170 328 Table 2 The Number of Locations in Each FSA with Draft Beer FSA Number of Locations with Draft Beer (% of sample) T5A 13 (8%) T5K 21 (13%) T5L 16 (10%) T5M 13 (8%) T5T 26 (16%) T6G 15 (9%) T6J 24 (15%) T6H 9 (6%) T6L 8 (5%) T6N 10 (6%) T6R 4 (3%) Total 159 (100%) Table 3 Establishments in the Sample Exclusive to Molson or Labatt, by Whether They Were in Violation of the Inducement Provisions of the Gaming and Liquor Act in 2000 All Establishments Chain Establishments with Taps Only Found in Violation 24 (51%) 21 (50%) Not Found in Violation 43 (38%) 23 (44%) Table 4 Exclusivity and Brand Presence by Aggregated FSAs Region Molsan Labatt Molson Products Products Exclusive (no other) T5A 5 (38%) 10 (77%) 2 (15%) T5T 22 (85%) 12 (46%) 10 (38%) T5L + T5M 21 (72%) 11 (38%) 10 (34%) T5K 9 (43%) 12 (57%) 4 (19%) T6G 3 (20%) 7 (47%) 1 (7%) T6J+T6N+T6R 23 (61%) 20 (53%) 10 (26%) T6K +T6L 13 (76%) 11 (65%) 6 (35%) Region Labatt Molson, Labatt, Exclusive no Labatt no Molson (no other) T5A 6 (46%) 2 (15%) 7 (54%) T5T 1 (4%) 13 (50%) 3 (12%) T5L + T5M 3 (10%) 15 (52%) 5 (17%) T5K 3 (14%) 6 (29%) 9 (43%) T6G 3 (20%) 2 (13%) 6 (40%) T6J+T6N+T6R 4 (11%) 13 (34%) 10 (26%) T6K +T6L 3 (18%) 6 (35%) 4 (24%) Table 5 Exclusivity and Brand Presence by Aggregated FSAs, Independents Only Region Molson Labatt Molson Products Products Exclusive (no other) T5A 2 (29%) 6 (86%) 1 (14%) T5T 5 (100%) 4 (80%) 1 (20%) T5L + T5M 8 (61%) 5 (38%) 2 (15%) T5K 4 (36%) 6 (55%) 2 (18%) T6G 2 (20%) 4 (40%) 1 (10%) T6J+T6N+T6R 5 (71%) 4 (57%) 2 (29%) T6K +T6L 9 (75%) 6 (50%) 6 (50%) Region Labatt Molson, Labatt, Exclusive no Labatt no Molson (no other) T5A 4 (57%) 1 (14%) 5 (71%) T5T 0 (0%) 1 (20%) 0 (0%) T5L + T5M 1 (8%) 5 (38%) 2 (15%) T5K 0 (0%) 3 (27%) 5 (45%) T6G 0 (0%) 1 (10%) 3 (30%) T6J+T6N+T6R 0 (0%) 2 (29%) 1 (14%) T6K +T6L 2 (17%) 6 (50%) 3 (25%) Table 6 Chains v. Independents and Type of License by Exclusivity with Major Vrewers All Taps: Molson or Labatt All locations with taps Locations with 3 or more taps Chains 44 (47%) 29 (3%) Independents 22 (34%) 5 (16%) Food 35 (49%) 24 (49%) Food/Liquor 14 (35%) 6 (21%) Liquor 17 (36%) 4 (14%) Table 7 Probit Estimation Results; Dependent Variable = 1 if Establishment is Exclusive (1) (2) Variable Coefficient (St. Coefficient (St. Error) Error) National 1.104 *** (0.39) 0.98 ** (0.42) Size -0.000012 0.002 (0.005) Fp 0.54 (0.36) 0.52 (0.38) Fplp 0.04 (0.36) 0.14(0.37) Over6 -1.18 *** (0.31) -1.34 *** (0.33) T5T -0.47 (0.51) -0.82 (0.57) T5L + T5M -0.06 (0.47) -0.05 (0.51) T5K 0.14 (0.52) 0.15 (0.55) T6G -0.07 (0.61) 0.04 (0.63) T6J+T6N+T6R -0.57 (0.48) -0.64 (0.51) T5A 0.20 (0.55) 0.23 (0.59) Constant -0.22 (0.39) -0.18 (0.40) Number of Observations 139 130 *** indicates significance at the 1% level; ** indicates significance at the 5% level; * indicates significance at the 10% level.
|Printer friendly Cite/link Email Feedback|
|Author:||Eckert, Andrew; West, Douglas S.|
|Date:||Jun 22, 2009|
|Previous Article:||Merger policy and the Antitrust Modernization Commission: economic power and the miasma of efficiency.|
|Next Article:||Competition law and policy of Japan.|