Advertising Restrictions and Competition in the Children’s Breakfast Cereal Industry C. Robert Clarky October 17, 2006

Abstract This paper analyses the consequences of advertising in the children’s breakfast cereal market. I take advantage of the fact that advertising directed at children is prohibited in the Canadian province of Quebec to examine the nature of advertising, and to determine whether the restriction hinders competition. I show that prices are higher in Quebec than in Canadian provinces that permit advertising. This suggests that the informative role of advertising dominates any persuasive role, since the most likely explanation for the higher prices is that the restriction prevented …rms from announcing existence or characteristics and so from overcoming perceived di¤erentiation. If advertising is informative, restricting it should increase the market shares of older, better-known brands and decrease the shares of newer and/or less well-known brands. Empirical analysis supports this prediction: established brands have higher shares in Quebec than in the rest of Canada, and the opposite is true for non-established brands.

JEL classi…cation: L0, L1, M3 I would like to thank the editor, Dennis Carlton, for greatly improving the paper, as well as Todd Stinebrickner, Arthur Robson, Srihari Govindan, Pierre Léger, Benoit Dostie, Andrew Leach, Eduardo Fajnzylber, Ulrich Doraszelski, Samuel Clark, and especially Ig Horstmann for helpful comments and advice. y Institute of Applied Economics, HEC Montréal, 3000 Côte-Sainte-Catherine, Montréal, QC, CANADA H3T 2A7 and CIRANO and CIRPÉE; [email protected]

1

1

INTRODUCTION

Advertising has come to represent a critical element of …rm strategy. In the United States, in the year 2000, the average advertising-to-sales ratio for the two hundred largest advertising spending industries was close to 5% (AdAge.com). However, for a variety of products, in an increasing number of jurisdictions, advertising is restricted. Alcohol advertising is often heavily regulated. Cigarette advertising is also restricted, to varying degrees, in many states, including all OECD countries. Sweden, Greece, and the Canadian province of Quebec have enacted legislation prohibiting at least some forms of advertising directed at children. Restrictions on advertising directed at children are also being considered in the U.S., where the American Psychological Association recently advocated a ban on advertising directed at children under the age of eight, and in the European Union, where a ban on junk food advertising directed at children is being considered (McKay 2005). Reducing consumption of certain products would seem to be the main purpose of these advertising bans.1 However, in addition to a¤ecting the overall level of sales, restricting advertising may have the unintended consequence of hindering competition. In particular, an advertising ban may limit price competition, and may have a di¤erential impact on speci…c …rms, advantaging some relative to others. Whether or not there is a di¤erential impact on speci…c …rms as a result of the ban, and which are advantaged or disadvantaged, depends on the role that advertising plays – whether it is informative or persuasive. If advertising’s role is largely to inform new consumers about a product’s existence and characteristics (see Stigler 1961; Butters 1977; and Grossman and Shapiro 1984), then older, better known brands will bene…t from the ban. Newer products cannot use advertising to generate awareness and so it now becomes more di¢ cult to inform potential customers of their existence. Word-of-mouth becomes a potentially more important avenue of information and so established products are likely to be harmed relatively less by an advertising restriction than are new products or those of which few consumers are aware. Alternatively, persuasion, and not information transmission, may be the more important function of advertising. Stigler and Becker (1977), and Becker and Murphy (1993), suggest that advertising a¤ects the utility that consumers derive from consuming a prod1

The European Union junk food advertising ban, for instance, is being considered in order to …ght

rising levels of obesity (BBC News (2005)).

2

uct.

Advertising creates prestige or image e¤ects by acting as a complement to the

consumption of the product. This type of advertising has also been referred to as goodwill advertising (Nerlove and Arrow 1962; Boyer 1974). Advertising expenditures contribute to a stock of goodwill that depreciates over time. Without the ability to advertise, it is more di¢ cult for brands to generate goodwill. Consequently if advertising is persuasive, then products with signi…cant stocks of goodwill should bene…t from the regulation (Doraszelski and Markovich 2005). In addition to these e¤ects on market shares, the two theories of advertising yield di¤erent predictions with respect to the impact on price of an advertising restriction. If advertising is informative, it should help overcome perceived product di¤erentiation and so should lead to price competition and lower prices. On the other hand, if advertising is persuasive, then it generates the perception that there are fewer substitutes for promoted brands and thus increases perceived di¤erentiation and prices. Therefore, if advertising is informative and is banned, prices should be higher than where it is not banned, while if it is persuasive and is banned, prices should be lower. In this paper I take advantage of the ban on advertising directed at children in Quebec to analyse the nature of advertising in the market for a particular children’s product, and to determine whether the advertising restrictions di¤erentially impact certain varieties. The Quebec Consumer Protection Act outlaws commercial advertising directed at persons under the age of thirteen (O¢ ce de la protection du consommateur). This legislation went into e¤ect in April 1980. From that time forward, advertising directed at children was prohibited under Sections 248 and 249 of the Act. The Act lists three criteria that must be considered in order to determine whether an advertisement is directed at children: i) the nature and intended purpose of the goods advertised, ii) the manner of presenting the advertisement, and iii) the time and place it is shown. If products are intended exclusively for the use of children or have a marked appeal for children, they cannot be advertised at all on children’s programmes (programmes for which children make up at least 15% of the audience), and can only be advertised on other programmes if they are treated so as not to appeal to the needs of children.2 Using a unique data set from the Print Measurement Bureau of Canada (PMB), which surveys the purchasing behavior of some …fteen thousand Canadian households annually, 2

There are some minor exceptions. For example, in-store displays are exempt from the legislation.

3

as well as advertising and price data from AC Nielsen, I examine the children’s breakfast cereal market in Canada. First, I show that prices are higher in Quebec than in other Canadian regions, where advertising directed at children is permitted, a …nding that suggests that the informative role of advertising dominates any persuasive role. In addition, as just said, if advertising is informative, the brands that stand to bene…t from an advertising ban are those that are older and better known. I label these brands established and then test whether the restriction on advertising increases their market shares and decreases the market shares of non-established brands (newer and/or less well-known brands). I do so by comparing market shares of established brands and non-established brands in regions where advertising is permitted and where it is not. The predictions are con…rmed in the data: established children’s breakfast cereal brands have higher market share in Quebec than in the rest of Canada, and the opposite is true for non-established brands. In order to con…rm that these e¤ects are a consequence of the advertising restriction and not of other unobservable Quebec-speci…c factors I present evidence that suggests that there are no obvious market-speci…c factors that have been omitted and that could be driving the results presented in this paper. In particular, I make use of information on adult/family cereal brands. Since the advertising of these brands is not restricted in Quebec by the Consumer Protection Act, these products should not be a¤ected in the same way as children’s products. That is, prices of adult/family brands in Quebec should not be higher relative to the rest of the country and market shares of established and non-established adult/family brands should not be di¤erentially a¤ected. The data con…rm that the average retail prices of all breakfast cereals are no higher in Quebec than in the rest of Canada, which supports the conclusion that higher children’s cereal prices in Quebec are indeed the result of the ban. I also use data on adult/family brands to con…rm that the market share e¤ect experienced by children’s brands in Quebec relative to the rest of Canada is the result of the ban on advertising and not because of either some unobservable preference for established brands or more brand loyalty in Quebec. I show that, unlike children’s brands, established adult/family brands do no better in Quebec than in the rest of the country, and non-established adult/family brands do no worse. The results presented in this paper are consistent with those from studies of the e¤ect of advertising restrictions on competition in di¤erent industries. Eckard (1991) studies the cigarette industry through an examination of the 1971 ban on television advertising for cigarettes in the United States. By comparing measures of competition in the cigarette 4

industry in the ten years prior to the ban and in the ten years after, he …nds that shares of leading brands were declining before the ban on advertising, but stable or increasing after its imposition. He also …nds that the ban impedes the entry of new …rms into the industry.

