Project Report - AgEcon Search

Loading...

RURAL ECONOMY Analysis of Value-Added Meat Product Choice Behaviour by Canadian Households Xu Zhang and Ellen Goddard Project Report # 10-04

Project Report

Department of Rural Economy Faculty of Agricultural, Life & Environmental Sciences University of Alberta Edmonton, Canada

1

Analysis of value-added meat product choice behaviour by Canadian households

Xu Zhang and Ellen Goddard

Project Report # 10-04

The authors are, respectively, Economist, Alberta Energy and Professor, Department of Rural Economy

2

Acknowledgements

Acknowledgements for funding are to the Alberta Prion Research Institute, the Alberta Livestock Industry Development Fund (ALIDF), Alberta Innovation and Science (INNSCI) and the Department of Rural Economy, University of Alberta.

3

Abstract ............................................................................................................................... 5 Background ......................................................................................................................... 7 The Canadian meat industry—an overview .................................................................... 7 Food retailing—store and brand choice .......................................................................... 8 Consumer demand and value-added meat ...................................................................... 9 Factors affecting meat demand ..................................................................................... 11 Economic problem ........................................................................................................ 14 Objectives ..................................................................................................................... 16 Implications................................................................................................................... 18 Literature Review And Methods ....................................................................................... 19 Introduction ................................................................................................................... 19 Definition of value-added meat products ...................................................................... 20 Overview of value added agricultural products demand .............................................. 24 Summary of Canadian meat demand studies ................................................................ 25 Hierarchy of consumer purchase decision making in the study ................................... 31 Model structure and econometric method..................................................................... 35 Demographic Data and Descriptive Statistics................................................................... 39 Introduction ................................................................................................................... 39 Socioeconomics and demographic information and definitions ................................... 41 Canadian Meat Demand Analysis By Level of Processing .............................................. 49 Introduction ................................................................................................................... 49 Data setup and descriptive statistics ............................................................................. 49 Model specification and econometric method .............................................................. 60 Model testing and empirical results .............................................................................. 63 Canadian Store Choice Analysis ....................................................................................... 73 Introduction ................................................................................................................... 73 Model specification and econometric method .............................................................. 77 Model testing and empirical results .............................................................................. 79 National and Store Brand Choice Analysis ...................................................................... 85 Introduction ................................................................................................................... 85 Data setup and descriptive statistics ............................................................................. 85 Model specification and econometric method .............................................................. 86 Model testing and empirical results .............................................................................. 89 Summary and Conclusions ............................................................................................... 99 Summary ..................................................................................................................... 100 Consumer Meat Behaviour and Level of Processing .................................................. 101 Consumer Meat Behaviour and Store Selection ......................................................... 104 Consumer Behaviour and Choice of National Brand versus Private Label Meat Products....................................................................................................................... 105 Conclusion .................................................................................................................. 107 References ....................................................................................................................... 109

4

Analysis of value-added meat product choice behavior by Canadian households

Abstract The competitive landscape in retailing has changed over the past decade. Moreover, the degree of product differentiation has been increasing: households are able to choose between an increasing number of store brands and national brands of similar products. The value added meat market is no different than any other sector of the grocery market – both national brands and private label brands are being developed to appeal to the consumer‘s desire for convenience, health, production and environmental attributes. Understanding the factors that are influencing consumers‘ value added meat product preferences is important for meat manufacturers who wish to add value to their firm‘s performance and increase market share. This knowledge is required in order to predict changes in demand and develop new products and marketing strategies that respond to changing consumer needs. The objective of the paper is to provide information on value added meat consumption patterns in Canada at the household level using household purchase information from a representative sample of the Canadian population collected through Nielsen Homescan™. Specifically the focus is on how meat consumers make their decision to purchase valueadded meat products – the impact of value added meat types, store choices and brands preference on meat demand. The study undertakes an empirical investigation of Canadian household value added meat demand for the period 2002 to 2007. A comparison of consumers‘ preferences is performed with respect to store-switching, brand loyalty and meat expenditure. Multivariate regression analysis is employed to explain consumer preferences for the examined stores, products and brands. We find that meat price, advertising, the number of stores visited, household sociodemographic characteristics and regional segments are strongly related to meat expenditure levels. Value added meat product preferences vary widely across meat types - for example, consumer behaviour towards pork is not a good predictor of behaviour towards poultry, in terms of national brand/store brand choice. The data developed in this analysis can highlight

5

marketing opportunities that exist for meat producers and processors to increase the value of total sales for their particular products. The results of this study highlight the impact of number of stores regularly shopped at on purchases of national brand versus private label meat products, the impact of expenditure on meat by product form on national brand versus private label and the impact of demographic and regional variables on all meat purchases, by animal species.

JEL Codes: D1, M3 Keywords: consumer behaviour, store loyalty, meat demand, value-added meat, national/store brand choice

6

Background

The Canadian meat sector is important to the Canadian economy. The meat processing industry is the largest food manufacturing industry. Changes in meat demand can have an impact on all segments of the food chain, which include agricultural input suppliers, farmers, processors, and distributors (Agriculture and Agri-Food Canada, 2009). Meat is an important component in the Canadian diet and it has been found to be the primary source of fat for both children and adults (Statistics Canada, 2007). Thus, understanding the factors that are influencing meat and value added meat demand in Canada is important for the Canadian agricultural sector. Moreover understanding consumer preferences for meat is increasingly important in the context of health concerns, animal disease and food safety outbreaks.

The Canadian meat industry—an overview

The meat and poultry industry is positioned as one of Canada's most important manufacturing industries (Agriculture and Agri-Food Canada, 2009). In 2008, Canada's annual shipments from the meat industry were $16.2 billion, which ranked it as the largest sector of the Canadian food manufacturing industry. Various processed meat products, including fresh/frozen, semi-processed, and processed meats (like smoked and cooked meats), as well as deli and sausage meats are well established in the marketplace and produced by Canada's meat processing companies. An increasing number of meat producers are expanding into niche markets, for example organic, and into value-added meat products (Agriculture and Agri-Food Canada, 2009).

In 2008, Canada's inventory of cattle and calves were 13.18 million on approximately 86,520 farms and ranches. With approximately 41 percent of this inventory, Alberta was the largest

7

cattle province. Farm cash receipts from the sale of cattle and calves in 2008 were $6.6 billion which represented 14 percent of total farm receipts. In the same year, in Canada, there were 12.4 million hogs on approximately 8,510 farms and 808,200 sheep and lambs on approximately 12,000 farms. Nearly three quarters of Canadian sheep production was located in Alberta, Ontario and Quebec. Farm cash receipts for sheep and lamb in 2008 were $124 million. 1.2 million tonnes of poultry meat were produced in 2008. The value of all poultry products was $3.2 billion in 2008. (Agriculture and Agri-Food Canada, 2009)\

Furthermore, the meat industry has undergone significant structural change in recent decades. The use of cost advantage strategies (low cost) and the use of more intensive product differentiation are only two examples of the strategies being pursued in the current meat industry. Intensification, concentration, and specialization are three structural forces behind meat industrialization (Bowler, 1985). As an example, Stull and Broadway (2004) suggest that industrialization in the meat industry has been focused on large volume production of uniform products at the lowest possible price, resulting in high-efficiency, high-volume cattle slaughter-dressing facilities. (Stull and Broadway, 2004).

Food retailing—store and brand choice

Retailers are the closest and most frequent point of contact for consumers to the meat industry and they can directly influence household meat consumption. In 2008, Canadian consumers spent around $69 billion on food in retail stores (Statistics Canada, 2009). The competitive landscape in retailing has changed over the past 40 years in Canada. The number of grocery stores has been declining whereas the size of the existing stores has been increasing, partially due to new entry by so-called supercenters e.g. Wal-Mart, Superstore and Costco (Agriculture and Agri-Food Canada, 2008). In 2005, approximately three quarters of the $71 billion in food and non-alcoholic sales were distributed through large chains (e.g., Loblaws, Sobeys, Safeway) and traditional grocery stores. Other format distributers, such as discount clubs (e.g., Costco and Sam‘s Club), large mass merchandising chains (e.g. Wal-

8

Mart), and convenience stores (e.g. Mac's, 711) have established a significant presence (27 per cent) of food sales in Canada (Agriculture and Agri-Food Canada, 2008).

Meanwhile, the degree of value added and product differentiation has been increasing: households are able to choose between private label and national brands of similar products (Sethuraman, 2003; Bonfrer and Chintagunta, 2004; Debbie, 2004; Hansen et al., 2006; Hassan and Monier-Dilhan, 2006; Tyagi, 2006; Kusum et al., 2008). The private label business consists of two categories: ―premium‖ private label such as President‘s Choice (Loblaws) or Our Compliments (Sobeys), and ―generic‖ such as no name and unbranded products. Private label brands have become one of the primary tools for grocery retailers to differentiate themselves from competition in retailing. The trend towards private label brand development is accelerating in all consumer product segments due to the profit potential.

Consumer demand and value-added meat

Changing consumer demand is one of the most important drivers behind the challenges and opportunities that are facing the agriculture and agri-food sector in Canada (Agriculture and Agri-Food Canada, 2009). Several studies have documented changes in meat consumption in the U.S over the past 30 years (Chavas, 1983; Moschini and Meilke, 1989; Thurman, 1987). Similar patterns can be observed in Canada. From 1970 to 2001, Canadian meat preferences s shifted from pork and beef to poultry meats (Chen and Veeman, 1991; Reynolds and Goddard, 1991).

The per capita growth in chicken consumption has been higher than for pork and beef products since early 1970. Pork and beef consumption peaked in 1976 when they accounted for 56 per cent of all Canadian meat consumption, while the share of chicken meats was 13.0 per cent. The consumption share of beef and pork meats fell to 40.6 per cent while the consumption share of chicken rose to 30.6 per cent by 2005. From 1975 to 2005, beef

9

appeared to have lost the biggest share of Canadian meat consumption falling from 36.0 per cent to 23.2 per cent while chicken's share more than doubled from 12.9 per cent to 30.6 per cent (Agriculture and Agri-Food Canada, 2009). Consumption of chicken increased by 136 per cent from 12.9 kg in 1975 to 30.6 kg in 2005 (Statistics Canada, 2008). Possibly due to Canadian consumers‘ health perceptions, chicken meat consumption has grown.

Table 1.1 Meat consumption trends in Canada from 1965 to 2005.

Year

Chicken Pork Beef

Per capita consumption (kg) 1965

10.0

18.6

28.8

1975

12.9

19.9

36.0

1985

19.3

22.0

28.0

1995

24.8

21.1

23.1

2005

30.6

17.4

23.2

2.8

-0.2

-0.5

Annual growth rates, per cent 1965-2005

Source: Statistics Canada, CANSIM table 002-0011, Accessed on March, 10th, 2009

Additionally another factor potentially affecting the demand for meat products is the changes in Canadian consumer dietary patterns over the past forty years. Many consumers want ready-made convenience food products, therefore there is an increasing demand for valueadded meat products (Agriculture and Agri-Food Canada, 2008). In the 2006 Canadian Consumer Perceptions of Food Safety and Quality survey(Agriculture and Agri-Food Canada, 2007), consumer perceptions of "nutritional value", "ingredients in the food", "brand or company name" and "convenience" are found to be closely linked to food at home consumption. Thus, more new meat products to market are concentrating on convenience,

10

variety, health and safety (See Table 1.2 below). The analysis is for U.S. market data but similar trends can be observed in Canada. Furthermore, consumers are becoming more aware of the production processes that go into their food. They are influenced by the origins of their food, how it is grown, processed and prepared.

Table 1.2: Attributes of 33 new meat products to market

Attributes

Numbers

Percentage

Convenience

30 of 33

91%

Natural

16 of 33

48%

Health benefits

17 of 33

52%

Easy cooking directions

20 of 33

61%

Better/unique tasting

21 of 33

64%

Others

5 of 33

15%

(*Source: Magazine of Meatingplace and Poultry, issues from 2006.1 to 2008.5, accessed in Sep. 2008)

Factors affecting meat demand

Aggregate consumers‘ food demand is potentially influenced by factors such as population growth, demographic profiles, changing household structure, changing consumer attitudes, advertising, food safety and growth of the economy. Population demographics, perceptions, awareness and attitudes are the key factors that influence meat demand (Verbeke et al., 2000; Reynolds-Zayak, 2004). Monitoring these factors over time can provide a comprehensive understanding of current consumer trends.

11

1. Household income and food expenditures

The household‗s income, to a large extent, influences what foods and what amount of foods are bought ( Stewart and Blisard, 2008). Households will spend more of their food dollar on meat consumption as income increases if meat is a normal good. Historical data suggests, as household income increases, the nominal level of spending on food increases. From 1961 to 2005, as per capita income increased, meat consumption increased at an annual rate of 1 per cent (Statistics Canada, 2008).

2. Household size

An important trend impacting meat demand is the growth of smaller households. Since 1966, the average number of Canadians per household has been continually decreasing (Statistics Canada, 2001). An increasing number of Canadians choose to live alone and married couples often live without children, thus the demand for smaller servings of foods and foods that require minimal preparation is increasing.

3. Population, Immigration, Education

Growth in food consumption is closely linked to population growth (Boserup, 1989). Canada's population is becoming more ethnically diverse and older. Canadian food patterns are influenced as much by the food preferences brought by immigrants from their home countries, as by exposure of the general population to different foods and methods of preparation. Education also plays an important role in the food demand of household.

12

4. Health and Nutrition

Health-related attitudes influence food choice and consumption (Steptoe et al., 1995; Geeroms et al., 2008; Hailu et al., 2009). Consumers are concerned that the food they eat may be harmful to their health (Holm and Kildevang, 1996). Research has shown that meat consumption has some relationship with colorectal cancer risk ( Norat and Riboli, 2001) and breast and prostate cancers (Biesalski, 2002). At the same time red meats are a good source of iron, something lacking in many Canadian‘s diets. Thus, a significant proportion of consumers are aware of both the health benefits and risks in their diet patterns. The 2006 Consumer Perceptions of Food Safety and Quality survey (Agriculture and Agri-Food Canada, 2007) also showed that 31 percent of consumers ranked nutrition as a top of mind issue for food at home consumption as compared to 24 percent in 2004.

5. Food safety

Food safety has become one of consumers‘ top concerns. There have been disease outbreaks and food recall issues, such as BSE, Avian Flu, foot-and-mouth, E. coli 0157, etc. in the beef cattle and poultry industries (Canadian Food Inspection Agency, 2007, refer to table 3);. Food safety concerns have dramatically increased in the past decade following incident of contaminated meat products in the U.S. and Canada (Doyle and Erickson, 2006). Food contamination is the subject of public attention and may adversely affect consumer demand for the implicated food products. Food borne diseases are very costly to society in terms of losses in public health (de Jonge et al., 2008). There is a growing interest in determining the effects of food safety concerns on meat demand. Therefore, understanding the consumers‘ responses to food safety incidents is important to policy analysts and the meat industry.

13

Table 1.3: 2000-2007 Food Recalls and Allergy Alerts from CFIA by Meat Category

2000

2001

2002

2003

2004

2005

2006

2007

Beef

2

11

10

16

4

4

0

11

General

0

11

8

11

9

4

5

9

Pork

5

3

1

0

5

0

1

4

Poultry

4

3

1

4

2

3

1

4

Seafood

1

7

9

9

2

0

4

5

Total

13

35

28

40

22

11

11

33

*Source: Canadian Food Inspection Agency (http://www.inspection.gc.ca, accessed on Sep. 2008)

6. Advertising

Many studies have had a focus on the effects of advertising on consumers‘ meat consumption. Different types of advertising, including both generic and brand advertising, have been found in meat demand analyses (Goddard, 1992; Verbeke and Ward, 2001; Wang, 2002; Freebairn, 2004; Lerohl et al., 2004; Halford et al., 2007; Amrouche et al., 2008; Chioveanu, 2008; Salma et al., 2009). Although some debate on the effects of advertising on market performance still exists in the economics literature, advertising has been a popular tool used by food processors and retailers to increase market share of a specific branded product or to launch new products to increase category sales. Generic advertising has also been used as a marketing strategy to combat health concerns.

Economic problem

The Canadian meat industry faces many challenges and it is important to understand the links between various factors and the industry, factors such as industry consolidation, value-added product development, introduction of private label products, product substitution across meat

14

types, changing household demographics and food safety and health perception can all influence demand affecting profits and revenues of farmers, processors and retailers. From a policy perspective, all of these issues can affect consumer health and welfare, industry profitability and possibly result in the need for new or changed regulations or policies.

On the other hand, from an industry marketing prospective, issues such as private label introduction, consumers' store and brand switching behaviour, meat type substitution, changing household demographics etc. will have an impact on developing a marketing strategy. Firms will also be interested in how consumers respond to new products, advertising and other sorts of promotion. These factors must be enunciated to understand how the Canadian meat industry can move forward to higher levels of customer satisfaction and value. The industry requires evolution to meet consumers‘ changing meat demand, especially for value-added meat products.

In this vein understanding consumer‘s value-added meat demand and behaviour, identifying historical and current trends in household demographics and testing for significant changes in household characteristics are all important. For example, household‘ attitudes and perceptions play a significant role in the store/brand and meat type choice behaviour, it is important to analyze how consumers determine their consumption decisions for the purchase of value-added meat products and brands. It is also important to understand, for policy formation how consumers choose between general grocery stores (including traditional retail cooperatives, such as Federated Co-operatives, etc) and multinational/ regional grocery chains and discount stores (such as Loblaws, METRO, Safeway, etc.). It is also necessary to find out how household spend their food dollar on meat products when their income increases.

Not only livestock producers, but also processors and retailers, need to understand meat demand changes in light of changing health perceptions, food safety concerns and trust in

15

brands and stores. This knowledge is required in order to predict changes in demand and develop new effective value added products and marketing strategies that respond to changing consumer needs, feeding into new product development; evaluating existing and potential policy opinion (such as, whether consumers respond as expected), which ultimately may increase the value of total sales.

Objectives

The overall objective of the study is to look at the structure of consumer value added meat purchasing behaviour (value added meat type choices, store choices as well as brand choices) in order to improve the understanding of recent food-at-home meat consumption patterns and discern new trends in value-added meat demand. Meat processors usually face two alternatives for branding policy: a processor either becomes a national company and sells meat products under its own brands (namely national brands), or cooperates with grocery store chains and produces meat products sold under the name of a store chain. Information related to this decision is related to the hierarchy of consumers‘ decision making: the process of selection decision among stores, meat by types (fresh, semi and fully processed meat) and meat by brands (national brands or private labels). For example, will the consumer choose a certain grocery store chain first and then make the meat type choice decision in-store? Or will they first make the decision of what types or brands of meat products they will purchase and then make relative grocery store shopping traffic?

How consumers‘ brand choice

(national brands vs. private labels) may be linked to store choices and subsequent in-store expenditure decisions? Which shopping scenario will drive store traffic in terms of volume of sales? Thus understanding the structure is important for the industry and meat producers to know where to introduce the new products and how to increase sales of value added meat products.

16

In particular, the study focuses on temporal and spatial patterns i.e. differences between similar households across geographic regions, as well as differences within individual households over time. In the study demographic and regional segments that historically and currently are purchasing different types and different levels of processing of meat will be identified, by segmenting them on total expenditure and share of meat expenditure. Trends in meat demand overtime, changes in demand between different value-added meat products, choices between grocery stores and national/store brands, demand for value-added/processed meat and UPC coded products are all examined in this study. Information on marketing variables such as market shares in grocery store chain will also be presented. Moreover the study will focus on household store and brand choice analysis related to value-added meat purchasing.

Specifically the research objectives for the study are threefold:

1. Using household level purchase data over the period 2002-2007 in order to: 

Understand how consumers make purchase decisions around fresh, semi-processed and fully processed products for four meat type categories: beef, pork, poultry and others (fish, lamb, etc)



Quantify the impact of demographic and regional characteristic differences in meat consumption behaviour, and these differences in the behaviour across meat types.

2. Using household level meat purchase data from 2002-2007 and store level advertising data(1999-2006) in order to: 

Find out whether Canadian consumers show consistency in purchasing patterns. Are they loyal to particular stores? Does this vary by region, by demographics, by store availability, is store advertising a factor?

17

3. Use household level purchase data from 2002-2007 and Nielsen Media Measurement's advertising data(2000-2008) in order to: 

Identify how consumers make the decisions about private label versus national brand products in their fully processed value-added meat category. Do product and brand advertising a factor? Does behaviour vary regionally and by demographics?

Implications

The analysis presented can be used to help Canadian industry participants to develop economically sustainable marketing strategies by identifying and matching consumer segments with product offerings, e.g. identify health-concerned consumers and quantify their willingness to pay for value-added products with fundamental health attributes. It can also be used to investigate the impact on meat expenditures of information such as advertising coverage, and new product introduction and marketing strategies. For example, Alberta Agriculture and Rural Development, developed an Alberta Livestock and Meat Strategy for the period 2008-2013. The Alberta Livestock and Meat Strategy is in line with efforts throughout Canada to strengthen the national livestock industry. Provinces such as Alberta, Manitoba, Saskatchewan, British Columbia, Nova Scotia and Prince Edward Island have all provided recent support to the livestock and meat sectors (Alberta Agriculture and Rural Development, 2008). The analysis presented can be used in developing new value-added meat products and marketing strategies that maximize carcass value for all suppliers along the valued-added meat supply chain. Economic benefits can be generated for the meat industry in terms of increased efficiency and increased demand for value-added meat products produced in Alberta and Canada.

18

Literature Review And Methods

Introduction

In today‘s food industry, ―value added‖ is a key term with various definitions. Value added is a very broad concept that encompasses many attributes such as seasoned, pre-cooked, healthy, convenience, prepackaged, etc. The term value-added can be interpreted in many ways (Kinsey et al., 1993; Gaquez-Abad and Sachez-Perez, 2009).

United States Department of Agriculture‘s (USDA, 2009) ―definition of Value-Added includes four categories that increase the value that is realized by the producer from an agricultural commodity or product as the result of:



A change in its physical state (a change in physical state is only achieved if the product cannot be returned to its original state.);



Differentiated production or marketing, as demonstrated in a business plan (the enhancement of value must be quantified by using a comparison with products produced or marketed in the standard manner, for example, organic carrots, free range chicken);



Product segregation (the enhancement of value should be quantified to the extent possible by using a comparison with products marketed without segregation.), for example genetically modified corn and non-genetically modified corn grown on the same farm; or



Agricultural commodities or products used as a source of farm or ranch based renewable energy.

19

Carrboro Farmers Markets, Inc (2007) defines value added products as ―Farm produced value added meat products are further processed meat products made from raw ingredients. Farmer vendors must raise a minimum of 51% of the raw ingredients in a value added meat product.‖ Statistics Canada (2007) defines value added as ― the value that is added to a product by, for instance, producing baked goods from flour, sugar, salt, yeast, eggs, water, and vegetable oils.‖

Definition of value-added meat products

One of the definitions for value added meat products is from Meat and Livestock Australia (MLA, 2008) includes: ‖ 

Adding extra ingredients to the raw meat, such as bread crumbs for schnitzel or vegetables for stir fries



Cooking the raw meat prior to selling, such as pre-cooked roasts



Processing meat into small goods, such as pastrami



Prepared products for retail such as sausages, patties or kebabs



Packaging meat for a longer shelf life, eg modified atmosphere packaging‖

The classification and definition of value added meat products in this study are according to the definition above and the availability of data from the sources used. "Value added" is defined as the level of value added processing in the meat products. There is great variety in the level of processing different meat products are subject to – in some cases products are processed to the point that they are ready to eat (luncheon meats) while others are merely seasoned or cut into small pieces ready for cooking. In this study an attempt is made to classify product by three different levels of processing, no other published study has examined meat by level of processing. Meat products are grouped into three categories: fresh, semi-processed and fully processed meat for four types of meat according to "meat cut" and "meat processed form" information provided by Nielsen Homescan™ database.

Both

20

"PRFRM" (meat processed form table, as shown in Table 2.1 below) and "PRTYP" (meat processed type table, as shown in Table 2.2 below) information are applied in the meat classification (Table 2.3). For example, if one product is in the fresh category in the "PRTYP" table, but is in the fully processed meat category in the "PRFRM" table, then it is grouped into fully processed meat category after combining both types of category information. UPC coded and random weighted meat products are all included in the sample data. Table 2.1: Nielsen Homescan™ panel product processed form (PRFRM) Fresh meat 340561 ALL TYPES 345061 ASSORTED 340531 BACKS 364811 BREAST 353575 CASINGS 340506 CHOPS 450802 CHOPS W/FILLET 436511 CHUB 351077 CHUNK 317632 CUBES 340533 CUT UP 129253 DICED 340530 DRUMSTICKS 345070 ESCALOPE 340513 FILLETS 365032 FINGERLINGS 353256 FLAP 129261 GROUND 340527 LONDON BROIL 340539 MEDALLIONS 340560 MINCED 129263 MINI 129227 N/A 129239 NOT APPLICABLE 468358 OSSO BUCCO 317578 PIECES 350888 PORTION 428240 RIB FINGERS 352967 RIB STRIP 345031 RIBLETS 340518 RIBS 370999 RINGOS 365036 RINGS 340507 ROAST 319240 ROLL

Semi processed 363885 BACON 340528 SAUSAGE 356417 ALOUETTE 394361 BROCHETTE 363900 BROCHETTES 365095 CARVED 425822 CHOPPETTES 340555 COTTAGE ROLL 371000 DRUMLETS 340558 HEAD 321308 KABOB 340509 KABOBS 364924 MEATBALL 340536 MEATBALLS 340526 ROULADEN 345006 SALT 345046 SAUSAGE MEAT 340748 SAUSAGES 363895 SKEWERS 363901 SOUVLAKI 363898 STIRFRY

340537 340524 363886 317447 345040 410596 129258 340563 129250 364953 365082 364861 340508 436512 365089 364975 340554 365090 365084 365046 364960 340562 340517 344949 340521 365129 346623 340540 356405 355660 345044 364961 365094 410823 365120

Fully processed SCALLOPINI SCHNITZEL SLICE SLICES BALLS BAVETTE BITES BURGERS CHIPS CHOMPERS CRISPS CUTLET CUTLETS CUTLETS/DRUMMETTES DINO SNACKS DUMPLING FINGERS FLINGS FRANKFURTERS FRIES FRITTERS MEATLOAF NUGGETS PATTIES PAUPIETTES PEROGIES POPCORN SATAY SAUSAGE CHAPLET SAUSAGE KABOB SAUSAGE PATTIES SNACKOSAURS SNAKE BITES SPIEDINI SPIRALS

21

356409 372928 353574 340516 356958 340512 375130 372576 363894 364111 129243 364830

ROSETTE SCRUNCHIONS SLAB SPLIT SPLIT/TIPPED STEAK STEAK CUBED STEAK/ROAST STEAKS UNSPECIFIED WHOLE WINGS

364979 340552 129249 365031 129260 364931 357815 340515 129242 351060

STEAKETTE STEW STICKS STIX STRIPS TEAZERS TENDERS TOURNEDOS SLICED SLICED/PIECE

Table 2.2: Nielsen Homescan™ panel meat processed type table (PRTYP)

343873 345502 446497 344999 355289 363270 310656 413242 454407 346191 99976 139654 139692 347426 382313 139662 354334 139655 345065 344954 344967 139688 344953 353258 354339 343879 355654 344950 360470 444255 353254 343210 416020

Fresh meat AIR CHILLED ANGUS ANGUS GRADE AAA BRAISING BROILER BROILER/GRADE A BUTTERFLIED CALIFORNIA STYLE CANADIAN ANGUS CUBED DRY FAST FRY FREE RANGE FRENCH STYLE FRENCH STYLE/ANGUS FRENCHED FRENCHED/GRILLING FRYER FRYER FREE RANGE FRYER GRADE A FRYER/UTILITY FRYING GRADE A GRADE A/MARINATED GRADE AAA GRAIN FED GRAIN FED/TENDERIZED GRILLING GRILLING/ANGUS HOTEL STYLE MATURE MILK FED MILK FED/HOTEL STYLE

139657 345068 355657 139693 349972 345060 139673 345100 345099 139670 350881 356688 363013 366374 357826 357823 352679 356402 139660 346983 344974 360469 354336 346197 352964 367197 345098 349791 345004 361539 45311 416019 345069

Semi processed BASTED BASTED/GRADE A BASTED/STUFFED BBQ BRAISING/SEASONED CORNED CORNMEALED CURED CURED/CORNMEAL DELICATED DOUBLE SMOKED FRENCH STYLE/MARINTD FRENCH STYLE/SEASOND FRENCHED SEASONED FRENCHED/GRAIN FED FRENCHED/SEASONED GARDEN STYLE GRILLING/MARINATED MARINATED MARINATED/SEASONED MARINATING MARINATING/ANGUS MATURE/SEASONED PEAMEAL PICKLED ROASTED/BASTED ROASTED/SEASONED ROASTING/STUFFED SALTED SALTED/CURED SEASONED SEASONED/ANGUS SEASONED/BBQ

370997 368110 340868 347249 361541 353577 368098 368096 355665 45337 368108 368113 368095 139689 352675 368114 368109 99973 368104 368105 139298 368117 462862 368387 45315 368091 350884 368094 368115 139219 368107 368090 368100

Fully processed BAKED BATTERED BREADED BREADED/FAST FRY BREADED/GRAIN FED BREADED/TENDERIZED BURRITOS CASSEROLE CHICKEN FRIED CHILI CHIMICHANGAS COOKED CORNDOGS COUNTRY STYLE CRISPY CROQUETTES DIM SUM DINNER EMPANADA ENCHILADAS FAJITA FILLO FILO FRENCHED/BREADED FRIED GRILLED MECHOUI PASTRY PATTIES PIE POTSTICKER PREPARED QUESADILLA

22

345007 345012 365511 45305 340746 345775 344945 370998 368093 139653 345063 348173 345032 352981 345015 346193 345041 351076 139663 365510 434599 344964 361952 352673 110204 139661 354337 346196 361950

MILK FED/TENDERIZED MINUTE MINUTE/FAST FRY N/A NEW ENGLAND STYLE NEW YORK STYLE NOT APPLICABLE POT ROAST ROAST ROASTER ROASTER GRADE A ROASTER UTILITY ROASTING ROLLED SIMMERING SIMMERING/FAST FRY STEWING SUGARBUSH TENDERIZED TENDERIZED/FAST FRY TENDERIZED/GRILLING TEXAS STYLE TRIMMED TUSCANY UNSPECIFIED UTILITY UTILITY/MATURE VERMONT YOUNG/GRADE A

407174 345027 344966 343877 344973 139671 314401 361544 139267 99965 310653 469255 353259 357819

SEASONED/DELICATED SEASONED/FAST FRY SEASONED/FRYER SEASONED/GRILLING SEASONED/STUFFED SMOKED ST LOUIS STYLE ST LOUIS/SEASONED STIR FRY STUFFED STUFFED/BASTED STUFFED/CURED STUFFED/FRYER STUFFED/MILK FED

374025 382315 345071 344989 110130 352970 368092 368102 368106 345028 368116 139676 368097 368101 353589 368118 368120 368099 110376

QUICK QUICK/ANGUS RANCH CUT ROASTED ROTI ROTISSERIE SAMOSAS SANDWICH SAUSAGE PASTA SEASONED/BREADED SHEPHERD PIE SLOW COOKED STEW TAQUITOS TENDERIZED/BREADED TORNADOS WONTON WRAPS BLACK FOREST

After classifying all meat products in the dataset into one of twelve categories the structure of the consumer choice problem for value added meat can be expressed as in Table 2.3. Consumers are in general assumed to determine how much spending they will entertain for meat and then to allocate that spending to different meats by type and by level of processing.

23

Table 2.3: Classification of value added meat in the study

Overview of value added agricultural products demand

Understanding recent food-at-home meat consumption patterns is important for meat manufacturers to develop and evaluate product development and marketing strategies and identify target consumer segments that are likely to increase their consumption of particular value-added meat products. From a public health perspective understanding consumer meat purchasing behaviour can facilitate the design of health recommendations and regulations, the recent public health focus on sodium is an example of a public health concern that could change the ways meats are processed. Understand consumer‘s decision making can also help

24

to maximize meat manufacturers‘ revenues and minimize their costs. Meat manufacturers can influence consumer purchase decision through various ways:

1. Product differentiation by pricing(Connor and Peterson, 1992; Hinloopen and Martin, 1997; Degeratu et al., 2000; Besanko et al., 2003; Fok et al., 2006; Bontemps et al., 2008; Yuxin et al., 2008; Gonzalez-Benito et al., 2009; Moon and Voss, 2009; Schnettler et al., 2009) 2. Product differentiation by investment in advertising (generic or branded advertising) (Cozzarin and Goddard, 1992; Alston et al., 2000; Verbeke and Ward, 2001; Boetel and Liu, 2003; Srinivasan and Bodapati, 2006; Erdem et al., 2008; Silberstein and Nield, 2008), 3. Product differentiation by distribution channels (through different grocery store chains, different store format, store loyalty) (Beaumont, 1988; Konishi, 2005; Ailawadi et al., 2008; Eacute et al., 2008; Litz and Rajaguru, 2008) 4. Product differentiation by quality/attributes, by amount of value adding (fresh, semi and fully processed, health and convenience) (Huang and Fu, 1993; Kinsey et al., 1993; Yiannaka et al., 2002; Enneking et al., 2007; Anders and Moeser, 2008) 5. Product differentiation by branding (make the market strategy on becoming a nationa company or coordinating with a grocery chain, brand loyalty), etc. (Connor and Peterson, 1992; Chintagunta, 1993b; Hinloopen and Martin, 1997; Chintagunta et al., 2001; Jin et al., 2005; Dolekoglu et al., 2008; Schnettler et al., 2008; Esbjerg and Bech-Larsen, 2009; Gaquez-Abad and Sachez-Perez, 2009; Liljander et al., 2009)

Summary of Canadian meat demand studies

A number of relevant meat demand studies have been conducted in Canada since the early 1970‘s. The first Canadian meat demand study in the literature was published in 1961 (Yeh, 1961), the author used annual disappearance data for the period 1929 to 1958 to investigate

25

how consumers reacted to changes in the prices of beef and pork and in disposable income. Kulshreshtha and Wilson (1972) focused only on beef demand (disappearance) in their study. Tryfos and Tryphonopoulos (1973) used annual disappearance data for the period 1954 to 1970 for beef, pork, chicken, lamb, veal and turkey demand analysis. Hassan and Katz (1975) applied Seemingly Unrelated Regression (SUR) analysis to estimate price and income elasticities of demand (disappearance) for beef, pork, lamb, veal, chicken and turkey. Hassan and Johnson (1979) applied Box-Cox transformations to select from a variety of functional forms (Linear, Double log, semi-log, log-inverse and general), and showed that that different specifications can lead to different meat demand elasticity results. Hassan and Johnson (1983) applied different estimation procedures (OLS, GLS and SUR) with seasonality hypotheses for the demand for beef, pork, veal, chicken and turkey. Young (1987) and Atkins et al. (1989) attempted to analyze the structural change in Canadian meat demand. Young (1987) used a single-equation approach, and found evidence of structural change in Canadian demand for pork, chicken and turkey, but no such evidence for beef (again using disappearance data). However Atkins et al. (1989) found a structural break in beef demand.