Holak and Reddy (1986) report similar results.

They …nd that the e¤ect of

past sales on current purchases is stronger after the 1971 ban on cigarette advertising. Sass and Saurman (1995) analyse advertising restrictions in the malt beverages industry. In this industry, restrictions on advertising vary by region, and the authors …nd increased market concentration in regions where advertising is restricted. The present study examines the reasons why brands are di¤erentially a¤ected by the advertising restrictions, whereas the studies mentioned above all focus on the e¤ect of advertising restrictions on brands with large market shares without modeling what it is about having a large market share that allows these brands to grow in the absence of advertising. Holak and Reddy (1986) do point out that there is some evidence that being an early entrant could be important.

They show, for some cigarette categories, that

purchase inertia for earlier entrants is greater after the ban. Sass and Saurman (1995) discuss what it is about large-share brands that allows them to enjoy greater market share in states where advertising is restricted: large-share brands are produced by large national brewers, while small-share brands are produced by local brewers.

Since these two sets

of brewers have very di¤erent marketing capabilities, however, it is not surprising that the local brewers are at a disadvantage where advertising is restricted.

The national

brewers are barely a¤ected by state restrictions since their marketing campaigns tend to be national. So the local brewers are the only ones that are really hurt by the ban. In my study all brands are marketed and sold nationally and yet may still be di¤erentially a¤ected by the ban depending on whether or not they are established. There are also several papers that examine the impact of price-advertising restrictions in di¤erent industries on prices. Benham (1972), Cady (1976), and Kwoka (1984) all compare prices in speci…c industries across states that allow and restrict advertising. Benham (1972) studies the eyeglasses industry; Cady (1976) examines prescription drugs; and Kwoka (1984) looks at optometry services. More recently, Milyo and Waldfogel (1999) investigate the impact of relaxing liquor price advertising restrictions in a speci…c state on prices in that state. In order to compare the results of these articles with my …ndings, I discuss these papers in the following section. The results presented here are also consistent with previous work on the breakfast 5

cereal industry.

Shum (2004) looks at this industry to determine whether advertising

increases or decreases brand loyalty. Consumers that are loyal to a particular brand will perceive fewer substitutes for it. He shows that the breakfast cereal market is characterized by considerable brand loyalty and that a brand’s advertising has a bigger impact on consumers that are not loyal to it– in other words advertising helps to overcome brand loyalty. This is in line with the predictions made in the present paper.

Without the

ability to advertise, newer brands cannot generate awareness and steal customers from older and better-known brands. The remainder of the paper proceeds as follows. In the next section, I analyse the impact of the advertising restriction in Quebec on prices. In Section 3, I examine the e¤ect of the ban on market shares. Speci…cally, I describe the empirical approach adopted to test the prediction that established brands do better in Quebec than in the rest of Canada and non-established brands do worse; and I explain the results. In Section 4, I examine adult/family cereals in order to provide evidence that these e¤ects are indeed the result of the advertising restriction, and not of some other unobserved Quebec-speci…c factors. Finally, in Section 5, I conclude.

2

THE IMPACT OF QUEBEC’S ADVERTISING RESTRICTION ON PRICES

In this section, I examine the e¤ect of the restriction on advertising directed at children in Quebec on prices in the children’s breakfast cereal industry. In addition to investigating whether price competition has been a¤ected by the ban, this analysis will allow me to determine which role of advertising (the informative role or the persuasive role) is more important in this industry. This distinction is necessary for the market share analysis presented in the next section, in order to be able to specify exactly what it means for a product to be established. The price data, collected by AC Nielsen, contain information on the retail prices for 17 children’s brands, for the years 1999 and 2000, for six Canadian regions: Atlantic Canada, Quebec, Ontario, Manitoba/Saskatchewan, Alberta, and British Columbia.3 Prices are 3

These data were used for academic purposes only and the results obtained do not necessarily re‡ect

the views of AC Nielsen. The 17 brands are those listed in the Print Measurement Bureau survey described in detail in section 3.2 below. The price for Post Pebbles cereal is the average of Fruity Pebbles and Cocoa

6

calculated using samples or aggregate census data for grocery supermarket banners. For each brand I choose the box size closest to 400 grams and I calculate the average price per 100 grams. I …nd that the prices of children’s cereals are higher in Quebec than in the rest of Canada. In Table 1, I present average prices for the di¤erent children’s brands in Quebec and in the rest of the country in 1999 and in 2000. Average prices per one hundred grams in Quebec are higher for 13 of the 17 brands. This di¤erence is con…rmed using regression analysis.

Following Milyo and Waldfogel (1999), who investigate the impact

on prices of relaxing liquor price advertising restrictions, I regress the log of price for each brand in each region in each year on a Quebec indicator, and year indicators in order to allow for time e¤ects. The average price per 100 grams in Canada is .87 and the estimated percentage increase in price if sold in Quebec is 0.025 (with p-value 0.080). This result provides evidence that the restriction in Quebec on advertising in the children’s breakfast cereal market is anti-competitive. It also suggests that, in this market, the informative role of advertising dominates the persuasive role, since the most likely explanation for the higher prices in Quebec is that the advertising restriction prevented brands from announcing to consumers their existence or their characteristics and so from overcoming perceived di¤erentiation, resulting in more limited price competition in the province. The result is also consistent with earlier work examining the e¤ects of advertising legislation di¤erences across jurisdictions. Benham (1972), Cady (1976), and Kwoka (1984) all …nd that prices are higher in jurisdictions that do not permit advertising. Milyo and Waldfogel (1999) use longitudinal data to investigate the impact on prices of relaxing liquor price advertising restrictions and …nd no signi…cant overall e¤ect on prices. They point out that single jurisdiction, before and after studies of the impact of advertising restrictions may be preferred to cross-jurisdictional studies since the latter do not allow for the control of omitted market-speci…c factors. In Section 4, I present evidence to suggest that, in many important ways, the Quebec market is very similar to the market in the rest of Canada, so that there are no obvious market-speci…c factors that have been omitted and that could be driving the results presented in this paper. I also show that in the adult/family cereal market, which is una¤ected by the advertising legislation, prices Pebbles, since no price is reported for Pebbles alone.

7

are no higher in Quebec than in the rest of Canada.

3

THE IMPACT OF QUEBEC’S ADVERTISING RESTRICTION ON MARKET SHARES

In addition to a¤ecting price competition, the restriction on advertising may have a differential e¤ect on the market shares of certain brands. Speci…cally, established brands may bene…t from the advertising restriction, and non-established brands may be disadvantaged. Before testing to determine whether this is the case, it is necessary to specify precisely what it means to be established or non-established in the present context. As mentioned in the Introduction, the de…nition of established or non-established depends on whether advertising in an industry is informative or persuasive. Based on the evidence provided by the price analysis performed above, I consider the informative role of advertising to be more important than the persuasive role in the children’s breakfast cereal industry. If advertising is informative, established brands are those of which consumers are aware even in the absence of advertising – those that are older and better-known. So if advertising is informative, economic theory predicts that older and better-known brands will have larger market shares in regions where advertising is restricted and that the opposite is true for non-established brands.4 Without the ability to advertise, it is more di¢ cult for newer and less recognizable brands – non-established brands –to increase market share.

3.1

Classi…cation of children’s cereal brands as established or non-established

Now that I have speci…ed what it means to be established, the next step is to classify the di¤erent children’s cereal brands according to the de…nition. How do I determine which brands are older and better-known? I consider a brand to be older if it was introduced prior to 1980.5 Longevity by itself though is not su¢ cient for ensuring a high level of awareness. Firms must also communicate the existence of their brands to consumers in 4

See, for instance, Doraszelski and Markovich (forthcoming), who develop a model of awareness adver-

tising in which, if there are existing asymmetries amongst …rms, banning advertising increases the market shares of …rms with higher levels of awareness. 5 Dates of introduction for child and adult/family brands were collected from Companies and their Brands and through direct contact with Kellogg’s, General Mills, Post and Quaker.

8

order to become well-known. In the context of children’s breakfast cereals, an important method of communication is brand promotion. Children learn of the existence of brands through advertisements and in-store displays; they also learn about brands from their parents, and so past promotional activity is important as well. I do not have data on all the types of promotion that …rms have engaged in over the years and so I capture a brand’s typical promotional activity by its past expenditure on national advertising. I use national advertising expenditure data for children’s brands in the pre-sample years for which AC Nielsen data are available (1989-1996). I examine the advertising expenditure levels and determine that, for children’s brands, expenditures over this period are clustered in two distinct groups: (i) little or no advertising, and (ii) heavy promotional activity. Between $400 000 and $550 000 there are very few observations, while above and below these amounts there are relatively many (all expenditure levels are in year 2000 dollars). Given the obvious break between these clusters, it seems that heavy promotional activity starts at around $550 000, and therefore I classify a brand as having engaged in heavy promotion over the years if it spent at least this amount in over half of the pre-sample years.6 Based on this, I denote a brand established if, in addition to having been introduced prior to 1980, more than $550 000 was spent promoting it nationally in at least …ve of the eight pre-sample years. Table 2 lists children’s brands and indicates whether they are classi…ed as established or non-established.

3.2

Data

In this subsection, I describe in more detail the data used to perform the market share analysis. In order to test for the di¤erential e¤ect of the ban the following variables are used: market shares, prices, income levels, and advertising expenditures. Table 3 contains summary statistics. Market share data: 6

I have examined the sensitivity of my results to the selection of this cuto¤. I also consider cuto¤s of

$500 000 and $600 000. Relative to the $550 000 cuto¤, one brand is re-classi…ed as established rather than non-established if the cuto¤ is moved to $500 000, while two brands are re-classi…ed as non-established rather than established if the cuto¤ is moved to $600 000. Furthermore, I have also generated a classi…cation based on the average total expenditure level for children’s brands over the eight year pre-sample period. Brands that spent more than the average in most years are considered established while those that spent less are not. These classi…cation methods yield results that are qualitatively equivalent to those arrived at with the main classi…cation method.

9

Market shares are calculated using Category Reports from the Print Measurement Bureau. Each year the PMB surveys approximately 15 000 Canadian households. The PMB survey is representative at the provincial level and the study design includes weighting to match Statistics Canada household and individual population data. Questionnaires are left behind by interviewers and contain questions about the household’s product usage. Respondents are asked to report on the brands of cold cereal used in their household in the previous year. PMB reports on the number of households that used a brand at least once.7 I use four years of PMB reports, 1998-2001. Each year’s PMB report provides results from a survey taken in the previous year (so my data cover 1997-2000). Usage is reported for six Canadian regions: Atlantic Canada, Quebec, Ontario, Manitoba/Saskatchewan, Alberta and British Columbia. For 1997, 55 cereals are reported on, out of which 15 are children’s brands. For 1998, 17 out of 57 are children’s brands, for 1999, 17 out of 67, and for 2000, 17 out of 74. My classi…cation of breakfast cereals into children’s cereals and non-children’s cereals is motivated by the classi…cations of Nevo (2000) and Shum (2004).8 Table 2 lists average market shares for the children’s brands in the sample in Quebec and the rest of Canada (ROC). Price data: As mentioned above, the price data are collected by AC Nielsen. For the 1997-2000 period that matches the PMB sample period, prices are calculated using samples or ag7

It is important to note that respondents are asked whether they use a brand "most often" or "some-

times", but the "usage" number that is reported is for brands "used most often/sometimes". The only way that this could be a problem is if, in Quebec, respondents are actually checking o¤ "sometimes" for established brands and "most often" for non-established brands (but these all count for the same), in which case the data could underestimate the actual shares of non-established brands, or if, in the rest of Canada, respondents are actually checking o¤ "sometimes" for non-established brands and "most often" for established brands, in which case the data could underestimate the actual shares of established brands. This is unlikely to be the case. 8 Certain Quaker Bagged Cereals might also be considered children’s brands.

These brands are not

included in the sample since PMB reports usage only in its 2001 survey and since price data were not available from AC Nielsen for these brands. Kellogg’s Cruncheroos was also not included because AC Nielsen did not have price data for this brand.

However, the inclusion of Cruncheroos would almost

certainly strengthen my results–it is a non-established brand whose average market share in Quebec is lower than its average market share in the rest of Canada. The same is true of three of the …ve Quaker bagged brands.

10

gregate census data for grocery supermarket banners by taking total revenue from boxes of all sizes divided by total quantity. This measure is not ideal as it does not represent a properly weighted average. In Section 3.4, I discuss the e¤ect on the market share analysis of the inclusion of the price data used in the price analysis of Section 2 (data that permit the calculation of an average price per 100 grams, but are available only for 1999 and 2000). Income data: Income data come from Statistics Canada’s Canadian Socio-Economic Information Management System II (CANSIM II) database. For each of the six Canadian regions I determine the average family income in each year in constant 2002 dollars. Advertising data: As mentioned above, I use pre-sample (1989-1996) national advertising expenditures to determine which brands are established. These data come from AC Nielsen Media Services annual estimates of Advertising Expenditures in Canada. I also use provincial annual advertising expenditure data (again from AC Nielsen Media Services) over the four sample years in the regression analysis.9 Despite the ban on advertising directed at children in Quebec, there is some advertising for children’s cereal brands in this province, presumably directed towards adults. Therefore it is important to control for the possibility that established brands have higher market shares in Quebec simply because they are more heavily promoted. Only advertising spending that originates in Canada is included; anything spilling over from the United States is omitted.