In many Canadian meat demand studies, the AIDS model, explaining expenditure shares in a system of equations, have been used since 1991. However the importance of functional form selection in producing meaningful economic characteristics of consumer behaviour cannot be underestimasted. Alston and Chalfant (1991) compared different functional forms and concluded that an incorrect use of functional form can lead to a finding of structural change in meat demand. The authors concluded that better data or better methods were needed for that demand study. Chalfant, Grey and White (1991) analysed meat demand using an AIDS demand system for beef, pork, poultry, and fish. They found a small positive cross price elasticity (economic substitution) between fish and pork. In their study the meat expenditure elasticity is positive for chicken and fish, but negative for beef and pork, suggesting that beef demand will decline as an individual‘s expenditure on meat increases. Chen and Veeman (1991) used a dynamic AIDS model of Canadian meat demand and compared it with a static AIDS model. The authors examined structural change in meat demand by testing for nonconstancy of the parameters of the non-linear system. The reason for the structural change

26

could be caused by increasing health concerns regarding diets and growth of fast food outlets. Reynolds and Goddard (1991) also focused on the structural change and analyzed demand for beef, pork and chicken. Their results showed that the structure of Canadian meat demand has changed gradually over the period 1975 to 1984. The elasticities were significantly different before and after the structural change. The results indicated that structural change was biased away from beef consumption, in favour of chicken consumption.

Cozzarin and Goddard (1992) first included advertising as a factor in meat demand. They compared two types of models

the Translog and AIDS demand systems to analyse

disappearance of beef, pork and chicken. Moschini and Vissa (1993) applied a mixed demand approach to analyze Canadian meat demand. They found that the estimated own price elasticity of chicken demand is greater in the mixed demand system, others are the same as those in a direct Rotterdam model. Eales (1996) used both the static and dynamic AIDS and IAIDS to test for endogenous RHS variables. All the AIDS estimates were in agreement as to the responsiveness of demand. The results indicated that IAIDS models were more "elastic" than AIDS models. Xu and Veeman (1996) applied joint non-nested testing for both the linearised almost ideal and Rotterdam models. The test results for structural change shows that the gradual transition AIDS model is preferred over the gradual-transition Rotterdam model for Canadian meat consumption. In a departure from the traditional approach of examining aggregate disappearance data on meat, Salvanes and DeVoretz (1997) focused on the specification of Canadian household demand for fish and meat products. The authors applied tests for separability by estimating different demand systems over different processed levels for fish and meat. The test indicates that fish is not weakly separable from the two other aggregated categories. And at an aggregated level Canadian fish demand cannot be modeled from meat.

Lerohl et al.(2004) and Lomeli (2005) included media influences on changes in consumption of meat products in Canada using both time series (disappearance) and cross sectional (household Family Food Expenditure Survey) data. Results found that pork-safety issues had

27

negative and significant own consumption effects. and positive cross-effects for beef. Pork generic advertising had own positive effects, while pork consumption was negatively affected by chicken generic advertising. Both beef brand and beef fast food restaurant advertising increased beef consumption. Lambert et al. (2006) analysed regional differences in meat and fish demand across Canada. A QUAIDS demand system was applied in the study using Canadian household food expenditure surveys conducted in 1992 and 1996. The authors found that various variables including prices, age, ethnicity and real total meat and fish expenditure affected the probabilities of purchase. Maynard et al.(2008) applied a double-hurdle count data model to test frequency of BSE media coverage which affected a household purchasing a beef entre in a restaurant. Anders and Moeser (2008) applied weekly retail and household scanner data to estimate consumer demand for organic and conventional fresh beef products in the Canadian retail market. The results indicated that ―organic beef was highly dependent on price and expenditures, whereas demand for conventional beef was mostly driven by income, habits and ‗typical‘ Canadian seasonal beef consumption patterns.‖ Table 2.4 Summary of Canadian meat demand studies Authors

Yeh, 1961

Kulshreshtha and Wilson, 1972

Meat types

Functional forms

beef and pork

Double logarithmic

Time series data Estimates were consistent for the period with those obtained in 1929 to 1958 previous studies

Linear

Time series data Estimates were consistent for the period with those obtained in 1949- 1969 previous studies

Linear

Time series data Estimates were consistent for the period with those obtained in 1954 to 1970 previous studies

Linear

In addition, most of the Time series data elasticities are in keeping for the period with comparable results 1954 to 1972 obtained from other studies

beef

Tryfos and beef, pork, chicken, Tryphonopoulos, lamb, veal and 1973 turkey

Hassan and Katz, 1975

beef, pork, lamb, veal, chicken and turkey

Data

Results

28

Hassan and Johnson, 1979

Hassan and Johnson, 1983

Young, 1987

Atkins, Kerr and McGivern, 1989

Alston and Chalfant, 1991

Linear, Double Time series data beef, pork, veal, log, semi-log, for the period chicken and turkey log-inverse 1965 to 1976 and general

beef, pork, veal, chicken and turkey

For the existence of fixed quarterly or seasonal Time series data effects, dummy variables for the period with fixed coefficients 1965 to 1977 should be used in the analysis.

Linear, Double Time series data beef, pork, chicken, log, linear-log for the period turkey and Box-Cox 1968 to 1986

beef, pork and chicken

Linear

Linear, Double beef, pork, poultry log, LA/AIDS, and fish Rotterdam

Chalfant, Grey beef, pork, poultry, and White, 1991 and fish

Chen and Veeman, 1991

Linear

beef, pork, chicken and turkey

LA/AIDS

LA/AIDS

different specifications can lead to different elasticity results.

found that the income elasticities were very sensitive to the model specifications and some specifications produced negative elasticities

Time series data Found a structural break in for the period beef demand. 1968 to 1986 time series observations from 1960 to 1988

incorrect use of functional form can lead to a finding of structural change in meat demand

time series observations from 1960 to 1988

small positive elasticity between fish and pork, consumption is positive for chicken and fish, but negative for beef and pork

structural change in meat demand, could be caused Quarterly timeby increasing health series data from concerns regarding diets 1967 to 1987 and growth of fast food outlets

29

The results indicated that Quarterly timestructural change was series data from biased away form beef 1968 to 1987 consumption and to chicken consumption.

Reynolds and Goddard, 1991

beef, pork and chicken

Cozzarin and Goddard, 1992

beef, pork and chicken

Translog and first included advertising time-series data AIDS factor in meat demand

Moschini and Vissa, 1993

beef, pork, and chicken

Rotterdam model

own price elasticity of chicken demand is greater in the mixed demand system, others are the same as those in a direct Rotterdam model.

Eales, 1996

beef, pork, and chicken

AIDS and IAIDS

The results indicated that Quarterly timeIAIDS models were more series data from "elastic" than AIDS 1970 to 1992 models.

AIDS and Rotterdam

The test results of structural change shows that the gradual transition quarterly retailalmost ideal model is level data from preferred over the gradual1967 to 1992 transition Rotterdam model for Canadian meat consumption.

Xu and Veeman, 1996

beef, pork and chicken

beef, pork and Salvanes and chicken, DeVoretz, 1997 fish(fresh/processed)

Lerohl et al., 2004; Lomeli, 2005

beef, pork, and chicken

LA/AIDS

LA/AIDS

time series observations from 1980 to 1990

Statistics Canada 1986 Food Expenditure Survey Public Use Microdata Files

Canadian fish demand cannot be modeled separately away from meat.

Canadian meat Generalized Pork generic advertising market data from Box-Cox has own positive effects 1976 to 2001

30

Lambert et al., 2006

Maynard et al., 2008

Anders and Moeser, 2008

fish, beef, pork, chicken, and other meats

beef entrees

organic and conventional fresh beef

QUAIDS

The authors find that various variables Canada‘s Food including prices, age, Expenditure ethnicity and real total Survey for 1992 meat and fish expenditure, and 1996 on the probabilities of purchase

BSE media coverage did Canadian FAFH Double-hurdle not systematically affect purchasesfrom model fast food purchases among 2000 to 2005 Alberta consumers.

AIDS

Nielsen retail scanner data 2000–2007

Organic beef is highly dependent on price and expenditures, whereas demand for conventional beef is mostly driven by income, habits and ‗typical‘ Canadian seasonal beef consumption patterns

Hierarchy of consumer purchase decision making in the study

The focus of this study is on how meat consumers make their decisions to purchase value added meat products: do they select store, then fresh versus semi-processed versus fully processed? Do they choose meat type (beef, pork, for example) at first, second or third stage of their decision structure (i.e. before store, before type, before brands).

Wrigley (1988) finds the sequence of shopping decisions that ―consumers choose a store knowing that they can obtain a desired brand there, then branding, promotion and advertising support are that much more important.‖ Brucks (1988) suggested a sequence of choices as first choose stores and then make the brand choices. Guadagni and Little (1998) concluded that a decision tree for a customer on a shopping trip that ‖ the customer may be viewed as deciding sequentially when to buy and then what to buy but with interaction between the

31

decisions‖. Bucklin and Lattin (1986) and Guadagni and Little (1987) both regard purchasing as a sequential process: choose product category at the first stage, then choose a brand. Krish- namurthi and Raj (1988) view brand choice and purchase quantity as related decisions and model them as such. Gupta (1988) models brand choice (what to buy), purchase quantity (how much to purchase) and interpurchase time (when to shop) decisions independently. Kahn and Schmittlein (1989) consider the hierarchical purchase process as that consumers must first decide to enter the store to shop before choosing brands.

Chiang (1991) views the decision process as "whether to buy," "what to buy" and "how much to buy". Chintagunta (1993a) concluded that household purchase behavior contains three components: purchase incidence, brand choice and purchase quantity. Wilkie (1994) described consumer decision process of three stages: sensing, selecting, and interpreting. Piedra et al. (1995) concluded that ―nearly two thirds of U.S. consumers purchase at least three different types of meat per week. Some meat choices are made prior to shopping, others are made after in-store visual inspection of cuts and prices.‖ Kamakura et al. (1996) conclude that ―some consumers may first choose what brand to buy, and then choose product form, size, or flavor. Others may first choose the flavour in a shopping occasion, and then choose among the brands offering that flavour.‖

Degeratu et al.(2000) divided the choice decision into a two-stage choice model in which customers first choose the store type in which they shop and then make brand choices. Sood et al. (2004) and Chernev (2006) views choice as ―a hierarchical decision process as two different stages (instead of two independent choices): first make an assortment selection and then selectan option from that assortment.‖ Hui et al (2009) divide a shopping path into three stages of visit, shop, and buy decisions. They conclude that factors of time pressure, licensing, and social influence of other shoppers influence the consumer in-store decision making process. Ailawadi et al. (2008) private label have an influence on consumers‘ expenditure share of different grocery stores. Gaquez-Abad and Sachez-Perez (2009) view the purchase of olive oil as a hierarchical process: ―consumers first decide what type of oil (e.g., soya,

32

olive, sunflower, etc.) they want. In this step, oil price is a function of quantity and production patterns. Then the consumers decide which brand to buy (brand choice behavior). Juhl et al., 2006; Esbjerg and Bech-Larsen, 2009)This is at least the case in the short run, as consumers typically will not visit another store if they cannot find their preferred brand in the store they have chosen. Some studies (Juhl et al., 2006; Esbjerg and Bech-Larsen, 2009) indicate that consumers choose stores before they choose brands, then manufacturers should focus on the assortments of the retail chains with the best locations.

Based on the previous hierarchy of choice studies, it is reasonable to assume that when consumers allocate budget shares within the meat category, weak separability of consumer preferences can be invoked to examine purely the hierarchical budgeting processes for meat in the shopping decision (Montgomery, 2002). The possible decision flows for the meat purchase decision are among: 1. Stores choice; 2. Meat choice by types (fresh, semi and fully processed meat); 3. Meat choice by brands (National brands vs. Private labels). The following three examples of decision flows are among many possible combinations that could be postulated, consumers could also use other decision processes. Assumption 1: one possible decision making process could be: consumers first choose where

to shop, and then make the decision of what type of meat to purchase, and finally choose among different brands.

33

Assumption 2: An alternative process could be: consumers first make the decision of what

types of meat they need to purchase, then they choose related brands, finally they decide where to buy the certain meat products.

Assumption 3: Or consumers could first make the decision of what types of meat they need to purchase, then they choose where to shop, finally they make the brand decision for the certain meat products.

34

Model structure and econometric method

Based on the comprehensive review of the issues related to meat demand analysis in the previous section, many different techniques are employed in this study to explain consumers‘ choice about meat types, meat brands and where to shop. Different functional form and model methodologies are applied to address the impact of prices and other economic determinants (elasticities) by demographic and regional characteristics in meat consumption behaviour, and these differences in the behaviour across meat types. In this study panel data, data on households purchasing behaviour across time, will be used. Panel data analysis can provide a large number of data points, hence improving the efficiency of econometric estimates. Hsiao (2003) demonstrates several benefits from using panel data, including controlling for individual heterogeneity while a time series study or a cross section study cannot (Hsiao, 2003). Panel data can provide more variability, more efficiency and more degree of freedom. Panel analysis is also able to identify and measure effects that are simply not detectable in pure cross section or pure time series data, because panel data have double subscripts on their variables (Baltagi, 2008), ie

yit    X it'   it i=1,…, N; t=1,…, T

Panel data sets are two-dimensional, where i represents households, individuals or countries (cross section dimension) and t denotes time points (time series dimension).  is a scalar, ' while  is K * 1 and X it is the it th observation on a vector of k nonstochastic regressors.

35

Different assumptions can be made on the precise behavioural structure using panel data. Two main models are the one-way fixed effects regressions and random effects regressions (Baltagi, 2008)

The fixed effects model is denoted as

yit = α + β'Xit + uit, uit = μi + νit,

where the μi are assumed to be elements of fixed parameters and they are fixed over time, this is called the fixed-effects model. Essentially variation across individuals is defined as a fixed effect difference between individuals.

The random effects model assumes in addition that the error terms for individuals are defined as random disturbances drawn from distributions with the following specifications :

i

IID(0,  2 )

and

it

IID(0,  v2 )

The two error components

i and  it are independent from each other (Baltagi, 2008).

Usually household samples vary in a random manner, so random models are appropriate specifications in dealing with household panel data (Baltagi, 2008). Thus, a random effects model is used in this study.

36

The general structure of panel data is based on a matrix of N units and T periods. When the number of time observations is the same as each individual observation unit (N units and T periods), the panel is called a balanced panel, in which case the matrix is completely filled. A more realistic alternative is when some observations are missing, the number of household observations per each period varies, and then the panel is called an unbalanced panel (Baltagi, 2008). A balanced panel approach is used in the study. In this study, store choice, brand preferences and household demographic characteristics are all assumed to affect the consumers‘ demand. The decision making process follows a hierarchical process. Due to the nature of our household-level panel data, with some zeroconsumption problems and based on previous related demand studies, a Working-Leser demand system is used in the analysis. The Working-Leser model was originally discussed by Working (1943) and Leser (1963). Working (1943) first applied the log-linear budget share specification to the model and Leser (1963) found that this functional form fit better than some other alternatives. Deaton and Muellbauer (1980a) provide more detailed information on this functional form. Basic Engel functions represent the relationship between consumption and consumer's income level. In addition, household consumption is also affected by demographic and socio-economic variables. In the Working-Leser model, each expenditure share is represented by a linear function of the log of prices and of the total expenditure and household demographic variables. The Working-Leser food demand function can be expressed as:

i  a0  ai *log x   j  ij *ln( p j )   k  ijH k   it where (i,j) represents given meat products; wi is the expenditure share of a particular meat i; pj is the price of meat j; and X is the total expenditure of all types of meat included in the model. Hk represents the household demographic variables. The expenditure elasticity formulae for the Working-Leser model (ei) can be shown as:

a  ei  1   i   wi 

37

The uncompensated own (j =i) and cross (j ≠i) price elasticities (eij) are defined as follows:

  ei   ij   ij   wi 

i, j  1,..., n

where  ij is the Kronecker's delta, it is a function of two variables, usually integers, which is 1 if they are equal (if i = j), and 0 otherwise. In this study, expenditure, own-price and crossprice elasticities are evaluated at sample means.