The data are collected through the following

sources: television, radio, daily newspapers, magazines, and out-of-home. Almost all of the advertising for ready-to-eat breakfast cereals is done through television.10 9

There are some brands that have related brands (for example: in addition to regular Cap’n Crunch,

in some years Quaker produces Cap’n Crunch and Berries cereal) that are not listed in the PMB but for which there is spending on advertising. In these cases, I attribute the advertising expenditures for the related brands to the main brand. The idea is that current market share of a brand, and also its level of awareness depend on the current and past advertising for the related brands. Alternatively, one could assume that the main brand is not a¤ected by the related brand’s advertising. The results do not change if this assumption is made. 10 The following networks are included in AC Nielsen’s estimates: Télé-Métropole, Radio Canada, Télévision Quatre Saisons, Canadian Broadcasting Corporation, Canadian Television, Baton Broadcasting Service-Ontario, Global, Maritimes Independent Television, The Sports Network, Musique Plus, MuchMusic, and Newsworld.

11

3.3

Empirical speci…cation

To test the prediction that established brands bene…t when advertising is prohibited, and that the opposite is true for non-established brands, I examine the relationship between market shares of established brands and non-established brands in regions where advertising is allowed and where it is not.

Speci…cally, I construct variables which indicate

whether a brand is established or not (E and N E) and whether the observation is from Quebec (Q). I then form an interaction term between the established indicator and the Quebec indicator, (E

Q), and an interaction term between the non-established indicator

and the Quebec indicator, (N E

Q). I use these variables to identify the e¤ect on market

share of being (i) non-established and sold in Quebec, (ii) non-established and sold in the rest of Canada, (iii) established and sold in Quebec, (iv) established and sold in the rest of Canada. As was true for the price analysis, in order for the empirical strategy proposed in this section to be valid the Quebec indicator must capture only the e¤ect of the ban and must not be correlated with any other unobservable factors which are speci…c to Quebec and which might a¤ect established and non-established brands di¤erently. In Section 4 below, I provide evidence which suggests that the di¤erential e¤ect is indeed the result of the ban. In addition to whether or not it is established and whether or not it is subject to the ban, a brand’s market share is likely in‡uenced by a number of other factors. I assume that market share depends on price, current and lagged advertising expenditures, and average income of consumers in the market. In order to allow brand j’s share in a particular market to depend on characteristics and prices of other brands I follow Berry (1994) and specify a relationship in which the dependent variable is the di¤erence between the log of each brand’s observed market share and the log of the share of an outside alternative, o: ln(Sj )

ln(So ). Berry’s speci…cation is derived from a discrete choice logit model of

demand in which consumers compare the utility they would get from consuming each brand and purchase the brand that yields the highest utility. So I regress the di¤erence between the log of each brand’s observed market share in region r in year t and the log of the share of the outside alternative in this market on price, advertising, lagged advertising, income, the established indicator, the E Q indicator, and the N E

Q indicator.

12

More speci…cally, I estimate the following equation: ln(Sjrt )

ln(Sort ) =

0

+

+

1 pjrt

4 Irt

+

+

5 Ej

2 ajrt

+

+

3 ajr(t 1)

6 (Ej

Q) +

7 (N Ej

Q) + "jrt ;

(1)

where pjrt and ajrt represent brand j’s price and advertising expenditure in region r in year t respectively, and where ajr(t r in year (t

1)

denotes brand j’s advertising expenditure in region

1), Irt is the average total income of all family units in region r in year t,

and "jrt is an error term. Year dummies are also included to control for any time trends. The omitted category is non-established in the rest of Canada, and so the coe¢ cient captures the e¤ect of being established relative to being non-established. parameters of interest are

6

and

7;

5

The primary

which are intended to capture the additional e¤ects

of being established and non-established in Quebec relative to (being established and non-established in) the rest of Canada respectively. I use two de…nitions of market: (i) the children’s breakfast cereal market, and (ii) the total breakfast food market which consists of all cereal users (children’s brands and adult/family brands) and all individuals who claimed not to use any cereal.11 The advantage of market de…nition (i) is that it provides a direct way to determine whether the relative share of a children’s established brand is higher or lower in Quebec. The advantage of market de…nition (ii) is that it captures the ability of consumers to opt for a product outside of the children’s market, if, for instance, the prices of all children’s cereals increased. If the market is de…ned to be the total breakfast food market, then the reference brand (or outside brand) is comprised of all alternatives to children’s cereal consumption. So, in this case Sort represents the share in region r in year t of adult/family cereal brands along with non-purchase of cereal. If, on the other hand, the market is restricted to the children’s cereal market, then selecting a reference brand is not as obvious since this choice can in‡uence results (or at least the interpretation of results). Suppose, for instance, that one of the established brands is selected to act as the reference good. In this case it is not clear what the sign of

6

should be. Relative to the particular established brand chosen

as the reference brand, the other established brands might do better or worse in Quebec 11

PMB asks respondants whether they did or did not purchase any cereal brands in the last six months.

I assume that those who did not buy in the previous six months are those that did not buy in the entire previous year.

13

than in the rest of the country. There is no reason to expect that, on average, the share of one established brand relative to another (the reference brand) should be any di¤erent in Quebec than in the rest of the country. On the other hand, we would expect that, at least on average, the share of non-established brands relative to the reference established brand should be smaller in Quebec than in the rest of the country. (If on the other hand, a non-established brand is selected as the reference good, established brands should do better in Quebec relative to this brand than in the rest of the country, and we should not expect a di¤erential e¤ect for non-established brands in Quebec.) Moreover, we cannot just test this for one particular established brand, since if we …nd that the share of non-established brands relative to this particular established brand is lower in Quebec than in the rest of Canada, it could be either that non-established brands have lower market share in Quebec, or that the reference established brand in particular has higher market share in the province. Therefore I construct a composite established reference brand for each market (regionyear). In order to allow for it to be the case that when choosing not to purchase an inside brand consumers are more likely to opt for an established brand with larger market share, I calculate the share-weighted average share of established brands. Relative to this composite established reference good, non-established brands should do worse in Quebec than they do elsewhere. Again, there is no reason to expect that individual established brands should do any di¤erently on average in Quebec than in the rest of the country relative to the composite established reference brand. Nonetheless, since, according to this market de…nition, the market is just composed of children’s cereals, if non-established brands are found to do worse on average in Quebec than in the rest of Canada, then established brands necessarily do better.12;13 Note also that equation (1) implies that prices and advertising expenditures of the outside good are normalized to be the same (zero) across markets. This is standard in the literature when the outside good represents the no-purchase (in the market being studied) 12

Alternatively I could construct a composite non-established brand. Relative to this composite reference

brand, established brands should do better in Quebec than in the rest of the country, and we should not expect a di¤erential e¤ect for non-established brands in Quebec. 13 Rather than constructing a composite reference brand I could also run multiple regressions, each time using a di¤erent established brand as the reference good. In each case it should be that, relative to the established brand chosen, non-established brands do worse in Quebec than in the rest of the country.