38

Demographic Data and Descriptive Statistics

Introduction This research project mainly contains three sections of analysis: consumers' meat demand analysis by level of processing, consumers' store choice analysis in meat purchasing, and analysis of brand choices between national brands and private labels (store brands) for the fully processed meat category. The data for the three analyses are sourced from the Nielsen Company Homescan™ panel data for calendar years 2002 through 2007. These data are taken from a sample of households that are representative for the Canadian population (as shown in table 3.1) by year. Each household was provided with a scanner machine by Nielsen in which they could scan and record all items purchased in different grocery stores in a given period, as well as demographic information about the household. Nielsen Homescan™ panel data is a unique dataset that consists, in this case, of all meat purchases by 16,515 Canadian households from 2002 to 2007, not necessarily all households are present in the sample for each year. Meat categories include fresh and frozen meat cuts of both random weighted and UPC coded products. The database also contains socioeconomic and demographic characteristics of the households such as age, income, region, household size and education, presence of children, etc.. Since not all participant households stayed in the panel in all six years from 2002 to 2007, Table 3.2 shows the proportion of households that stayed in the panel for each year. Some of the households dropped out of the panel and other households participated in the panel for the subsequent year. In order to effectively address the study objectives, the data used for the empirical analysis is a balanced panel from 2002 to 2007 after excluding households with missing information on important variables and households not participating over the entire six-year period. The final balanced panel data sample covers households who stayed in the panel and had purchase information in all six years, leading to a total of 4322 households at the national panel and 508 households in Alberta and 1036 households in Ontario. All the expenditure and quantity data have been

39

aggregated to yearly data to control for the large number of zero observations, at a monthly level. Meat and store expenditure data are expressed in terms of Canadian dollars. Table 3.1: Comparing Sample Balanced Data with 2006 Census Profile of Canada Nielsen Homescan™

2006 Census Profile Canada

Region (n=4322) Maritimes 14% 8% Quebec 25% 24% Ontario 25% 39% Man/Sask 10% 7% Alberta 13% 10% BC 13% 13% Household Head Age Ontario Alberta Canada Ontario 18-34 2% 5% 19% 19% 35-44 19% 18% 15% 16% 45-54 26% 30% 16% 15% 55-64 22% 22% 12% 11% 65+ 31% 24% 14% 14% Household Size Ontario Alberta Canada Ontario Single Member 25% 27% 27% 24% Two Members 40% 40% 34% 32% Three Members 14% 12% 16% 17% Four Members 13% 14% 15% 17% Five - Nine Plus Members 8% 7% 9% 11% Age & Presence of Children Ontario Alberta Canada Ontario No children 78% 78% 77% 75% Have children 22% 22% 23% 25% Household Head Education Ontario Alberta Canada Ontario NOT HIGH SCHOOL GRAD 14% 13% 24% 22% HIGH SCHOOL GRADUATE 15% 18% 26% 27% COLLEGE OR UNIVERSITY 71% 69% 51% 51% Income Ontario Alberta Canada Ontario < $20,000 9% 8% 7% 7% $20,000-$29,999 12% 14% 9% 8% $30,000-$39,999 12% 13% 13% 11% $40,000-$49,999 11% 11% 13% 11% $50,000-$69,999 19% 19% 22% 21% $70,000+ 38% 36% 36% 42% National Urban vs. Rural Ontario Alberta Canada Ontario RURAL 32% 31% 19% 15% URBAN 68% 69% 81% 85% Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Alberta 22% 15% 16% 10% 11% Alberta 25% 34% 16% 16% 10% Alberta 82% 18% Alberta 23% 26% 50% Alberta 5% 6% 10% 11% 22% 45% Alberta 17% 83%

40

Table: 3.2 The proportion of households who participated in the panel from 2002-07 Year

Number of participating Canadian households

2002 9580 2003 9231 2004 10044 2005 9933 2006 9304 2007 9582 Source: Nielsen Homescan™ panel data 2002-2007

Socioeconomics and demographic information and definitions

Nielsen Homescan™ panel data has detailed information on household socioeconomic and demographic characteristics for each of the panellist. The sample data used in all three studies in the project focus on household panellists in Ontario and Alberta for calendar years 2002 through 2007. The socioeconomic and household demographics used in all three studies in the project include: household size, household income, household head age, education, and presence of children, language, urbanization, and province. In this section, the definition of household demographic variables used in the empirical analyses are provided. In addition, descriptive statistics associated with the study sample comparing households in the provinces of Ontario and Alberta are presented. Income Household income levels are recorded as a categorical variable (income falls within a range such $25,000 to $34,000) in the Nielsen Homescan™ panel data. Mid-points are used to approximate a continuous income measure. Table 3.3 and 3.4 present the income classes and mid-point values for the sample data and comparable Canadian Census data, for 2006. The

41

frequency distribution by year implies that the study sample data is roughly representative of income classes in the Census data.

Table 3.3 The income classes and mid-point value for the sample data for Ontario Nielsen Homescan™ panel data 2002-2007

Income class (CAD$)

Midpoints

< $20,000

10000

$20,000-$29,999

24999.5

$30,000-$39,999

34999.5

$40,000-$49,999

44999.5

$50,000-$69,999

59999.5

$70,000+

74999.5

Total

YEAR

2002

2003

2004

2005

2006

2007

Census, 2006 Total

Canada

Ontario

7.1%

6.6%

9.2%

7.6%

12.6%

10.9%

12.6%

11.3%

22.3%

21.5%

36.3%

42.0%

Count

100

100

89

90

90

73

542

HH%

9.7%

9.7%

8.6%

8.7%

8.7%

7.0%

8.7%

Count

123

123

140

125

125

114

750

HH%

11.9%

11.9%

13.5%

12.1%

12.1%

11.0%

12.1%

Count

126

126

131

122

122

119

746

HH%

12.2%

12.2%

12.6%

11.8%

11.8%

11.5%

12.0%

Count

115

115

109

119

119

112

689

HH%

11.1%

11.1%

10.5%

11.5%

11.5%

10.8%

11.1%

Count

206

206

186

189

189

179

1155

HH%

19.9%

19.9%

18.0%

18.2%

18.2%

17.3%

18.6%

Count

366

366

381

391

391

439

2334

HH%

35.3%

35.3%

36.8%

37.7%

37.7%

42.4%

37.5%

Count

1036

1036

1036

1036

1036

1036

6216

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Table 3.4 The income classes and mid-point value for the sample data for Alberta Income class (CAD$) < $20,000 $20,000-$29,999 $30,000-$39,999 $40,000-$49,999 $50,000-$69,999 $70,000+ Total

Nielsen Homescan™ panel data 2002-2007 Midpoints 10000 24999.5 34999.5 44999.5 59999.5 74999.5

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Canada

Alberta

Count

43

43

38

39

39

30

232

7.1%

5.4%

9.2%

6.4%

12.6%

10.2%

12.6%

10.9%

22.3%

21.7%

36.3%

45.5%

HH%

8.5%

8.5%

7.5%

7.7%

7.7%

5.9%

7.6%

Count

78

78

74

68

68

55

421

HH%

15.4%

15.4%

14.6%

13.4%

13.4%

10.8%

13.8%

Count

69

69

62

63

63

65

391

HH%

13.6%

13.6%

12.2%

12.4%

12.4%

12.8%

12.8%

Count

55

55

56

54

54

55

329

HH%

10.8%

10.8%

11.0%

10.6%

10.6%

10.8%

10.8%

Count

107

107

104

92

92

76

578

HH%

21.1%

21.1%

20.5%

18.1%

18.1%

15.0%

19.0%

Count

156

156

174

192

192

227

1097

HH%

30.7%

30.7%

34.3%

37.8%

37.8%

44.7%

36.0%

Count

508

508

508

508

508

508

3048

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

42

As appears in Table 3.3, in Ontario the aggregate frequency of households falling into income classes: less than $20,000, $20,000-$29,999 and $30,000-$39,999 are higher in the Nielsen panel data than in Census 2006. This difference is compensated for with a lower frequency of income class of $50,000-$69,999 and $70,000+ in Nielsen Homescan™ panel than in Census 2006. The same distribution also appears in Alberta. The aggregate frequency of households falling into income classes in Alberta: less than $20,000, $20,000-$29,999 and $30,000-$39,999 are higher in Nielsen Homescan™ panel data than in Census 2006, and $50,000-$69,999 and $70,000+ income class have a lower frequency in Nielsen Homescan™ panel data than in the Census 2006 data. The difference indicates that lower income households participated more in the data collection activities than households in the higher income class. When compared over time, it appears that for both Alberta and Ontario, the proportion of households falling into higher income classes (such as more than $70,000) is increasing and the proportion falling into lower income classes (such as less than $20,000) is decreasing. The increase in the percentage of households with higher incomes is observed over the study period, implying that households remaining in the panel over the period 20022007 exhibited increasing incomes.

Household head age

Household head age is recorded as a categorical variable in the Nielsen panel data.. The same mid-point method is used to approximate household head age levels as a continuous measure. Table 3.5 and 3.6 present the household head age classes and mid-point values for the sample data. As appears in both tables 1 and 2, the aggregate frequency of younger household age classes: 18-34 are much lower in Nielsen panel sample data than in the Census 2006 data. However the percentage of older household heads in the classes: 45-54, 55-64, 65+ are higher in the Nielsen Homescan™ panel data. This implies that households with younger heads do not participate in the panel at the same rate as households with middle aged - or older heads do. Both tables also show that the proportion of households with older heads

43

is increasing over the time frame of this study, the households that stayed in the panel tended to have older heads.

Table 3.5 the household head age classes and mid-point value for the sample data of Ontario HH age class

Nielsen Homescan™ panel data 2002-2007 Midpoints

18-34

26

35-44

39.5

45-54

49.5

55-64

59.5

65+

69.5 Total

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Count

42

42

16

16

16

6

138

HH%

4.1%

4.1%

1.5%

1.5%

1.5%

.6%

2.2%

Count

230

230

204

189

189

146

1188

HH%

22.2%

22.2%

19.7%

18.2%

18.2%

14.1%

19.1%

Count

268

268

271

265

265

265

1602

HH%

25.9%

25.9%

26.2%

25.6%

25.6%

25.6%

25.8%

Count

233

233

226

233

233

227

1385

HH%

22.5%

22.5%

21.8%

22.5%

22.5%

21.9%

22.3%

Count

263

263

319

333

333

392

1903

HH%

25.4%

25.4%

30.8%

32.1%

32.1%

37.8%

30.6%

Count

1036

1036

1036

1036

1036

1036

6216

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Ontario

19.0%

19.0%

15.0%

16.0%

16.0%

15.0%

12.0%

11.0%

14.0%

14.0%

Source: Source: Statistics Canada - 2006 Census. Catalogue Number 97-551-XCB2006012.and Nielsen Homescan™ panel data 2002-2007

Table 3.6 Household head age classes and mid-point value for the sample data of Alberta Nielsen Homescan™ panel data 2002-2007

HH age class

Midpoints

18-34

26

35-44

39.5

45-54

49.5

55-64

59.5

65+

69.5 Total

YEAR

2002

2003

2004

2005

2006

2007

Census, 2006 Total

Count

46

46

25

16

16

8

157

HH%

9.1%

9.1%

4.9%

3.1%

3.1%

1.6%

5.2%

Count

108

108

99

87

87

73

562

HH%

21.3%

21.3%

19.5%

17.1%

17.1%

14.4%

18.4%

Count

150

150

155

158

158

151

922

HH%

29.5%

29.5%

30.5%

31.1%

31.1%

29.7%

30.2%

Count

99

99

111

120

120

129

678

HH%

19.5%

19.5%

21.9%

23.6%

23.6%

25.4%

22.2%

Count

105

105

118

127

127

147

729

HH%

20.7%

20.7%

23.2%

25.0%

25.0%

28.9%

23.9%

Count

508

508

508

508

508

508

3048

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Alberta

19.0%

22.0%

15.0%

15.0%

16.0%

16.0%

12.0%

10.0%

14.0%

11.0%

Source: Source: Statistics Canada - 2006 Census. Catalogue Number 97-551-XCB2006012.and Nielsen Homescan™ panel data 2002-2007

44

Household size Household size variable measures the number of members of the household. The Nielsen panel records the household size in five groups. Household size equal to one, means there is only a single member of the household, two means two members in the household, and so forth. Household size equal to five means there are five or more than five members in the household. Table 3.7 and 3.8 show the proportion of households with different household sizes for the sample data and the comparable Canadian Census data for 2006. Table 3.7 Household sizes for the sample data of Ontario and Census 2006 HH size 1 2 3 4 5 or 5+ Total

Nielsen Homescan™ panel data 2002-2007 YEAR

2002

2003

2004

2005

2006

2007

Census, 2006 Total

Count

247

247

255

259

259

261

1528

HH%

23.8%

23.8%

24.6%

25.0%

25.0%

25.2%

24.6%

Count

396

396

410

417

417

452

2488

HH%

38.2%

38.2%

39.6%

40.3%

40.3%

43.6%

40.0%

Count

166

166

156

132

132

133

885

HH%

16.0%

16.0%

15.1%

12.7%

12.7%

12.8%

14.2%

Count

137

137

139

149

149

124

835

HH%

13.2%

13.2%

13.4%

14.4%

14.4%

12.0%

13.4%

Count

90

90

76

79

79

66

480

HH%

8.7%

8.7%

7.3%

7.6%

7.6%

6.4%

7.7%

Count

1036

1036

1036

1036

1036

1036

6216

Canada

Ontario

27.0%

24.0%

34.0%

32.0%

16.0%

17.0%

15.0%

17.0%

9.0%

11.0%

HH% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Table 3.8 Household sizes for the sample data of Alberta and Census 2006 HH size 1 2 3 4 5 or 5+ Total

Nielsen Homescan™ panel data 2002-2007

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Count

133

133

134

137

137

146

820

HH%

26.2%

26.2%

26.4%

27.0%

27.0%

28.7%

26.9%

Count

192

192

210

211

211

213

1229

HH%

37.8%

37.8%

41.3%

41.5%

41.5%

41.9%

40.3%

Count

63

63

64

62

62

55

369

HH%

12.4%

12.4%

12.6%

12.2%

12.2%

10.8%

12.1%

Count

78

78

61

67

67

61

412

HH%

15.4%

15.4%

12.0%

13.2%

13.2%

12.0%

13.5%

Count

42

42

39

31

31

33

218

HH%

8.3%

8.3%

7.7%

6.1%

6.1%

6.5%

7.2%

Count

508

508

508

508

508

508

3048

Canada

Alberta

27.0%

25.0%

34.0%

34.0%

16.0%

16.0%

15.0%

16.0%

9.0%

10.0%

HH% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

45

Household education The household education variable indicates the level of the household head's education achieved. The Nielsen panel records the household education in six levels: no high school graduation; high school graduate; some college or technical school; college or technical school graduate; some university; university graduate. The six categories of education level are reduced to two groups: no high school graduation and otherwise. The education dummy variable (HHEDU1) is then created with a value of one if the household has high school or higher education and zero otherwise. The descriptive statistics for the household education level are listed below in tables 3.9 and 3.10.

Table 3.9 Household head education for the sample data of Ontario and Census 2006 Nielsen Homescan™ panel data 2002-2007

Education levels

Dummy

No high school education

HHEDU1=0

Otherwise

HHEDU1=1 Total

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Count

151

151

145

145

145

131

868

HH%

14.6%

14.6%

14.0%

14.0%

14.0%

12.6%

14.0%

Count

885

885

891

891

891

905

5348

HH%

85.4%

85.4%

86.0%

86.0%

86.0%

87.4%

86.0%

Count

1036

1036

1036

1036

1036

1036

6216

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Ontario

24.0%

22.0%

77.0%

78.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Table 3.10 Household head education for the sample data of Alberta and Census 2006 Nielsen Homescan panel data 2002-2007

Education levels

Dummy

No high school education

HHEDU1=0

Otherwise

HHEDU1=1 Total

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Count

73

73

67

63

63

57

396

HH%

14.4%

14.4%

13.2%

12.4%

12.4%

11.2%

13.0%

Count

435

435

441

445

445

451

2652

HH%

85.6%

85.6%

86.8%

87.6%

87.6%

88.8%

87.0%

Count

508

508

508

508

508

508

3048

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Alberta

24.0%

23.0%

77.0%

76.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Presence of children The Nielsen Homescan™ panel records the presence and the age of children information under nine categories: under 6 only; age 6 to 12 only; age 13 to 17 only; under 6 and age 6 to 12; under 6 and age 13 to 17; age 6 to 12 and age 13 to 17; under 6, age 6 to 12 and age

46

13 to 17 and no children under 18. In the study, we group and create two dummy variables to define the presence of children information. The dummy variable (Child1) is created with a value of one if the household has the presence of children (aged under 18) and zero otherwise. The descriptive statistics for the presence of children are listed below in tables 3.11 and 3.12. In the study sample, it appears that over three quarters of the households do not have children under the age of 18. An increase of the percentage of households without children can be observed in both Ontario and Alberta over the study period, implying the households had older children to start who left the home during the sample.

Table 3.11 Household presence of children for the sample data of Ontario and Census 2006 Nielsen Homescan™ panel data 2002-2007

Children

Dummy

No children

CHILD1=0

Have children

CHILD1=1

Total

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Count

781

781

809

807

807

844

4829

HH%

75.4%

75.4%

78.1%

77.9%

77.9%

81.5%

77.7%

Count

255

255

227

229

229

192

1387

HH%

24.6%

24.6%

21.9%

22.1%

22.1%

18.5%

22.3%

Count

1036

1036

1036

1036

1036

1036

6216

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Ontario

77.0%

75.0%

23.0%

25.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Table 3.12 Household presence of children for the sample data of Alberta and Census 2006 Nielsen Homescan™ panel data 2002-2007

Children

Dummy

No children

CHILD1=0

Have children

CHILD1=1

Total

YEAR

2002

2003

2004

2005

2006

Census, 2006

2007

Total

Count

380

380

398

402

402

409

2371

HH%

74.8%

74.8%

78.3%

79.1%

79.1%

80.5%

77.8%

Count

128

128

110

106

106

99

677

HH%

25.2%

25.2%

21.7%

20.9%

20.9%

19.5%

22.2%

Count

508

508

508

508

508

508

3048

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Alberta

77.0%

67.0%

23.0%

33.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Urban and Rural The location where household reside are recorded by urban and rural variables in the study sample data. Two dummy variables are created to define the urbanization information of household. The dummy variable (Urban) is created with a value of one if the household

47

reside in an urban area and zero otherwise. On the other hand, the dummy variable (Rural) have a value of one if the household resides in the rural area and zero otherwise. Table 3.13 Household urbanization for the sample data of Ontario and Census 2006 Nielsen Homescan™ panel data 2002-2007

Urbanization

Dummy

Rural

Urban=0

Urban

Urban=1

Total

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Count

334

333

322

323

324

324

1960

HH%

32.2%

32.1%

31.1%

31.2%

31.3%

31.3%

31.5%

Count

702

703

714

713

712

712

4256

HH%

67.8%

67.9%

68.9%

68.8%

68.7%

68.7%

68.5%

Count

1036

1036

1036

1036

1036

1036

6216

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Ontario

19.0%

15.0%

81.0%

85.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

Table 3.14 Household urbanization for the sample data of Alberta and Census 2006 Nielsen Homescan™ panel data 2002-2007

Urbanization

Dummy

Rural

Urban=0

Urban

Urban=1

Total

Census, 2006

YEAR

2002

2003

2004

2005

2006

2007

Total

Count

159

160

160

161

160

160

960

HH%

31.3%

31.5%

31.5%

31.7%

31.5%

31.5%

31.5%

Count

349

348

348

347

348

348

2088

HH%

68.7%

68.5%

68.5%

68.3%

68.5%

68.5%

68.5%

Count

508

508

508

508

508

508

3048

HH%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

100.0%

Canada

Alberta

19.0%

17.0%

81.0%

83.0%

Source: Statistics Canada, Census 2006 and Nielsen Homescan™ panel data 2002-2007

In summary, the descriptive statistic results for most of the variables discussed in this section are consistent and relatively close to Canadian Census data for 2006. The sample data are a balanced panel which covers households which stayed in the panel over the study period from 2002 and 2007. So it is observed that households included tended to have older heads and have higher education and income levels than the Canadian Census data. Behavioural models reported in this study will be more representative for the better educated, more urban, higher income and older households than for the 2006 Canadian population as a whole. The next section will provide more data descriptive statistics on household meat and store expenditures.