14

option. In the current analysis, however, depending of the market de…nition, the reference good might be an inside good and so shares of children’s brands might not be independent of the price and advertising of the reference brand. To address this issue I also consider the following speci…cation: ln(Sjrt )

ln(Sort ) =

0

+ where port , aort , and aor(t

+

1 (pjrt

4 Irt

1)

+

5 Ej

port ) + +

2 (ajrt

6 (Ej

Q) +

aort ) + 7 (N Ej

3 (ajr(t 1)

Q) + "jrt ;

aor(t

1) )

(2)

represent the price, advertising expenditure, and lagged

advertising expenditure of the reference good in region r in year t. For market de…nition (i), port , aort , and aor(t

1)

represent respectively the share-weighted average price, advertising

expenditure, and lagged advertising expenditure of the composite established brand. It is clear in the case of market de…nition (i) that the reference good is in fact an inside good whose price and advertising will in‡uence consumer choices and therefore should be included in the model. If this is not done, and non-established brands are found to do better in Quebec relative to the reference brand than they do in the rest of the country, it could actually be the result of di¤erent price setting and advertising behavior for the reference brand in Quebec than elsewhere. This speci…cation cannot be estimated for market de…nition (ii) since price data for adult/family brands and for non-cereal purchases are not available. However, for this market de…nition it is less problematic to consider the reference good as being an outside alternative whose value is constant since there is no reason to think that pricing or advertising for adult/family cereals and non-cereal purchases should be systematically di¤erent in Quebec. 3.3.1

Endogeneity of price and advertising

At this point, it is important to deal with the possibility that "jrt is correlated with certain explanatory variables. If unobserved (by the econometrician) determinants of a brand’s market share also a¤ect its price or advertising expenditure, parameter estimates may be inconsistent.

For instance, unobservable promotional activity, such as in-store displays

and sponsorships of sporting/schooling events, could a¤ect market shares directly but also in‡uence prices and expenditures on traditional means of advertising.

Shocks to

demand such as those generated by health warnings (for example announcements about the increasing obesity of children) can also a¤ect market share as well as advertising and 15

pricing strategies. In addition, brand equity might further compromise the exogeneity of the price and advertising expenditure variables. Adding …rm dummies to the model will address some of these issues. This is the appropriate approach when there are unobservable characteristics at the …rm level that a¤ect market shares and advertising budgets simultaneously. In other words, …rms could have general marketing strategies that they apply to all of their brands. For instance, Kellogg’s may decide to pay for premium shelf space at the supermarket for its brands. This might increase market share directly and could be part of an overall promotional strategy that sees the …rm spending more on television advertising as well. However, controlling for …rm-speci…c unobservables is not enough to resolve all potential endogeneity problems. While …rm dummies should control for any unobserved heterogeneity stemming from the manufacturer that a¤ects a brand’s market share and that also in‡uences its price and the amount spent promoting it, there may also be brand-speci…c unobserved heterogeneity.

To address this I adopt a brand …xed-e¤ects model. Including …xed

e¤ects means that the indicator for being established, Ej , and …rm dummies cannot be separately identi…ed since they do not vary from one market to another. Nonetheless, I can consistently estimate the coe¢ cients of interest since they do vary across markets. Once brand e¤ects are included, the error term is the unobserved region-year deviation from the overall mean valuation of the brand. Firms are assumed to be able to observe this deviation and account for it when choosing their marketing strategies. It seems likely that when setting their prices in each market (region-year) …rms take the deviation from the mean valuation into account. To deal with this possibility, I adopt an instrumental variables approach proposed by Hausman (1997). This method relies on the assumption that region-speci…c valuations of brands are independent across regions controlling for brand-speci…c intercepts. Under this assumption, prices of a brand in other regions are valid instruments. This is the case since prices of a particular brand in two regions are correlated owing to the common marginal cost of producing the brand but the price in one region is uncorrelated with the market-speci…c valuation of the brand in the other. Therefore, I use the average of prices in other regions (so not including the region being instrumented) as an instrument.14 The assumption that valuations are independent across 14

A common set of instruments employed to deal with the endogeneity of price in discrete choice models

(Nevo 2000; Berry, Levinsohn, and Pakes 1995) includes the observed brand characteristics, the sums of the values of the same characteristics of other brands sold by that …rm, and the sums of the values of

16

regions will not hold in a number of circumstances. For instance, if there is some national shock, then all regions will be a¤ected.

The inclusion of time dummies should help to

correct this problem. Also if local advertising or in-store promotions are correlated across regions, then the independence assumption will be violated. Nevo (2000) points out that the larger the regions, the less likely it is that there will be correlation across borders. In my study regions are entire Canadian provinces (or multiple provinces in the cases of Manitoba/Saskatchewan and Atlantic Canada); it is therefore doubtful that in-store promotional strategies are coordinated across such vast distances. In an e¤ort to con…rm that the potential endogeneity of price and/or advertising does not obscure the estimated e¤ects of being established or non-established in Quebec, I also analyse a speci…cation in which I do not include price and advertising. I estimate ln(Sjrt )

ln(Sort ) =

0

+

1 Irt

+

2 Ej

+

3 (Ej

Q) +

4 (N Ej

Q) + "jrt ;

(3)

using market de…nitions (i) and (ii). Below I present results which reveal that omitting these variables does not signi…cantly a¤ect the coe¢ cients of interest.

3.4

Empirical results

Table 2 lists average shares of the children’s cereal market for brands in Quebec and in the rest of Canada respectively and shows that the majority of established brands do better in Quebec than in the rest of Canada and the majority of non-established brands do worse in Quebec than in the rest of Canada. A more complete analysis of the e¤ect of the ban is provided by the estimation of speci…cations (1)-(3) using the two market de…nitions given above. The results, presented in Table 4 of the appendix, are consistent across speci…cations and reveal that established brands do better in Quebec than they do in the rest of Canada, and that the opposite is true for non-established brands.15 Without the same characteristics of the brands o¤ered by other …rms. These variables are meant to proxy for the closeness of the competition in the market. I do not adopt this approach here since in my case there is almost no variation between markets in these instruments. This is because there is little variation over time and no variation across regions in the brands available for sale in my data and so this amounts to little more than adding brand dummies. 15 Reported p-values for all regressions are calculated using a generalized White-like formula to compute the standard errors in order to allow for serial correlation between observations for each brand in a given province.