48

Canadian Meat Demand Analysis By Level of Processing Introduction The first objective of the study is to understand how Canadian households make purchase decisions around fresh, semi-processed and fully processed meat products for four meat type categories: beef, pork, poultry and others (fish, lamb, etc.). The analysis aims to quantify the impact of price, advertising, demographic and regional characteristic differences on meat consumption behaviour, and differences in the behaviour across meat types. In this section, the data setup for the analysis followed by the data descriptive statistics are provided. Then the explanation of model specification and econometric methods are presented. The model results and summary are finally provided in the section.

Data setup and descriptive statistics Nielsen Homescan™ data is used in this analysis, the data contains all individual panellist's meat purchase information, by size, by product processed form, by brand, and by meat type. The panel data also includes the household demographic data, including age of household head, presence of children, income, education, urban and rural residence information,as described above. The meat demand analysis focuses on the meat products purchased by household in the provinces of Ontario and Alberta over the time period 2002 to 2007. In total, 1036 households in Ontario and 508 households in Alberta are observed in the balanced panels. Value added meat products are grouped into the twelve meat categories according to their "PRTYP" (meat processed type table) and "PRFRM" (meat processed from table) information recorded by Nielsen Homescan™ data (as discussed in the first section). Twelve choice alternatives in this analysis were identified: (1) fresh pork, beef, poultry and other meats; (2) semi-processed pork, beef, poultry and other meats; (3) fully processed pork, beef, poultry and other meats. These product purchases across a year were aggregated into annual expenditures, on the twelve products, for each household.

49

1 .Total expenditure on value added meat

Aggregate annual expenditures on the meat products for the period 2002 to 2007 are described in this section. In Table 4.1 and 4.2 below, aggregate market (expenditure) shares for each of the twelve meat categories in Ontario and Alberta are reported. Table 4.1. Market share for each meat category in Ontario. 2002

2003

2004

2005

2006

2007

Fresh pork

13%

13%

11%

12%

12%

11%

Fresh beef

32%

30%

30%

28%

29%

29%

Fresh poultry

24%

24%

24%

24%

25%

25%

Fresh others

3%

5%

6%

5%

5%

5%

Semi processed pork

4%

5%

5%

6%

5%

6%

Semi processed beef

1%

1%

1%

1%

1%

1%

Semi processed poultry

1%

2%

2%

2%

2%

2%

Semi processed others

3%

2%

2%

2%

2%

2%

Fully processed pork

3%

3%

4%

4%

4%

4%

Fully processed beef

1%

1%

1%

1%

1%

0%

Fully processed poultry

8%

8%

7%

6%

6%

6%

Twelve meat categories

Fully processed others Total

7%

7%

7%

8%

9%

9%

100%

100%

100%

100%

100%

100%

By value added levels Fresh meat total

72%

73%

71%

70%

70%

70%

Semi processed meat total

9%

10%

10%

11%

10%

11%

Fully processed meat total

18%

18%

19%

19%

20%

19%

100%

100%

100%

100%

100%

100%

Pork total

20%

20%

20%

22%

21%

21%

Beef total

33%

31%

31%

30%

31%

31%

Poultry total

34%

34%

34%

33%

33%

33%

Total By meat types

Others total Total

13%

14%

15%

15%

16%

15%

100%

100%

100%

100%

100%

100%

Source: Nielsen Homescan™ Panel, Ontario 2002 to 2007

50

Table 4.2. Market share for each meat category in Alberta. Data

2002

2003

2004

2005

2006

2007

Fresh pork

16%

17%

16%

17%

15%

14%

Fresh beef

37%

36%

30%

30%

33%

33%

Fresh poultry

24%

23%

24%

24%

25%

25%

Twelve meat categories

Fresh others

3%

4%

5%

4%

3%

4%

Semi processed pork

1%

1%

3%

3%

2%

2%

Semi processed beef

1%

1%

1%

1%

1%

1%

Semi processed poultry

1%

1%

1%

1%

2%

2%

Semi processed others

2%

1%

2%

2%

1%

1%

Fully processed pork

4%

4%

5%

5%

5%

5%

Fully processed beef

1%

1%

1%

1%

1%

1%

Fully processed poultry

6%

5%

6%

6%

5%

6%

Fully processed others

5%

5%

8%

7%

7%

8%

Total

100%

100%

100%

100%

100%

100%

By value added levels Fresh meat total

80%

80%

75%

75%

76%

75%

Semi processed meat total

5%

5%

6%

7%

6%

6%

Fully processed meat total

15%

15%

19%

19%

19%

19%

Total

100%

100%

100%

100%

100%

100%

Pork total

21%

23%

24%

25%

22%

21%

Beef total

38%

38%

32%

32%

35%

34%

Poultry total

31%

29%

30%

30%

32%

32%

Others total

10%

11%

14%

13%

12%

12%

Total

100%

100%

100%

100%

100%

100%

By meat types

Source: Nielsen Homescan™ Panel, Alberta 2002 to 2007

Error! Reference source not found.In Tables 4.3 and 4.4 report the average spending per household per year for each meat category from 2002 to 2007 is presented. Average annual household total meat expenditure increased from $336 to $398 in Ontario and $382 to $406 in Alberta.

51

Table 4.3. Annual average expenditure., dollars Data

2002

2003

2004

2005

2006

2007

Twelve meat categories Fresh pork

44.5

48.6

45.2

48.8

46.8

45.8

Fresh beef

107.0

110.3

121.0

115.8

114.6

115.9

Fresh poultry

80.9

89.7

97.7

99.7

97.8

99.3

Fresh others

10.8

18.0

22.5

22.3

18.8

19.4

Semi processed pork

15.0

16.9

21.6

23.8

20.4

22.2

Semi processed beef

1.8

2.4

3.2

4.6

4.7

5.7

Semi processed poultry

4.8

6.6

8.8

8.2

8.2

9.6

Semi processed others

9.5

9.2

8.3

6.9

6.3

6.5

Fully processed pork

8.8

9.5

14.8

16.1

17.0

15.4

Fully processed beef

2.4

2.2

2.6

2.4

2.2

2.0

Fully processed poultry

28.2

28.7

29.5

26.2

23.8

22.6

Fully processed others

22.3

25.2

28.6

33.4

36.6

34.5

336.1

367.3

403.7

408.1

397.4

398.9

243.2

266.6

286.4

286.6

278.1

280.4

Semi processed meat total

31.2

35.0

41.9

43.5

39.7

44.0

Fully processed meat total

61.7

65.7

75.4

78.0

79.6

74.5

Pork total

68.4

75.0

81.6

88.7

84.3

83.5

Beef total

111.1

114.9

126.8

122.7

121.5

123.6

Poultry total

114.0

125.1

136.0

134.1

129.8

131.5

Others total

42.6

52.3

59.4

62.6

61.8

60.4

Total By value added levels Fresh meat total

By meat types

Source: Nielsen Homescan™ Panel, Ontario 2002 to 2007

For 2007, on average, household total meat expenditure averaged $398 in Ontario. Fresh meat consumption is the large market share in meat consumption, in which fresh beef has the single largest share.

52

Table 2.4 Alberta Annual average expenditure, dollars. Data

2002

2003

2004

2005

2006

2007

Fresh pork

62.4

72.7

69.7

72.4

59.6

57.3

Fresh beef

140.7

152.4

131.1

131.7

134.4

132.3

Fresh poultry

90.3

95.8

102.6

102.8

103.0

99.9

Fresh others

Twelve meat categories

12.1

16.5

20.2

18.7

13.9

15.1

Semi processed pork

4.2

6.2

11.3

14.4

8.2

7.4

Semi processed beef

3.6

4.2

4.6

4.2

3.4

3.6

Semi processed poultry

2.9

2.9

3.1

4.0

6.2

7.4

Semi processed others

7.2

5.9

6.8

6.7

5.5

4.5

Fully processed pork

13.6

17.0

22.7

23.4

21.4

21.0

Fully processed beef

2.5

2.4

2.2

3.1

3.8

3.4

24.7

23.1

24.3

24.3

21.7

23.3

Fully processed poultry Fully processed others Total

18.4

22.2

33.0

31.2

29.1

30.9

382.7

421.3

431.7

436.8

410.3

406.2

By value added levels Fresh meat total

305.5

337.5

323.7

325.6

310.9

304.5

Semi processed meat total

18.0

19.2

25.8

29.3

23.3

23.0

Fully processed meat total

59.2

64.6

82.2

82.0

76.1

78.6

Pork total

80.1

95.9

103.7

110.2

89.2

85.7

Beef total

146.8

159.0

137.9

139.0

141.6

139.3

Poultry total

118.0

121.8

130.1

131.1

131.0

130.6

Others total

37.7

44.7

60.1

56.6

48.5

50.5

By meat types

Source: Nielsen Homescan™ Panel, Alberta 2002 to 2007

For 2007, on average, household total meat expenditure averaged $406 in Alberta. Fresh meat consumption is also the large market share category in meat consumption.

53

In Tables 4.5 and 4.6, the coefficients of variation for expenditure on each of the meat categories are reported. The coefficient of variation is a normalized measure of the dispersion of sample data. It is calculated as the ratio of the standard deviation to the mean. The coefficient of variation can provide a comparison across market segments when the means across segments vary. The higher the level of the coefficient of variation, the greater is the degree of variability in the data.

Table 4.5 Coefficients of variation of household purchases in Ontario.

Fresh pork

2002 1.3

2003 1.4

2004 1.3

2005 1.3

2006 1.3

2007 1.2

Fresh beef

1.2

1.2

1.2

1.2

1.3

1.3

Fresh poultry

1.1

1.0

1.0

1.0

1.2

1.1

Fresh others

2.5

1.9

1.8

2.0

2.5

2.5

Semi processed pork

2.0

1.8

1.9

1.6

1.6

1.4

Semi processed beef

3.6

3.2

2.6

2.5

2.8

2.6

Semi processed poultry

2.5

2.6

2.1

2.2

2.2

2.0

Semi processed others

1.7

1.8

1.9

1.9

2.1

2.3

Fully processed pork

1.8

1.8

1.5

1.5

1.5

1.5

Fully processed beef

3.7

3.8

3.5

3.3

3.5

3.3

Fully processed poultry

1.6

1.8

1.7

1.9

1.8

2.4

Fully processed others

1.7

1.6

1.6

1.3

1.4

1.5

Pork total

1.1

1.2

1.1

1.1

1.0

1.0

Beef total

1.2

1.2

1.2

1.2

1.2

1.3

Poultry total

0.9

0.9

0.9

0.9

1.0

1.0

Others total

1.3

1.2

1.2

1.2

1.3

1.3

Fresh meat total

1.0

0.9

0.9

0.9

1.0

1.0

Semi processed meat total

1.4

1.3

1.3

1.2

1.2

1.1

Fully processed meat total Total

1.2

1.2

1.1

1.1

1.0

1.2

0.9

0.8

0.8

0.8

0.9

0.9

Source: Nielsen Homescan™ Panel, Ontario 2002 to 2007

54

Table 4.6. Coefficients of variation of household purchases in Alberta.

Fresh pork

2002 1.2

2003 1.2

2004 1.3

2005 1.3

2006 1.3

2007 1.3

Fresh beef

1.1

1.1

1.1

1.1

1.1

1.2

Fresh poultry

1.0

1.0

1.0

1.0

1.0

1.0

Fresh others

2.3

1.8

2.1

1.7

1.8

1.7

Semi processed pork

3.0

2.4

2.1

2.6

2.0

2.3

Semi processed beef

2.6

2.4

2.3

2.9

3.2

3.1

Semi processed poultry

2.8

2.6

2.9

2.5

2.2

2.3

Semi processed others

1.7

1.8

2.0

2.2

2.3

2.4

Fully processed pork

1.5

1.3

1.2

1.2

1.4

1.4

Fully processed beef

3.3

3.0

3.2

3.6

3.0

2.9

Fully processed poultry

1.7

1.7

1.8

1.9

1.8

1.9

Fully processed others

1.5

1.4

1.3

1.3

1.4

1.3

Pork total

1.1

1.1

1.1

1.1

1.1

1.1

Beef total

1.1

1.1

1.1

1.1

1.1

1.1

Poultry total

0.9

0.9

0.9

0.9

0.9

0.9

Others total

1.2

1.2

1.2

1.2

1.1

1.1

Fresh meat total

0.9

0.9

0.9

0.9

0.9

0.9

Semi processed meat total

1.3

1.4

1.4

1.6

1.4

1.6

Fully processed meat total Total

1.0

0.9

0.9

0.9

0.9

0.9

0.8

0.8

0.8

0.8

0.8

0.8

Source: Nielsen Homescan™ Panel, Alberta 2002 to 2007

The coefficients of variation for most of the meat categories is greater than one in Ontario and Alberta, except for the poultry total and fresh meat total categories. It means the standard deviation is greater than the mean in the aforementioned categories and using the mean per household expenditure on each meat category to represent the population could become problematic, as spending patterns vary widely within the population. Hence, a segmentation approach (segment consumers into groups) is applied in purchasing patterns among households across the years in next section.

55

2. Household food expenditure patterns, levels

Consumers usually have heterogeneous preferences, so it is useful to segment consumers into groups with similar needs and background. Segmentation variables used in the section are the household demographic variables. The value added meat expenditure patterns are shown in the following tables. In Tables 4.7-4.14, the households are grouped into seven categories based on expenditure levels on all meat categories. The seven expenditure levels are 0 dollar (no consumption), less than 25 dollars, 25 to 50 dollars, 51 to 100 dollars, 101 to 300 dollars, 301 to 500 dollars, and more than 500 dollars. The aggregate data for 2002 to 2007(six years) are presented in tables below. Table 4.7 Meat expenditure by fresh meat category in Ontario from 2002-2007

56

Table 4.8 Meat expenditure by fresh meat category in Alberta from 20022007

Table 4.9 Meat expenditure by semi processed meat category in Ontario from 2002-2007

57

Table 4.10 Meat expenditure by semi processed meat category in Alberta from 2002-2007

Table 4.11 Meat expenditure by fully- processed meat category in Ontario from 2002-2007

58

Table 4.12 Meat expenditure by fully- processed meat category in Ontario from 2002-2007

Table 4.13 Meat expenditure by all value added meat categories in Ontario from 2002-2007

59

Table 4.14 Meat expenditure by all value added meat categories in Alberta from 2002-2007

Model specification and econometric method

In this study a balanced panel of sample data in Ontario and Alberta are analyzed. Not all household have positive expenditures on all twelve meat categories. The positive value added meat expenditure shows that household have already made the decision to purchase and are able to choose one or more products from the twelve value added meat categories. We assume each household faces a two-step hierarchy in their decision making: the household first makes the decision of what types of value added meat to purchase (participation step), then they will decide how much expenditure they will spend once they choose the meat product types to purchase (expenditure step). Therefore a two-step estimation following the Heien and Wessels (1990) Working-Leser demand system procedure is applied in the value added meat demand analysis. In the first step, a probit regression is computed that determines the probability that a given household

60

will purchase a particular meat type. The probability of purchase is then used as an instrument in the second-stage estimation of the Working-Leser demand system. 1.Participation decision by value added meat products

The first stage of the demand system is modeled as a participation choice problem: the dependent variable is represented by a binary choice variable yiht  1 if household h decides to purchase value added meat i at period t and is yiht  0 if the household does not consume the meat product of i at period t. Then E ( yiht )  1* piht  0*(1  piht )  piht and this is usually modeled as a function of household demographic variables and total meat expenditure. The inverse mills ratio is actually the expectation of the structural residual, where the model is given by: (TSP 5.0 reference manual): yi  X i    i

 i ~ N(0,1)

Di  1( yi  0)

And the inverse mills ratio is the value of the following two expressions, depending on whether D=0 or 1:

E ( D  1) 

Norm( Xb) Norm( Xb)   Dlcnorm( Xb) 1  Cnorm( Xb) Cnorm( Xb)

E ( D  0) 

Norm( Xb)   Dlcnorm( Xb) Cnorm( Xb)

where Norm is the normal density, Cnorm is the cumulative normal and Dlcnorm is the derivative of the log cumulative normal with respect to its argument. So the likelihood of household participation decision by value added meat type (Pr[ yiht  1]) for a random effects panel can be expressed as:   Pr[ yiht  1]  Pr[ X iht   aiht   ih  0]   ( X iht )

61

and the likelihood of households that do not purchase a particular value added meat is:   Pr[ yiht  0]  Pr[ X iht   aiht   ih  0]  1   ( X iht )

where  X iht   0  1 *MTotal  2 * hage  3 * hhedu  4 * urban  5 * hhsize  6 *T

2.Expenditure decision by value added meat products

The second step is the estimation of the expenditure share equations of the Working-Leser demand system via seemingly unrelated regression (SUR) of the expenditure share that household h spends on value added meat i in time period t. In the Working-Leser model, each expenditure share of the meat product is a linear function of the log of prices and of the total expenditure on all the meat items. The general form of the second stage equations of Working-Leser food demand function can be expressed as:

i  a0  a1 *log( Mtotal )   j a2 *ln( p j )  a3 *log[ M i (1)]  a4 * Mills  a5 * AD  a6 * hhinc  a7 * KID  a8 * chains  a9 * hhsize  a10 * T   it

where (i,j) represents the twelve value added meat products. wi is the expenditure share of meat product i among the twelve value added meat products; pj is the price of meat j; Mtotal is the total expenditure of all meat products M(-1) is the lagged meat i expenditure which may lead to a habit formation, where past consumption decisions serve as predictors of future purchase decisions

62

AD is the advertising information of meat i and other meat HHINC is the household income Kid is the presence of children in the household. Chains represents the number of grocery store chains where household purchased the twelve meat products. T is the time trend variable.