17

the ability to advertise, less established brands are at a competitive disadvantage relative to established brands. The …rst two columns of Table 4 contain results from the brand …xed-e¤ects estimation of (1) using the instrument described above to control for the endogeneity of price. In Column 1, I report results for market de…nition (i) (children’s cereals only) with the composite established brand acting as the reference brand. As pointed out above, in this case there is no reason to expect that the coe¢ cient on E

Q should be signi…cantly

di¤erent from zero since, on average, the share of established brands relative to the composite established brand should be no di¤erent in Quebec than in the rest of the country. This is what I …nd. On the other hand, we should expect that relative to the composite established brand, non-established brands do worse in Quebec than in the rest of the country. The coe¢ cient on N E

Q is -.38 (p-value of .028), which indicates that, relative

to the reference good, market shares of non-established brands are 38% lower in Quebec than in the rest of the country.16 Since the market is de…ned to be only children’s cereal brands, the fact that non-established brands do worse in Quebec than they do in the rest of Canada implies that established brands do better in Quebec than they do in the rest of the country.17;18 16

As mentioned above, another way to analyse this would be to run seven di¤erent regressions each time

using a di¤erent established brand as the reference good. I do not report them here, but results from these regressions strongly support the results derived using the composite established brand. In each case non-established brands do worse in Quebec than in the rest of Canada, although when Lucky Charms is the reference brand the marginal signi…cance level is quite high. 17 It is important to note that these regressions were run using all four years of data, and, consequently, the less accurate price data. Given the manner in which prices were calculated for the full four-year sample, I test the robustness of my results to the inclusion of more accurate price data. I run speci…cation (1) using just the …nal two years of data since for these two years price data that represent a properly constructed weighted average are available. The results are consistent with those derived using the full four years of data. 18 Speci…cation (1) restricts the e¤ects of interest to be the same in each of the regions that make up the ROC. A Chow test of this restriction suggests that the average e¤ects estimated using this speci…cation cannot be attributed exactly to each region. Inspection of the results, however, reveals that there are no regions in which the e¤ects of interest move in the opposite direction from speci…cation (1). This con…rms that the restriction imposed in (1) is rejected because the regional e¤ects di¤er in magnitude and not because they di¤er in sign. That is, the e¤ects measured in (1) represent averages of regional e¤ects that all move at least weakly (since a number of the e¤ects are statistically insigni…cant) in the same direction, but the con…dence intervals of the average e¤ects do not encompass each of the individual regional e¤ects.

18

In Column 2, I report results from the same speci…cation using market de…nition (ii) (total breakfast food market). In this case the outside alternative is adult/family cereals and non-purchases, and so we can interpret the coe¢ cient on E

Q as the additional

e¤ect of being established in Quebec relative to the rest of the country. The coe¢ cient on E

Q is .20 (p-value of .126), suggesting a 20% increase in market share relative to the

outside good for established brands in Quebec. The coe¢ cient on N E

Q is -.30 (p-value

of .093), which indicates a 30% decrease in market share relative to the outside good for non-established brands in Quebec. Comparing results from the two market de…nitions it is important to note …rst that the coe¢ cient on N E

Q using market de…nition (i) captures both the decrease in share for

non-established brands, but also some of the increase in share for established brands since the reference good is a composite of established brands. In contrast, using market de…nition (ii) the reference good, since it is made up of adult/family cereals and non-purchases, should not be a¤ected by the ban, and so the coe¢ cient on N E

Q represents only the

decrease in share for non-established children’s brands. Note also that with de…nition (ii) consumers have the option of substituting to something other than a children’s cereal. In some sense de…nition (i) restricts consumers that have not heard of certain non-established brands to buy another children’s cereal rather than opting instead for and adult/family brand or some outside option. In Column 3 of Table 4, I present results from the estimation of speci…cation (2). With this speci…cation I consider relative prices, advertising expenditures, and lagged advertising expenditures in order to address the fact that under market de…nition (i), the outside good is actually an inside good and so its price may a¤ect brand selection on the part of consumers. Results are consistent with those derived using speci…cation (1). In Quebec, non-established brands do 41% worse than they do in Ontario relative to the composite established brand (p-value of .03). As mentioned above, I do not estimate Speci…cation (2) for market de…nition (ii) since I do not have price data for adult/family cereals or the non-purchase alternative. Finally, in Columns 4 and 5, I present results from the estimation of speci…cation (3). In contrast with speci…cations (1) and (2), prices and advertising are not included in this speci…cation, but it provides similar evidence against the null. The coe¢ cients of interest are unchanged (statistically) from those derived using the main speci…cation presented in (1). Column 4 lists results for market de…nition (i): the coe¢ cient on N E 19

Q is -.36

(p-value of .043). In Column 5, I report results using market de…nition (ii). I …nd that being established in Quebec increases market share relative to being established in the rest of Canada by about 14% (p-value of .15), while being non-established in Quebec reduces market share relative to being non-established in the rest of the country by 29% (p-value of .11).

4

ARE THE PRICE AND MARKET SHARE EFFECTS REALLY CAUSED BY THE BAN? EVIDENCE FROM THE ADULT/FAMILY CEREAL MARKET

Now that I have shown that, as compared to the rest of Canada, in Quebec prices are higher and market shares for established and non-established children’s brands are di¤erent, it is important to provide evidence that these e¤ects are the result of the ban on advertising and not of some other unobservable Quebec-speci…c factor. For instance, one possibility is that Quebecers watch less television anyway and so the ban actually has little impact on behaviour. However, in its Window on Quebec 2001, AC Nielsen reports that people in Quebec watch approximately the same amount of television as people in Canada as a whole (22.5 hours per week in Quebec vs. 21.5 hours per week in Canada). Another possibility is that consumption habits in Quebec are very di¤erent from those in the rest of Canada.

This does not appear to be the case. AC Nielsen …nds that Quebecers

spend approximately the same amount on food ($114.90 per week in Quebec vs. $112.09 per week in Canada) and on baked goods and cereals in particular ($13.42 per week in Quebec vs. $12.47 per week in Canada). Moreover Quebec consumers do not have a bias against pre-sweetened cereals. The PMB data show that in 1999, 25% of cereal usage in Quebec was of pre-sweetened cereals, versus 26% nationally. The PMB also reports that the fraction of households in Quebec with children under the age of twelve is not much di¤erent than in Canada as a whole (in 2000, 25.5% in Quebec vs. 27% nationally). There is a signi…cant di¤erence between Quebec and the rest of Canada in language. Only 43% of the population of Quebec can speak English, de…ned by Statistics Canada as being able to carry on a conversation (1996 Census), but a much larger fraction understands enough English to make sense of any breakfast cereal commercials they might see on the English stations available to them. Children in Quebec begin learning English in Grade 3 (Grade

20

4 prior to 1999). Despite the similarities between the di¤erent markets, there may still be omitted market-speci…c factors that a¤ect established and non-established brands di¤erently. For instance some national advertising campaigns might be primarily designed for English television programmes and so these advertisements would have little e¤ect on Francophones.

Francophones would therefore be more likely to purchase brands that had been

around for a long time because those are the ones they have heard of. This would be particularly problematic if established meant only old. However, I have de…ned established brands to be those that have been around for a long time and that regularly advertise nationally.

The fact that established brands are found to have bigger market share in

Quebec therefore implies that regular national advertising likely has an e¤ect on French speaking people. As mentioned in Section 3.2 above, it is also the case that, despite the ban on advertising in Quebec, there is some advertising expenditure on children’s cereal brands in this province.19 Therefore one might also be concerned that established brands have higher market shares in Quebec simply because they are more heavily promoted. One possibility is that laws to support the French language in Quebec have increased promotional costs in such a way as to encourage …rms to promote their established brands more heavily. Since I use region-speci…c advertising expenditure data, I am controlling for this possibility. Finally, and perhaps most importantly, Quebecers may be more brand loyal or just have some speci…c preference for established brands. However, if Quebecers have some speci…c preference for established brands, it should also be the case that market shares for established adult/family cereal brands are di¤erentially a¤ected in Quebec despite the fact that the advertising of these brands is not restricted in Quebec by the Consumer Protection Act. This does not appear to be the case. I compare the children’s market with the adult/family cereal market. If adult/family established brands are found to do no better in Quebec than in the rest of the country, and non-established brands to do no worse, then this would provide support for the conclusion 19

Advertisements must be presented by agencies and/or advertisers to a review panel prior to being

cleared to air. Additionally, some advertisements for children’s products may appear on programs that are directed at an adult audience. Nielsen Media Research examined advertisements for General Mills’ Nesquick and Lucky Charms for me and found that they appeared on such adult-directed shows as Chicago Hope.

21

that the e¤ect experienced by children’s brands in Quebec is indeed the result of the ban on advertising and not because of either some unobservable preference for established brands or more brand loyalty on the part of Quebecers. Brand level price data for adult/family cereal brands are not available for the sample period, and so I employ speci…cation (3) in order to determine whether an additional e¤ect exists for established and non-established adult/family brands. Since the results for children’s brands show that the coe¢ cients on the e¤ects of interest are unchanged using speci…cation (3) rather than (1), little is lost from the use of this speci…cation. Adopting the same method as for children’s cereals I classify adult/family brands as established or not. A brand is established if it was introduced prior to 1980 and if it engaged in heavy promotional activity. I examine advertising expenditure levels for adult/family cereals over the 1989-1996 pre-sample period. Again, expenditures appear to be clustered in two groups, although for adult/family brands the cuto¤ is not quite as obvious as it was for children’s brands. The break is somewhere between $1 million and $1.5 million.

Since it is less clear where the break occurs, I also calculate the average

expenditure level for adult cereals and …nd it to be close to $1 million. Therefore, I consider two possible cuto¤s for heavy promotional activity: (i) spent at least $1.5 million on advertising in over half of the pre-sample years, and (ii) spent at least $1 million on advertising in over half of the pre-sample years. Table 5 contains results from the estimation of speci…cation (3) using only observations on adult/family brands.20 The results reveal that, in contrast with children’s brands, adult/family established brands do no better in Quebec than in the rest of Canada and non-established adult/family brands do no worse in Quebec than they do in the rest of the country. These results provide evidence that the increase in market shares of established children’s brands is the result of the advertising legislation. In a comparable market, untouched by the legislation, established brands are not better o¤ and non-established brands are not worse o¤ in Quebec than in the rest of Canada. With regards to the price analysis, one might still be worried, despite the similarities between Quebec and the rest of Canada alluded to at the beginning of this section, that there are some omitted market-speci…c factors that cause prices in general to be higher in Quebec. Again, I turn to the adult/family cereal market. Since the advertising of 20

Again, I employ two de…nitions of market. In the …rst, I consider the market to be comprised only of

adult cereals. In the second I also include children’s cereals and non-purchases.

22

adult/family cereal brands is not restricted in Quebec, prices of adult/family brands should not be higher in the province. Although brand level price data are not available, AC Nielsen reports an average retail price per box for all breakfast cereals nationally and in each region in each year. Table 6, lists these prices for Canada and Quebec for the years 1995-2001. Prices for all breakfast cereals are no higher in Quebec than they are nationally. Since the adult/family brands are untouched by the legislation, this provides support for the hypothesis that the higher children’s brands prices are indeed the result of the advertising restriction.

5

CONCLUSIONS

In this paper I have examined the possibility that an unintended consequence of the ban on advertising directed at children in Quebec is to hinder competition. While the average price of all breakfast cereals is no higher in Quebec than in the rest of the country, prices of children’s brands are higher in Quebec. This result suggests that, in this industry, the informative role of advertising dominates the persuasive role, since the most likely explanation for the higher prices is that, as a result of the advertising restriction, perceived product di¤erentiation is increased and price competition is therefore hindered. Since the informative role of advertising is dominant in this industry, older and betterknown brands stand to bene…t when advertising is restricted. These brands should have bigger market shares in Quebec than in regions where advertising is allowed, and the opposite should be true for non-established brands. The empirical analysis of the breakfast cereal market performed in this paper supports these predictions. Established children’s brands–those that have been around for a long time and have advertised heavily and regularly over the years–enjoy bigger market share in the province of Quebec where advertising is banned than they do in the rest of Canada. Regardless of the empirical speci…cation or de…nition of market used, the advertising restriction appears to drive up market shares for these brands. On the other hand, non-established children’s brands do better in the rest of Canada than they do in Quebec. Newer products, it would seem, are less successful at penetrating the children’s breakfast cereal market in Quebec since they are unable to announce their existence to children. In contrast, the market shares of established and non-established adult/family brands do not appear to be signi…cantly di¤erent in Quebec than in the rest of the country. 23

This provides con…rmation that the e¤ect on children’s brands stems from the ban on advertising in the province and not because of either some unobservable preference for established brands or more brand loyalty on the part of Quebecers.

24

6

REFERENCES

AC Nielsen Company of Canada.1995-2000. Market Track-Grocery Banner. — — –.2001. Window on Québec 2001. AdAge website: www.adage.com. BBC News. 2005. EU takes Aim at Junk Food Adverts. http://news.bbc.co.uk/2/hi/business/4190313.stm (last updated January 20, 2005). Becker, G., and Murphy, K. 1993. A Simple Theory of Advertising as a Good or Bad. The Quarterly Journal of Economics 108: 941-964. Benham, L. 1972. The E¤ect of Advertising on the Price of Eyeglasses. Journal of Law and Economics 25: 242-262. Berry, S. 1994. Estimating Discrete-Choice Models of Product Di¤erentiation. The RAND Journal of Economics 63: 841-890. Berry, S., and Levinsohn, J., and Pakes, A. 1995. Automobile Prices in Market Equilibrium. Econometrica 63: 841-890. Boyer, K. 1974. Informative and Goodwill Advertising. The Review of Economics and Statistics 56: 541-548. Butters, G. 1977. Equilibrium Distributions of Sales and Advertising Prices. The Review of Economic Studies 44: 465-491. Cady, J. 1976. An Estimate of the Price E¤ects of Restrictions on Drug Price Advertising. Economic Inquiry 14: 493-510. Doraszelski, U., and Markovich, S. Forthcoming. Advertising Dynamics and Competitive Advantage. Rand Journal of Economics. Eckard, W. Jr. 1991 Competition and the Cigarette TV Advertising Ban. Economic Inquiry 29: 119-133. Gale Research Group. 1990-2000. Companies and their Brands. 25