Model testing and empirical results TSP International 5.0 was the econometric software used for the estimation of parameters in this study. Likelihood ratio tests (LRT) were applied to select the best fitting model among a number of models. The definitions of variables used for the analysis are listed in Table 4.20 below.

Table 4.20 Definition and sample statistics of variables used for value added meat choice analysis Variables

Definitions

First stage: binary dependent variables B11 1 if choose fresh pork, 0 otherwise B12 1 if choose fresh beef, 0 otherwise B13 1 if choose fresh poultry, 0 otherwise B14 1 if choose fresh others, 0 otherwise B21 1 if choose semi-processed pork, 0 otherwise B22 1 if choose semi-processed beef, 0 otherwise B23 1 if choose semi-processed poultry, 0 otherwise B24 1 if choose semi-processed others, 0 otherwise B31 1 if choose fully-processed pork, 0 otherwise B32 1 if choose fully-processed beef, 0 otherwise B33 1 if choose fully-processed poultry, 0 otherwise B34 1 if choose fully-processed others, 0 otherwise Second stage: expenditure share dependent variables SH11 share of fresh pork expenditure SH12 share of fresh beef expenditure

Ontario Mean SD 0.78 0.91 0.93 0.53 0.60 0.24 0.35 0.45 0.57 0.17 0.60 0.72

0.42 0.29 0.26 0.50 0.49 0.43 0.48 0.50 0.49 0.38 0.49 0.45

0.11 0.11 0.27 0.19

Alberta Mean SD 0.81 0.89 0.93 0.51 0.41 0.27 0.23 0.41 0.65 0.23 0.57 0.70

0.39 0.31 0.26 0.50 0.49 0.44 0.42 0.49 0.48 0.42 0.50 0.46

0.14 0.13 0.28 0.19

63

SH13 share of fresh poultry expenditure SH14 share of fresh others expenditure SH21 share of semi-processed pork expenditure SH22 share of semi-processed beef expenditure SH23 share of semi-processed poultry expenditure SH24 share of semi-processed others expenditure SH31 share of fully-processed pork expenditure SH32 share of fully-processed beef expenditure SH33 share of fully-processed poultry expenditure SH34 share of fully-processed others expenditure Logged form of meat expenditure LM11 logged fresh pork expenditure LM12 logged fresh beef expenditure LM13 logged fresh poultry expenditure LM14 logged fresh others expenditure LM21 logged semi-processed pork expenditure LM22 logged semi-processed beef expenditure LM23 logged semi-processed poultry expenditure LM24 logged semi-processed others expenditure LM31 logged fully-processed pork expenditure LM32 logged fully-processed beef expenditure LM33 logged fully-processed poultry expenditure LM34 logged fully-processed others expenditure Logged form of meat price LP11 logged fresh pork price 1.96 LP12 logged fresh beef price 2.11 LP13 logged fresh poultry price 1.85 LP14 logged fresh others price 1.70 LP21 logged semi-processed pork price 1.44 LP22 logged semi-processed beef price 1.73 LP23 logged semi-processed poultry price 2.55 LP24 logged semi-processed others price 2.57 LP31 logged fully-processed pork price 1.69 LP32 logged fully-processed beef price 1.06 LP33 logged fully-processed poultry price 1.36 LP34 logged fully-processed others price 1.40 LP11oth logged price except for fresh pork 1.85 LP12oth logged price except for fresh beef 1.74 LP13oth logged price except for fresh poultry 1.87 LP14oth logged price except for fresh others 1.88 LP21oth logged price except for semi- pork 1.88 LP22oth logged price except for semi- beef 1.87

0.25 0.05 0.05 0.01 0.03 0.02 0.04 0.01 0.08 0.10

0.18 0.10 0.07 0.03 0.07 0.05 0.07 0.02 0.12 0.14

0.25 0.04 0.02 0.01 0.01 0.02 0.06 0.01 0.07 0.09

0.17 0.08 0.05 0.02 0.04 0.05 0.09 0.02 0.12 0.14

1.21 1.67 1.66 0.68 0.78 0.24 0.40 0.47 0.69 0.16 0.83 1.01

0.77 0.73 0.66 0.73 0.72 0.47 0.60 0.59 0.67 0.39 0.78 0.74

1.36 1.73 1.71 0.65 0.45 0.27 0.26 0.41 0.85 0.20 0.79 0.98

0.79 0.79 0.64 0.72 0.61 0.48 0.51 0.55 0.71 0.43 0.77 0.73

2.00 2.08 1.90 1.62 1.74 2.14 2.58 2.61 1.77 1.16 1.45 1.79 1.91 1.84 1.93 1.94 1.93 1.92

0.10 0.05 0.02 0.08 0.22 0.13 0.14 0.08 0.06 0.13 0.05 0.04 0.03 0.02 0.03 0.02 0.03 0.02

0.09 0.03 0.03 0.07 0.05 0.05 0.05 0.07 0.10 0.12 0.01 0.05 0.03 0.03 0.04 0.04 0.03 0.03

64

LP23oth logged price except for semi- poultry LP24oth logged price except for semi- others LP31oth logged price except for fully- pork LP32oth logged price except for fully- beef LP33oth logged price except for fully- poultry LP34oth logged price except for fully- others HH demographic and purchase information MTotal Total expenditure on all types of meat LTE logged total exp on all types of meat HHINC Annual HH income(C$, midpoint) HAGE Household head age(midpoint) KID1 1 if HH with children, 0 otherwise KID0 1 if HH without children , 0 otherwise HHEDU0 1 if no high school edu, 0 otherwise HHEDU1 1 if higher edu, 0 otherwise URBAN 1 if in urban area, 0 otherwise RURAL 1 if in rural area, 0 otherwise HHSIZE Number of members in household T year 1-6 Chains Number of grocery chains HH visited

Variables

Definitions

Advertising expenditure by meat types AD11 fresh pork AD AD12 fresh beef AD AD13 fresh poultry AD AD14 fresh others AD AD21 semi-processed pork AD AD22 semi-processed beef AD AD23 semi-processed poultry AD AD24 semi-processed others AD AD31 fully-processed pork AD AD32 fully-processed beef AD AD33 fully-processed poultry AD AD34 fully-processed others AD AD11oth Total AD except for fresh pork AD12oth Total AD except for fresh beef AD13oth Total AD except for fresh poultry AD14oth Total AD except for fresh others

1.85 1.87 1.95 1.88 1.90 1.90

0.04 0.03 0.03 0.03 0.03 0.03

1.88 1.92 1.99 1.93 1.94 1.94

0.05 0.02 0.02 0.02 0.02 0.03

385.3 325.4 414.8 337.7 2.42 0.42 2.46 0.41 52386 22189 51932 21909 55.42 11.88 53.45 12.22 0.22 0.42 0.22 0.42 0.78 0.42 0.78 0.42 0.14 0.35 0.13 0.34 0.86 0.35 0.87 0.34 0.68 0.46 0.69 0.46 0.32 0.46 0.31 0.46 2.40 1.21 2.34 1.21 3.50 1.71 3.50 1.71 2.60 0.89 2.84 1.17

Ontario & Alberta Mean SD 1726248 1288502 8250415 375781 470236 0 39451 212510 3591602 71519 1320833 1178453 1.67993D+07 1.72370D+07 1.02751D+07 1.81498D+07

960208 580839 1275109 354160 427841 0 54549 234255 1465079 98572 1878608 352855 1631576 1735378 1575829 1697066

65

AD21oth AD22oth AD23oth AD24oth AD31oth AD32oth AD33oth AD34oth

Total AD Total AD Total AD Total AD Total AD Total AD Total AD Total AD

except for semi-processed pork except for semi-processed beef except for semi-processed poultry except for semi-processed others except for fully-processed pork except for fully-processed beef except for fully-processed poultry except for fully-processed others

1.80553D+07 1.85256D+07 1.84861D+07 1.83130D+07 1.49339D+07 1.84540D+07 1.72047D+07 1.73471D+07

2175070 1828122 1843340 1802960 1886875 1829559 2383818 2033027

Note: The source of these data is Nielsen Homescan™ Panel, Ontario & Alberta, 2002-2007 and Nielsen Media Measurement.

First stage: household participation decision results by types of value added meat

Tables 4.22 and 4.23 report the probability results for the Probit model for Ontario and Alberta (participation step) TABLE 4.22 First-Step Probit Estimates for Ontario Variables Constant MTOTAL HAGE HHEDU0 URBAN HHSIZE T

Constant MTOTAL HAGE HHEDU0 URBAN HHSIZE T

Constant

fresh pork Coeff. t -0.618 *** -5.023 0.003 *** 26.384 0.014 *** 8.092 0.113 * 1.836 -0.312 *** -7.092 -0.013 -0.735 -0.007 -0.604 fresh beef Coeff. t 0.312 ** 2.046 0.004 *** 20.357 0.007 *** 3.297 0.012 0.152 0.097 * 1.813 -0.137 *** -6.165 -0.009 -0.637 fresh poultry Coeff. t 0.133 0.793

semi- pork Coeff. t -0.990 *** -8.943 0.002 *** 25.270 0.007 *** 4.436 0.092 * 1.820 -0.108 ** -2.937 0.029 * 1.808 0.062 *** 6.206 semi- beef Coeff. t -1.439 *** -12.058 0.001 *** 16.890 0.002 1.273 -0.027 -0.517 -0.090 ** -2.340 -0.020 -1.163 0.093 *** 8.695 semi- poultry Coeff. t -0.655 *** -6.023

fully- pork Coeff. t -1.304 *** -11.880 0.001 *** 20.834 0.012 *** 7.784 0.155 ** 3.122 -0.172 *** -4.732 0.079 *** 4.986 0.067 *** 6.853 fully- beef Coeff. t -1.523 *** -11.807 0.001 *** 14.052 0.003 * 1.772 -0.115 ** -1.997 -0.106 ** -2.577 0.065 *** 3.598 -0.006 -0.524 fully- poultry Coeff. t 0.797 *** 7.325

66

MTOTAL HAGE HHEDU0 URBAN HHSIZE T

0.005 *** 19.404 0.001 *** 13.215 0.008 *** 3.163 -0.005 *** -3.529 0.085 0.969 0.184 *** 3.803 0.121 ** 2.042 -0.153 *** -4.292 -0.042 * -1.723 0.048 ** 3.072 -0.041 ** -2.557 0.073 *** 7.350 fresh others semi- others Coeff. t Coeff. t Constant -1.231 *** -11.476 -0.378 *** -3.537 MTOTAL 0.001 *** 17.031 0.001 *** 17.728 HAGE 0.012 *** 7.941 0.002 1.276 HHEDU0 -0.062 -1.293 0.216 *** 4.469 URBAN 0.123 *** 3.497 -0.022 -0.611 HHSIZE 0.063 *** 4.050 0.064 *** 4.158 T 0.011 1.096 -0.119 *** -12.137 Note:***, **, * = significance at 1%, 5%, 10%

0.001 *** 12.612 -0.020 *** -12.780 0.144 ** 2.907 -0.060 * -1.666 0.179 *** 11.059 -0.034 *** -3.452 fully- others Coeff. t -0.428 *** -3.813 0.001 *** 13.190 0.000 -0.272 -0.097 ** -1.923 -0.010 -0.265 0.183 *** 10.666 0.088 *** 8.576

TABLE 4.23 First-Step Probit Estimates for Alberta Variables Constant MTOTAL HAGE HHEDU0 URBAN HHSIZE T

Constant MTOTAL HAGE HHEDU0 URBAN HHSIZE T

Constant

fresh pork Coeff. t -0.633 *** -3.211 0.004 *** 19.795 0.012 *** 4.431 0.015 0.179 0.034 0.377 0.000 -1.444 -0.032 * -1.840 fresh beef Coeff. t -0.189 -0.825 0.005 *** 16.550 0.006 ** 2.051 -0.102 -0.956 0.260 * 2.557 0.000 * -1.754 -0.045 ** -2.161 fresh poultry Coeff. t 0.090 0.359

semi- pork Coeff. t -1.172 *** -7.000 0.001 *** 15.599 0.004 * 1.719 -0.061 -0.919 -0.132 * -1.827 0.000 1.524 0.064 *** 4.534 semi- beef Coeff. t -1.298 *** -7.294 0.001 *** 12.552 0.006 * 2.451 0.223 *** 3.284 0.032 0.408 0.000 0.132 -0.045 ** -3.008 semi- poultry Coeff. t -1.480 *** -8.064

fully- pork Coeff. t -0.902 *** -5.338 0.002 *** 15.747 0.012 *** 5.198 0.126 * 1.840 -0.209 ** -2.728 0.000 1.459 0.042 ** 2.887 fully- beef Coeff. t -1.528 *** -8.347 0.001 *** 9.903 0.006 ** 2.386 0.175 ** 2.483 -0.054 -0.705 0.000 -1.070 0.056 *** 3.661 fully- poultry Coeff. t 0.407 ** 2.521

67

0.004 *** 13.637 0.001 *** 10.067 0.005 1.482 -0.006 * -2.311 -0.087 -0.743 0.018 0.259 0.181 1.599 0.061 0.745 0.000 -1.123 0.000 *** 4.101 0.007 0.285 0.105 *** 6.819 fresh others semi- others Coeff. t Coeff. t -0.807 *** -4.978 -0.533 *** -3.255 Constant 0.001 *** 12.836 0.001 *** 13.170 MTOTAL 0.003 1.493 0.002 0.702 HAGE -0.051 -0.786 0.086 1.330 HHEDU0 0.005 0.075 -0.007 -0.091 URBAN 0.000 * 1.881 0.000 0.494 HHSIZE 0.039 ** 2.862 -0.070 *** -5.029 T Note:***, **, * = significance at 1%, 5%, 10% MTOTAL HAGE HHEDU0 URBAN HHSIZE T

0.001 *** 9.797 -0.009 *** -3.982 0.510 *** 7.573 -0.180 ** -2.521 0.000 0.433 -0.010 -0.750 fully- others Coeff. t -0.035 -0.207 0.001 *** 9.409 -0.002 -0.785 0.262 *** 3.710 -0.157 ** -2.088 0.000 ** 2.703 0.076 *** 5.202

TABLE 4.24 Second-Step Working-Leser Model Estimates for Ontario

Parameter Constant LTE Mills LM(-1) LP AD ADOTH HHINC KID1 CHAINS LPOTH HHSIZE T Parameter Constant LTE Mills LM(-1)

fresh pork Coeff. 0.121333 0.049849 *** 8.17E-04 4.00E-03 *** -0.14105 ** -3.86E-09 5.83E-10 -4.02E-07 *** -0.01734 *** 3.00E-03 * 0.108861 -3.26E-03 * -0.01097 ** fresh beef Coeff. 1.40566 ** 0.123538 *** -0.03832 *** 5.99E-04

t 0.442735 11.4302 0.762153 3.57762 -1.95817 -1.3331 0.458554 -5.69493 -4.06616 1.87378 1.21511 -1.85936 -2.18149 t 1.96388 15.691 -18.1595 0.434862

semi- pork Coeff. -0.72437 ** 9.35E-03 ** 0.012289 *** 4.42E-03 *** 0.079179 ** 7.43E-09 1.16E-09 -8.66E-08 * -4.42E-03 -2.09E-03 ** 0.31361 ** 2.01E-03 * 7.01E-03 *** semi- beef Coeff. -0.09667 5.33E-03 *** 6.04E-03 *** -9.43E-03 ***

t -2.82433 2.82861 16.7339 4.95776 2.28146 1.35328 1.24433 -1.94837 -1.58415 -2.13245 2.76292 1.7434 4.36105 t -1.56232 4.10106 7.86727 -11.2116

fully- pork Coeff. 0.508032 ** -0.01715 *** 0.012534 *** 4.23E-03 *** -0.01542 4.16E-09 * -1.19E-09 ** -2.71E-08 -0.01013 *** -1.23E-03 -0.21814 * 2.91E-03 ** 6.85E-03 fully- beef Coeff. -1.90258 *** 3.92E-03 ** 1.72E-03 * -0.01575 ***

t 2.68751 -5.00643 17.3432 4.24575 -0.33732 1.7978 -2.06606 -0.62719 -4.72189 -1.30823 -1.76436 3.0194 1.53851 t -3.20577 2.96444 1.74297 -13.274

68

LP AD ADOTH HHINC KID1 CHAINS LPOTH HHSIZE T Parameter Constant LTE Mills LM(-1) LP AD ADOTH HHINC KID1 CHAINS LPOTH HHSIZE T Parameter Constant LTE Mills LM(-1) LP AD ADOTH HHINC KID1 CHAINS LPOTH HHSIZE T

-0.71 ** -2.15402 -1.79E-08 ** -3.00707 1.22E-09 0.760741 -5.37E-07 *** -4.44169 -1.11E-03 -0.14872 -5.25E-03 ** -1.95763 0.114787 0.579503 -0.01971 *** -6.5135 -0.01464 ** -2.66782 fresh poultry Coeff. t -1.91098 ** -2.92238 7.38E-03 0.888812 -0.03808 *** -17.176 -2.42E-03 * -1.65724 1.15074 ** 2.76044 -7.70E-09 * -1.66226 -1.56E-08 ** -3.0206 6.28E-07 *** 5.33943 1.86E-03 0.253736 2.19E-03 0.828031 0.101108 0.363617 -9.30E-03 ** -2.97854 7.38E-03 1.46607 fresh others Coeff. t 1.65917 *** 4.4946 -0.01939 *** -5.21889 0.01154 *** 15.1608 5.50E-03 *** 5.43072 0.222764 ** 2.28927 3.03E-09 0.477318 1.46E-08 *** 3.33407 8.08E-08 1.35996 -7.97E-03 ** -1.96611 2.99E-03 ** 2.18471 -1.14324 *** -3.85992 2.47E-03 1.57023 -0.02075 *** -3.63169

-0.07583 -1.27315 7.51E-08 1.52711 -3.33E-10 * -1.72975 -7.25E-08 *** -4.35607 -1.31E-03 -1.10675 7.13E-04 * 1.75875 0.12228 1.54442 -7.95E-05 -0.18539 1.80E-03 *** 3.51926 semi- poultry Coeff. t 0.35164 1.08674 -0.02128 *** -5.78217 0.010847 *** 12.1888 1.35E-04 0.108294 -0.02629 -0.85571 -1.19E-08 -0.29601 5.10E-10 0.768641 6.73E-08 1.44785 1.44E-03 0.604245 -5.58E-04 -0.53583 -0.12082 -0.83782 1.25E-03 1.1916 4.38E-04 0.165583 semi- others Coeff. t 0.077665 0.531027 -0.01044 *** -3.85522 0.011703 *** 15.9965 6.94E-05 0.070209 -0.02265 -1.29608 5.04E-10 0.171767 -2.51E-10 -0.70288 -1.46E-07 *** -4.05897 -2.65E-03 -1.38873 1.25E-04 0.161426 0.019497 0.336287 4.18E-03 *** 4.67067 -1.30E-03 -1.53459

-0.12269 *** -3.27899 -5.04E-08 ** -2.63309 9.39E-10 ** 2.35481 -2.39E-08 -1.46191 -1.41E-03 -1.32692 -3.97E-04 -0.96735 1.02152 *** 3.21509 1.51E-03 *** 3.66988 0.026509 *** 3.18899 fully- poultry Coeff. t 0.753191 1.51246 -0.04798 *** -8.78903 5.98E-03 *** 7.35398 5.31E-03 *** 5.31726 -0.13091 -0.70695 2.75E-09 0.972586 -5.74E-10 -0.40544 3.03E-07 *** 3.93839 0.03031 *** 5.54961 1.65E-03 0.919288 -0.20235 -0.72054 9.86E-03 *** 4.49191 -0.01339 * -1.77896 fully- others Coeff. t 0.757904 ** 2.2859 -0.08313 *** -12.2641 2.93E-03 *** 3.69196 3.34E-03 *** 3.32149 -0.20785 -1.20903 -1.18E-09 -0.14289 -1.06E-09 -0.86403 2.16E-07 ** 2.53011 0.012736 ** 2.73983 -1.15E-03 -0.60647 -0.11711 -0.60907 8.16E-03 *** 3.60735 0.011058 1.48188

TABLE 4.25 Second-Step Working-Leser Model Estimates for Alberta

69

Parameter Constant LTE Mills LM(-1) LP AD ADOTH HHINC kid0 CHAINS LPOTH T Parameter Constant LTE Mills LM(-1) LP AD ADOTH HHINC kid0 CHAINS LPOTH T Parameter Constant LTE Mills LM(-1) LP AD ADOTH HHINC kid0 CHAINS LPOTH T

fresh pork Coeff. t -0.36094 -1.21071 0.072953 *** 11.1937 -8.33E-04 -0.4805 2.42E-03 1.49012 0.114247 * 1.78142 -2.85E-09 -0.92505 -4.98E-10 -0.26117 -5.83E-07 *** -5.49193 0.010719 ** 2.11242 -1.14E-03 -0.58121 0.067413 0.595339 -3.16E-05 -7.52E-03 fresh beef Coeff. t 0.351397 0.375913 0.165816 *** 14.9388 -0.02996 *** -9.78286 9.24E-03 *** 4.67642 0.233931 1.21344 2.71E-09 0.227366 6.61E-10 0.258197 -6.68E-07 *** -4.29446 0.030908 *** 3.87672 -5.25E-03 * -1.88588 -0.5109 -1.441 -6.62E-03 ** -2.08502 fresh poultry Coeff. t 1.27271 * 1.85226 -0.01671 -1.35518 -0.03841 *** -10.1141 -1.14E-03 -0.48785 -0.60068 ** -2.0219 1.95E-09 0.821185 -1.68E-09 -0.6981 3.11E-07 * 1.89554 -7.07E-03 -0.92891 8.97E-04 0.332165 0.08225 0.552735 -3.95E-03 -1.00352

semi- pork Coeff. t -0.54849 ** -2.57109 -2.33E-03 -0.83465 0.011623 *** 10.8757 -7.42E-04 -0.5571 0.052929 *** 3.79966 1.42E-08 ** 3.15502 8.66E-10 1.08029 -3.19E-08 -0.74184 -9.77E-04 -0.51114 -7.71E-04 -1.06931 0.244734 ** 2.39185 -1.47E-03 ** -1.97216 semi- beef Coeff. t 6.94E-03 0.064486 2.60E-03 * 1.78283 7.37E-03 *** 8.83481 -2.38E-03 ** -2.11802 2.63E-03 0.218598 6.77E-08 * 1.75914 2.18E-10 0.508088 -2.71E-08 -1.44794 -2.68E-03 ** -2.41227 9.77E-04 ** 2.68324 -5.37E-03 -0.1128 -5.98E-04 -1.21761 semi- poultry Coeff. t -0.27883 ** -3.0463 -7.59E-03 ** -2.53948 0.010576 *** 7.55185 -7.63E-03 *** -5.62058 0.033497 ** 2.28764 -2.61E-08 -0.8098 2.37E-10 0.590536 1.25E-07 *** 3.57833 -1.31E-03 -0.86632 1.30E-03 ** 2.36912 0.103963 *** 3.36522 5.27E-03 *** 5.24157

fully- pork Coeff. t -0.59213 -1.64228 -0.03013 *** -4.76154 7.99E-03 *** 6.82324 2.78E-03 ** 1.98723 0.117679 ** 2.34933 -3.36E-09 -1.42474 5.78E-10 0.413931 -7.78E-08 -1.06732 7.27E-03 ** 2.23362 -3.70E-04 -0.25247 0.262009 1.299 -1.23E-03 -0.38882 fully- beef Coeff. t -0.07591 -0.65227 2.15E-03 1.00377 3.20E-03 ** 3.07584 -5.70E-03 *** -3.81235 0.011191 0.791032 -2.22E-08 ** -2.19256 -7.57E-10 ** -2.76303 -3.08E-08 -1.5842 -1.62E-03 -1.60216 7.61E-04 ** 2.23761 0.040548 0.61035 1.07E-03 0.764119 fully- poultry Coeff. t -0.54284 -0.84665 -0.04904 *** -5.98104 5.61E-03 *** 4.8768 5.12E-03 *** 3.75769 0.057502 0.503953 -1.98E-09 -0.63167 -1.96E-09 -0.79922 2.96E-07 ** 2.74588 -0.02471 *** -4.88868 2.52E-03 1.35102 0.349559 1.24489 -9.55E-05 -0.02401

70

Parameter Constant LTE Mills LM(-1) LP AD ADOTH HHINC kid0 CHAINS LPOTH T

fresh others Coeff. t 0.670195 1.20833 -0.02139 *** -3.3616 0.015338 *** 13.4247 3.47E-03 ** 2.57077 0.078099 0.656588 -9.28E-09 -0.59227 1.52E-09 1.3386 1.54E-07 ** 2.34258 8.14E-03 ** 2.95743 2.59E-03 ** 2.47472 -0.38507 -1.03333 -6.92E-04 -0.4386

semi- others Coeff. t 0.236347 * 1.92964 -0.01262 ** -3.06677 0.01039 *** 9.33132 -5.50E-03 *** -4.97079 -0.09806 ** -2.67214 -1.73E-08 *** -3.50032 -8.56E-10 -1.39026 1.02E-08 0.324391 -5.58E-03 ** -2.19674 1.67E-03 ** 2.69032 0.0402 0.875707 3.51E-03 ** 2.10669

fully- others Coeff. t 0.86154 0.674804 -0.10372 *** -9.89158 -2.91E-03 ** -2.49723 6.51E-05 0.046402 -2.97E-03 -0.01169 -3.41E-09 -0.24573 1.67E-09 0.783073 5.21E-07 *** 4.19447 -0.01309 ** -2.24579 -3.19E-03 -1.60862 -0.28933 -0.65954 4.83E-03 ** 2.26344

Results The results for each of the above models show some similarities and some differences across regions. In the first stage of the model in Ontario, older aged, better educated households with larger household sizes are all more likely to purchase fresh, semi-processed and fully processed pork products but urban dwellers are less likely to purchases each of the pork products. In Alberta, only age is a significant explanatory for any pork purchase decision although urban dwellers are less likely to purchase semi or fully processed pork. In both provinces there is evidence of an increased tendency to purchase pork products over time. Comparing pork and all other meat products in Ontario the results suggest that older aged households are more likely to purchase semi and fully processed beef products, are less likely to purchase semi and fully processed poultry products and more likely to purchase semi and fully processed other meat products (mainly seafood). Household size has a positive impact on purchases of all semi and fully processed meat products (except semi processed beef) and higher education levels have positive impacts on the decision to purchase semi and fully processed poultry but mixed effects on beef and other meats. The results are much less consistent for Alberta with age of household head being the most consistent explanatory of the decision to purchase any meat in fresh, semi or fully processed form. In the models explaining the level of expenditure for each of the twelve

71

meat types the consistent explanators appear to be the household size and/or having children in the household (in both provinces). Price responses, when statistically significant, suggest inelastic demands for most of the twelve meat types in both provinces (a 1% decrease (increase) in price results in a less than one percent increase (decrease ) in quantity sold). Advertising effects do not appear to be significant across the twelve meat types in explaining the level of meat expenditure. Households with children are likely to spend less on semi and fully processed pork and beef but

likely to spend more on fully processed poultry

products. In Ontario

households with higher levels of income are likely to spend less on all types of pork, on fresh and semi-processed beef but more on all types of poultry and fresh and semi-processed other meat products. In Alberta the effects of income are negative for fresh pork and beef but positive for all types of poultry products.

72

Canadian Store Choice Analysis Introduction The second objective of the study is to investigate how Canadian households make store choice decisions in purchasing meat products. In particular, the analysis focuses on the impact of store advertising and household demographic variables on store choice purchasing patterns. First, this section provides the data generation for the analysis followed by the data descriptive statistics. Then the explanation of model specification and econometric method are given. The model results and summary are provided in the conclusion.

Data setup and descriptive statistics

Nielsen Homescan™ data is the source of data in this analysis. The store choice analysis focuses on the Canadian household purchase information in the provinces of Ontario and Alberta over the time period 2002 to 2007. According to estimated marketing shares, shopping trips and regional differences, six grocery chains are selected for specific analysis in each province (Ontario and Alberta). In Ontario, the six grocery chains include: Loblaws, Metro, Safeway, Co-op, Sobeys (Empire), and all others. In Alberta, the six grocery chains are Loblaws, Safeway, Co-op, Empire, JPG (Save On Foods) and all others. Loblaws, Safeway, and Co-op are used in both provincial store choice models to make a comparison. In the following section a summary and short history for each of the grocery chains is provided.

Market share

Aggregate annual meat expenditure market share for each of the grocery store chains for the period 2002 to 2007 (for the Homescan panelists in this study) are reported in this section.

73

Error! Reference source not found. below reports aggregate market shares for each of six grocery store chains in Ontario and Alberta.

Table 5.1. Market share for store meat expenditure in Ontario and Alberta.

YEAR

2002

2003

2004

2005

2006

2007

Coop

0%

0%

0%

0%

0%

0%

METRO

28%

31%

29%

27%

27%

26%

Safeway

0%

0%

1%

0%

0%

0%

others

3%

4%

5%

5%

6%

6%

Loblaws

52%

47%

48%

48%

48%

49%

Empire

15%

18%

18%

19%

18%

18%

Source: Nielsen Homescan™ Panel, Ontario 2002 to 2007

YEAR

2002

2003

2004

2005

2006

2007

Coop

17%

15%

15%

15%

17%

17%

Empire

17%

16%

16%

17%

18%

16%

8%

7%

6%

5%

6%

7%

Loblaws

15%

17%

16%

18%

21%

23%

Safeway

39%

42%

43%

40%

34%

31%

4%

4%

5%

5%

5%

6%

JPG

others

Source: Nielsen Homescan™ Panel, Alberta 2002 to 2007

74

Table 5.2 Market Share and Household Spending, by category, , in Ontario and Alberta $

Coop

Empire

Loblaws

METRO

Safeway

others

0 [0-50] [50-100] [100-300] [300-500] [500-1000] [1000+]

6199 16 1 6216

2710 1682 639 816 219 127 23

706 1456 941 1856 660 493 104

1653 1716 847 1367 400 197 36

6157 17 16 17 6 3

3733 1806 391 259 20 7

0

0

%

Coop 6199 13 1 1 1 1

Empire 2710 1714 701 458 337 296

Loblaws 706 1257 1060 919 832 1442

METRO 1653 1548 1012 739 614 650

Safeway 6157 14 12 15 9 9

others 3733 1777 383 150 81 92

0 0-20% 20-40% 40-60% 60-80% 80%<

0 0 0

Source: Nielsen Homescan™ Panel, Ontario 2002 to 2007 $ 0

Coop

Empire

JPG

Loblaws

Safeway

others

1907

1381

2231

1298

914

1905

[0-50]

357

769

431

741

625

787

[50-100]

192

301

142

347

341

189

[100-300]

359

398

185

443

625

154

[300-500]

139

113

41

134

273

10

[500-1000]

87

80

16

73

221

3

7

6

2

12

49

[1000+]

% 0

Coop

Empire

JPG

Loblaws

Safeway

others

1907

1381

2231

1298

914

1905

0-20%

389

874

455

813

602

835

20-40%

226

304

137

357

374

166

40-60%

138

186

90

204

306

68

60-80%

133

138

48

158

319

43

80%<

255

165

87

218

533

31

Source: Nielsen Homescan™ Panel, Alberta 2002 to 2007

75

Table 5.3. Number of grocery store chains visited in Ontario and Alberta, 2002-2007

Source: Nielsen Homescan™ Panel, Ontario 2002 to 2007

76

Source: Nielsen Homescan™ Panel, Alberta 2002 to 2007

Model specification and econometric method

The source of data used in the store choice analysis is the same balanced panel of sample data that is used in the meat choice analysis in the fourth section of this report. Due to the zero consumption problem, not all households in Ontario and Alberta have positive expenditures at all six grocery chains. Each household is assumed to face a two-step hierarchy in decision making: households first make the decision of where to shop (participation step), then they will decide how much to spend in the chosen grocery store once they have made the store choice decision (expenditure step). Therefore a two-step estimation procedure following the Heien and Wessels (1990) Working-Leser demand system procedure is applied in the store choice demand analysis. In the first step, a probit regression is computed that determines the probability that a given household will shop at each grocery store. The probability of participation is then used as an instrument in the second-stage estimation of the Working-Leser demand system

77

. Participation decision for grocery stores (where to shop)

The first stage of the demand system is modeled as a participation choice problem: the dependent variable is represented by a binary choice variable yiht  1 if household h decides to shop at a given grocery store i at period t and is yiht  0 if the household does not choose to shop at period t. Then given E ( yiht )  1* piht  0*(1  piht )  piht , followed by same method as in Chapter 4, the grocery store participation decision is modeled as a function of household demographic variables and total meat expenditure in all grocery stores.

So the likelihood of household grocery store participation decision (Pr[ yiht  1]) for a random effects panel can be expressed as:   Pr[ yiht  1]  Pr[ X iht   aiht   ih  0]   ( X iht )

and the likelihood of households that do not shop at a given grocery store is:   Pr[ yiht  0]  Pr[ X iht   aiht   ih  0]  1   ( X iht )

where  X iht   0  1 *Texp  2 * hhinc  3 * hage  4 * hages  5 * urban  6 * hhsize  7 *T

Expenditure decision for grocery stores (how much to spend)

The second step is the estimation of the store expenditure share equations of the WorkingLeser demand system via seemingly unrelated regression (SUR) of the expenditure share that household h spends in a given grocery store i in time period t. In the Working-Leser model, each store expenditure share is a linear function of the log of the total expenditure in all grocery store chains and household demographic variables, lagged store advertising variables.