Grossman, G., and Shapiro, C. 1984. Informative Advertising with Di¤erentiated Products. The Review of Economic Studies 51: 63-81. Hausman, J. 1997. Valuation of New Goods Under Perfect and Imperfect Competition. Pp. 209-237 in The Economics of New Goods, volume 58 of the NBER Studies in Income and Wealth, edited by T.F. Bresnahan and R.J. Gordon. Chicago, Ill.: University of Chicago Press. Holak, S., and Reddy, S. 1986 E¤ects of a Television and Radio Advertising Ban: A Study of the Cigarette Industry. Journal of Marketing 50: 219-227. Kwoka, J. 1984. Advertising and the Price and Quality of Optometric Services. The American Economic Review 74: 211-216. Mckay, B. 2005. Study Tries to Link Obesity in Children with Food Marketing. The Wall Street Journal Online, January 27. Milyo, J., and Waldfogel, J. 1999. The E¤ect of Price Advertising on Prices: Evidence in the Wake of 44 Liquormart. The American Economic Review 89: 1081-1096. Nerlove, M., and Arrow, K.J. 1962. Optimal Advertising Policy under Dynamic Conditions. Economica 29: 129-142. Nevo, A. 2000. Mergers with Di¤erentiated Products: The Case of the Ready-to-Eat Cereal Industry. RAND Journal of Economics 31: 395-421. Nielsen Media Services. 1989-2001. Annual Summary of Advertising Expenditures in Canada. Nielsen Marketing Research, Markham, ON. O¢ ce de la protection du consommateur. 2006. Consumer Protection Act. http://www.opc.gouv.qc.ca/e_presentation/laws.asp (last updated October 2, 2006). Petrin, A. 2002. Quantifying the Bene…ts of New Products: The Case of the Minivan. The Journal of Political Economy 110: 705-729. Print Measurement Bureau. 1997-2001. PMB Category Reports.

26

Sass, T., and Saurman, D. 1995. Advertising Restrictions and Concentration: The Case of Malt Beverages. The Review of Economics and Statistics 77: 66-81. Shum, M. 2004. Does Advertising Overcome Brand Loyalty? Evidence from the Breakfast Cereals Market. Journal of Economics and Management Strategy 13: 241-272. Stigler, G. 1961. The Economics of Information. The Journal of Political Economy 69: 213-225. Stigler, G., and Becker, G. 1977. De Gustibus Non Est Disputandum. The American Economic Review 67: 76-90.

27

7

APPENDIX A:

Table 1: Average prices in 1999 and 2000 per 100 grams in Quebec and the rest of Canada.

Brands

Average price per 100g

Average price per 100g

Quebec

Rest of Canada

GM Cinnamon Toast Crunch

$0.93

$0.88

GM Count Chocula

$1.08

$1.03

GM French Toast Crunch

$0.83

$0.82

GM Golden Grahams

$0.93

$0.89

GM Lucky Charms

$0.91

$0.88

GM Nesquick

$0.89

$0.88

GM Reese’s Pu¤s

$0.94

$0.95

GM Trix

$1.08

$0.93

Kellogg’s Cinnamon Mini Buns

$0.96

$0.99

Kellogg’s Corn Pops

$0.92

$0.88

Kellogg’s Froot Loops

$0.89

$0.88

Kellogg’s Frosted Flakes

$0.81

$0.75

Post Alphabits

$0.68

$0.65

Post Honeycombs

$0.751

$0.746

Post Pebbles

$0.83

$0.95

Post Sugar Crisp

$0.77

$0.73

Quaker Cap’n Crunch

$0.87

$0.92

28

Table 2: Market shares in Quebec and the rest of Canada (ROC) Market De…nition (i) Brand

In PMB 98

Established

Average

Average

share in

share in

Quebec

ROC

GM Cinn. Toast Crunch

y

n

0.0113

0.0514

GM Count Chocula

y

n

0.0031

0.0130

GM French Toast Crunch

n

n

0.0498

0.0348

GM Golden Grahams

y

n

0.0417

0.0494

GM Lucky Charms

y

y

0.0463

0.0613

GM Nesquick

n

n

0.0308

0.0186

GM Reese’s Pu¤s

y

n

0.0125

0.0206

GM Trix

y

n

0.0139

0.0199

Kellogg’s Cinn. Mini Buns

y

n

0.0223

0.0190

Kellogg’s Corn Pops

y

y

0.1127

0.1065

Kellogg’s Froot Loops

y

y

0.1669

0.1597

Kellogg’s Frosted Flakes

y

y

0.1414

0.1507

Post Alphabits

y

y

0.0869

0.0945

Post Honeycombs

y

y

0.1394

0.0942

Post Pebbles

y

n

0.0091

0.0129

Post Sugar Crisp

y

y

0.0683

0.0419

Quaker Cap’n Crunch

y

n

0.0640

0.0650

29

Table 3: Summary statistics for children’s breakfast cereals 1997-2000. Min

Max

Mean

St.Dev

Share of market (%) - Market De…nition (i)

.00123

.194

.061

.048

Share of market (%) - Market De…nition (ii)

.00030

.054

.015

.012

Provincial Advertising ($ millions per year)

0

1.40

.085

.15

Income ($ ‘000’s)

44.40

67.20

53.53

6.31

Price ($ per box)

2.06

5.59

3.87

.53

Price ($ per 100g in 1999, 2000)

.46

1.24

.87

.12

30

31

1)

ajr(t

aor(t

Q

Q

1)

(:428) :082 (:806)

(:392) :092 (:736)

:30 (:093)

:38 (:028)

included

(:126)

(:475)

included

:20

:082

(:861)

:15

:13

(:458)

(:054)

(:121)

:0017

:95

:61

:0058

Coe¢ cient

Coe¢ cient

Market Def. (ii)

included

(:031)

:41

(:736)

:035

(:718)

:0025

(:828)

:04

(:114)

:28

(:738)

:24

Coe¢ cient

Market Def. (i)

Speci…cation (2)

correlation are reported in parentheses.

Brand …xed e¤ects included included included Note - The number of observations is 396. The dependent variable is ln(Sjrt )

Year …xed e¤ects

NE

E

Established

Income

aort

port

ajrt

pjrt

Lagged Advertising

Advertising

Price

Variable

Market Def. (i)

Speci…cation (1)

included

(:11)

:29

(:15)

:14

(:496)

:0034

Coe¢ cient

Market Def. (ii)

included included ln(Sort ). P-values that are robust to serial

included

(:043)

:36

(:513)

:059

(:415)

:0038

Coe¢ cient

Market Def. (i)

Speci…cation (3)

Table 4: Results from the estimation of Speci…cations (1)-(3) for Market De…nitions (i) and (ii)

Table 5: Results from the estimation of Speci…cation (3) : Adult/Family brands

Market De…nition (i)

Variable Income

Market De…nition (ii)

Classi…cation 1

Classi…cation 2

Classi…cation 1

Classi…cation 2

Coe¢ cient

Coe¢ cient

Coe¢ cient

Coe¢ cient

:012

:012

:0086

:0086

(:000)

(:000)

(:000)

(:000)

Established E

NE

:10

Q

Q

Year …xed e¤ects

:030

:080

:041

(:237)

(:812)

(:361)

(:747)

:071

:095

:044

:080

(:424)

(:268)

(:627)

(:360)

included

included

included

included

Brand …xed e¤ects included included included included Note - The number of observations is 912. The dependent variable is ln(Sjrt ) ln(Sort ): P-values that are robust to serial correlation are reported in parentheses.

32

Table 6: Average retail prices of all breakfast cereals in dollars per box at regional grocery banners, 1995-2001.

2001

2000

1999

1998

1997

1996

1995

Canada

3.77

3.44

3.54

3.59

3.53

3.56

3.59

Quebec

3.50

3.26

3.35

3.65

3.57

3.54

3.49

33