78

The general form of the second stage equations of Working-Leser demand function can be expressed as:

i  a0  a1 *log(T exp)  a2 * hage  a3 *log[ M i (1)]  a4 * Mills  a5 * hhedu a6 *hhinc  a7 * KID  a8 * urban  a9 * hhsize  a10 * AD  a11 * ADoth  a12 * ch sin s  a13 *T   it

where (i) represents the one of the six grocery store chains in Ontario and Alberta; wi is the store expenditure share of grocery chain i; Texp is the total expenditure of all grocery store chains; M(-1) is the lagged store i expenditure (on year lag) which may lead to a habit formation, where past consumption decisions serve as predictors of future purchase decisions; AD is the advertising information(one year lag) of grocery chain i in a given year; ADoth is the advertising information(one year lag) of other grocery chains in a given year; HHINC is the household income; Kid is the presence of children in the household; Chains represents the number of grocery store chains where household visited. T is the time trend variable. Expenditure elasticity: Model testing and empirical results

Likelihood ratio tests (LRT) are applied to select the best fitting model among a number of models. The definitions of variables used for the analysis are listed below.

79

Table 5.4 Definition and sample statistics of variables used for store choice analysis Ontario Variables

Definitions

First stage binary dependent variables 1 if choose Coop, 0 otherwise PCOOP 1 if choose Sobeys(empire), 0 otherwise PEMP 1 if choose Loblaws, 0 otherwise PLOB 1 if choose Metro, 0 otherwise PMET 1 if choose Save on foods(JPG), 0 otherwise PJPG 1 if choose Safeway, 0 otherwise PSAFE 1 if choose other stores, 0 otherwise POTH Second stage expenditure share dependent variables store expenditure share of Coop COOPSH store expenditure share of Sobeys EMPSH store expenditure share of Loblaws LOBSH METROSH store expenditure share of Metro store expenditure share of Save on foods JPGSH store expenditure share of Safeway SAFESH store expenditure share of others OTHSH Logged form of meat expenditure logged store expenditure of Coop LCOOP logged store expenditure of Sobeys LEMP logged store expenditure of Loblaws LLOB logged store expenditure of Metro LMET logged store expenditure of Save on foods LJPG logged store expenditure of Safeway LSAFE logged store expenditure of others LOTH Total expenditure on all stores TEXP logged total exp on all stores LTE HH demographic and purchase information Annual HH income(C$, midpoint) HHINC Household head age(midpoint) HAGE Squared household head age(midpoint) HAGES 1 if HH with children , 0 otherwise KID1 1 if HH without children , 0 otherwise KID0 HHEDU0 1 if no high school edu, 0 otherwise HHEDU1 1 if higher edu, 0 otherwise URBAN 1 if in urban area, 0 otherwise RURAL 1 if in rural area, 0 otherwise HHSIZE Number of members in household T year 1-6 CHAINS Number of grocery chains HH visited

Alberta

Mean

SD

Mean

SD

0.003 0.564 0.886 0.734 N/A 0.009 0.399

0.052 0.496 0.317 0.442 N/A 0.097 0.490

0.374 0.547 0.574 N/A 0.268 0.700 0.375

0.484 0.498 0.495 N/A 0.443 0.458 0.484

0.001 0.174 0.455 0.294 N/A 0.004 0.072

0.017 0.260 0.343 0.314 N/A 0.054 0.165

0.165 0.162 0.191 N/A 0.078 0.339 0.065

0.296 0.262 0.283 N/A 0.201 0.358 0.156

0.003 0.969 1.796 1.356 N/A 0.018 0.562 385 5.579

0.064 0.962 0.840 0.958 N/A 0.191 0.751 325 0.968

0.730 0.942 1.024 N/A 0.435 1.410 0.549 416 5.670

1.010 0.973 0.986 N/A 0.783 1.058 0.762 338 0.955

52386 22189 51932 21909 55 12 53 12 3212 1281 3006 1272 0.223 0.416 0.222 0.416 0.777 0.416 0.778 0.416 0.140 0.347 0.130 0.336 0.860 0.347 0.870 0.336 0.685 0.465 0.685 0.465 0.315 0.465 0.315 0.465 2.397 1.210 2.337 1.209 3.500 1.708 3.500 1.708 2.596 0.894 2.839 1.175

80

Table continued...

Variables

Definitions

Ontario

Alberta

Mean

SD

Mean

SD

Advertising expenditure by grocery store chains TA1COOP

One year lag of AD for Coop

1903227

652783

1903227

652837

TA1EMP

One year lag of AD for Sobeys

8604716

1003701

8604716

1003784

TA1LOB

One year lag of AD for Loblaws

9552057

932734

9552057

932812

TA1MET

One year lag of AD for Metro

5505401

960972

N/A

N/A

TA1JPG

One year lag of AD for Save on foods

N/A

N/A

4712313

1906918

TA1SAFE

One year lag of AD for Safeway

1.41689D+07

2681527

1.41689D+07

2681752

TA1OTH

One year lag of AD for othes

2.72842D+07

3295569

2.80773D+07

1437932

TA2COOP

two years lag of AD for Coop

1707535

685734

1707535

685791

TA2EMP

two years lag of AD for Sobeys

8731168

1150445

8731168

1150542

TA2LOB

two years lag of AD for Loblaws

9018072

1064810

9018072

1064899

TA2MET

two years lag of AD for Metro

5465500

883694

N/A

N/A

TA2JPG

two years lag of AD for Save on foods

N/A

N/A

3748053

2110160

TA2SAFE

two years lag of AD for Safeway

1.27778D+07

2467236

1.27778D+07

2467442

TA2OTH

two years lag of AD for others

2.68634D+07

3335412

2.85809D+07

3215373

The first stage estimation results are reported in Tables 5.5 and 5.6 below. In Ontario households who spend more on meat are more likely to shop at Empire, Loblaws, Metro and less likely to shop at ‗other‘ grocery stores. Higher incomes, older household head age and larger household size explain the decision to purchase meat at a Loblaws store (including all stores owned by Loblaws). Over time more households are choosing to purchase meat at Metro, Empire and other grocery stores. Households in urban areas are less likely to shop at Empire, Loblaws but more likely to shop at Metro and other grocery stores. In comparison, in Alberta households who spend more on meat are more likely to spend that money at Co-op, Empire, Loblaws and Safeway grocery stores (reflecting the increasing concentration in the grocery retailing industry in Canada). Urban dwellers are less likely to buy meat at Co-op, Empire, Loblaws but more likely to make meat purchases at JPG and Safeway. Larger household size suggests an increased probability of purchasing meat at Co-op, Empire, Loblaws and other grocery stores. In Alberta higher income households are more likely to make meat purchases at Loblaws, JPG and other grocery stores and less likely to make meat purchases at Co-op stores. Over time, for these households, the probability of purchasing meat is growing at Empire, Loblaws and other grocery stores and declining at Safeway stores.

81

TABLE 5.5. First-Step Probit Estimates of Ontario Variables

Co-op Coeff.

C TEXP HHINC

Empire t

Coeff.

-6.78106

**

-2.79104

0.239926

-7.13E-04

*

-1.87758

3.60E-04

0.282428

-1.01E-06

1.28E-06

Loblaws t

***

Coeff.

t

0.728832

-0.752197

**

-1.82496

6.42638

9.12E-04

***

9.68237

-1.23395

2.19E-06

**

2.02138

HAGE

0.184495

*

1.90583

-0.017894

-1.43564

0.046938

**

2.99261

HAGES

-1.96E-03

**

-2.05538

2.28E-04

**

1.97019

-3.83E-04

**

-2.61733

T

-0.075856

-1.47433

0.031567

***

3.31633

3.21E-03

URBAN

0.030476

0.158377

-0.220275

***

-6.18276

-0.103392

HHSIZE

0.100847

1.44663

0.057763

***

3.6628

0.110101

Variables

Metro Coeff.

C TEXP

Safeway t

***

3.96679

-4.29054

1.52E-04

**

2.72058

-2.59E-04

t ***

**

-2.14611

***

5.00125

Others

Coeff.

1.42765

0.252021

Coeff.

t

-3.39826

-0.875792

**

-2.622

-1.34681

-2.33E-04

***

-4.20999

***

3.06324

HHINC

-1.10E-06

-1.27254

-1.26E-06

-0.507808

2.57E-06

HAGE

-0.040289

**

-2.96282

0.078985

*

1.6783

-0.019865

HAGES

4.01E-04

***

3.18104

-7.43E-04

*

-1.73004

2.04E-04

*

T

0.016788

*

1.65743

0.03414

1.17678

0.102469

***

10.4763

URBAN

0.190689

***

5.12177

-0.050178

-0.475992

0.131301

***

3.61173

HHSIZE

-0.01832

-1.10424

4.45E-03

0.090857

0.233075

***

14.4603

-1.5694 1.738

TABLE 5.6. First-Step Probit Estimates of Alberta Variables

Co-op Coeff.

C

Empire t

Coeff.

-0.426286

***

-4.649930

-0.048825

TEXP

0.000199

**

2.694440

0.000645

KID1

-0.276462

***

-3.292030

HHINC

-0.000002

**

-2.139450

T

0.011435

Loblaws t

Coeff.

t

-0.531572

-0.275020

**

-3.016070

8.313550

0.000436

***

5.538200

0.001383

0.016389

0.025023

-0.000001

-0.989910

0.000002

**

***

0.298622 2.172410

0.832403

0.026846

**

1.967770

0.041672

**

3.057700

URBAN

-0.204854

***

-4.071010

-0.447434

***

-8.723360

-0.226411

***

-4.461060

HHSIZE

0.132551

***

4.244100

0.076410

**

2.431640

0.069205

**

2.212200

Variables

JPG Coeff.

C

t ***

Coeff.

-7.770250

0.165426

*

0.000056

0.737098

0.000958

***

KID1

0.143320

1.639150

HHINC

0.000003

**

-0.015196

URBAN

0.150495

HHSIZE

-0.044241

**

Others t

TEXP

T

-0.758335

Safeway

Coeff.

t

1.695730

-1.062700

10.107500

0.000044

***

-11.233200

0.085657

0.954490

-0.143739

*

-1.716510

2.122960

0.000000

0.148954

0.000004

**

3.056050

-1.056880

-0.045352

**

-3.084570

0.077385

***

5.543680

2.778470

0.762524

***

14.582500

-0.278282

***

-5.477800

-1.343190

-0.159071

***

-4.740220

0.202521

***

6.399090

0.590272

82

The second stage estimation results are reported in Tables 5.7 and 5.8. These results show significant explanatory variables for the decision on level of spending on meat at each of the grocery store chains. The number of grocery store chains shopped at by a household is a significant determinant of level of meat spending at all chains except Loblaws in Ontario. In Alberta the number of chains shopped at is positively related to the level of spending at Empire, JPG and other stores but is negatively related to the level of spending at Safeway and Loblaws brand stores. In Alberta, households with higher levels of education spend more on meat at Co-op, Loblaws and JPG and less at Empire and Safeway. In Ontario households with higher levels of education spend more on meat at Metro and less at Empire. Households with larger sizes spend more on meat at Loblaws and other stores in Ontario and at Co-op, Loblaws and others in Alberta. Store advertising has no significant effects in Alberta but has small positive effect on meat spending for Co-op and Safeway stores in Ontario – both nontraditional and small retailers in Ontario. TABLE 5.7. Second-Step Working-Leser Demand Estimates for Ontario

Parameter Constant LTE AD ADoth HHEDU1 Chains HHINC T HHSIZE Mills Urban LM(-1) Parameter Constant LTE AD ADoth HHEDU1

Co-op Coeff.

t

Empire Coeff.

-0.0889

***

-3.76

0.312097

5.23E-03

***

4.79

-3.84E-04

1.28E-08

***

8.87

-5.27E-09

0.09

2.74E-11 9.58E-05 0.023847

***

3.73E-08 -7.71E-03

***

-1.28E-04 -0.243014

***

-7.28E-04 -0.35268

***

Metro Coeff. 0.389731

4.41

0.134206 0.036371

**

-3.08

5.00E-09

1.58

-2.75E-09

**

-2.24

2.57E-09

1.55

0.07

-0.034429

***

-4.00

-9.78E-03

8.43

0.022994

***

5.39

-0.084334

***

1.38

-3.94E-07

**

-2.77

5.63E-07

**

2.92

-6.62

7.26E-03

*

1.74

-0.012336

**

-2.13

-0.18

-1.10E-03

-0.44

0.011428

***

3.23

-5.02

0.047491

***

5.35

0.030311

**

2.10

-0.53

-0.042543

***

-6.88

-0.013403

*

-1.64

-10.98

0.06971

***

19.70

0.081649

***

18.90

Safeway Coeff.

t

1.32 ***

7.52

-0.93

Others Coeff.

-17.63

t

4.17

0.062283

**

2.98

0.190583

***

4.10

-4.88E-03

-1.14

4.86E-03

***

8.32

-0.041199

***

-14.96

-4.30E-09

-1.37

1.59E-09

***

4.23

-9.78E-09

***

-9.96

-2.23E-09

-1.46

-2.63E-09

***

-5.52

5.02E-09

***

6.08

4.93

5.71E-04

0.53

-2.11E-03

0.045643

***

t

-0.12

t

***

t

Loblaws Coeff.

***

-0.45

83

Chains HHINC T HHSIZE Mills Urban LM(-1)

0.016492

***

3.35

0.012716

7.14

8.28E-03

**

2.60

-7.81E-07

***

-4.42

-1.23E-08

-0.71

5.87E-07

***

6.94

-0.62

5.68E-04

0.70

0.01549

***

5.68

-1.14E-03

***

-3.36

0.017652

***

9.60

-3.27E-03

***

-0.026709

***

-8.42

0.035866

**

3.15

0.07339

***

3.52

0.055955

***

9.78

0.047226

***

6.54

-2.21E-03

**

-1.97

0.011657

***

3.48

0.070696

***

18.23

0.087882

***

3.35

0.042744

***

12.45

TABLE 5.8. Second-Step Working-Leser Demand Estimates for Alberta

Parameter Constant LTE Mills LM(-1) AD ADoth HHINC HHEDU1 KID1 HHSIZE T Chains Parameter Constant LTE Mills LM(-1) AD ADoth HHINC HHEDU1 KID1 HHSIZE T Chains

Co-op Coeff. 0.043411

t 0.36

4.61E-03

Empire Coeff. 0.060867

0.92

7.21E-03

0.045439

***

19.43

-7.33E-03

0.021652

***

11.37 0.50

2.66E-09

t

Loblaws Coeff.

0.48

2.24E-01

*

t 1.71

1.29

-2.99E-02

***

-4.91

-3.30

1.11E-02

***

4.98

2.85E-03

1.53

1.11E-02

***

6.30

1.07E-09

0.31

1.69E-10

0.06

***

1.62E-09

0.79

7.74E-10

0.35

-6.46E-10

-0.29

-9.02E-08

-0.40

-3.84E-07

*

-1.70

3.36E-07

1.37

**

-2.01

0.038686

-0.67

0.019088

0.027603

**

2.16

-0.032567

-0.039031

**

-2.25

-1.21E-02

0.026764

***

-4.73E-03 -0.031057

***

JPG Coeff. 1.36E-01

**

3.04 1.01

4.01

7.30E-03

1.08

0.019427

**

2.85

-0.67

-2.84E-03

-0.38

0.013741

*

1.85

-6.95

0.016712

3.94

4.17E-03

t

***

Safeway Coeff.

t

1.39

4.24E-01

**

2.71

0.88

Others Coeff. 1.11E-01

t 1.33

-2.53E-02

***

-4.80

8.50E-02

***

12.24

-4.15E-02

***

-8.27

-1.47E-02

***

-6.51

1.88E-02

***

7.86

-0.053319

***

-24.56

-5.15E-03

**

-2.22

0.021272

***

12.94

-0.051711

***

-19.83

-5.40E-10

-0.17

-2.83E-09

-1.08

-5.33E-10

-0.33

6.80E-10

0.41

-3.51E-09

-1.29

1.08E-09

0.73

9.01E-08

0.49

-1.51E-07

-0.49

1.99E-07

1.35

3.43E-02

***

3.89

-0.080747

***

-4.26

0.012714

0.027611

**

2.18

0.056311

**

2.74

-0.051892

***

-4.37

-5.96E-03

-1.30

-0.077532

***

-10.03

0.030002

***

6.47

-2.08E-03

-0.33

-5.85E-03

-0.67

1.76E-03

2.97

-3.66E-02

-6.63

0.036881

9.86E-03

**

***

*

1.64

0.37 ***

14.50

84

National and Store Brand Choice Analysis Introduction

The third objective of the study is to identify how consumers make decisions about private label versus national branded meat products in their fully processed value-added meat category. The analysis aims to quantify the impact of price, advertising, demographic and regional characteristic differences in brand choice behaviour, and these differences in the behaviour across meat types. In this chapter, the data setup for the analysis is provided followed by the data descriptive statistics. Then the explanation of model specification and econometric methods is given. The model results and summary are finally provided in the final section of the chapter. Data setup and descriptive statistics

Nielsen Homescan™ data is sourced for the brand choice analysis. The brand choice demand analysis focuses on the fully processed meat purchase information in the provinces of Ontario and Alberta over the time period 2002 to 2007. The same household panel as used in sectionr 4 and 5 was analysed in the brand choice analysis. The panel totalled 1036 households in Ontario and 508 households in Alberta in the balanced panel. Three fully processed meat types: pork, poultry, and other meat (mainly fish products) are used in the analysis, there was almost no shares of branded beef purchased, so beef was excluded in this analysis. In order to better understand the brand choice decisions, the national brands and private label products were grouped into four brand categories in detail according to their marketing shares, the four shares are the leading national branded products, other national branded products, the leading store branded products, and other store branded products. Then twelve choice alternatives in this analysis were identified: (1) leading national branded pork, poultry and others; (2) other national branded pork, poultry and others; (3) leading store branded pork, poultry and others; (4) other store branded pork, poultry and others. These product purchases were aggregated into annual expenditures by each household.

85

Table 6.1 Brand Categories Brand Categories

Leading National Brands

Meat Types Pork, Poultry, Others (Mainly Fish) Other Meats (Mainly Fish)

Other National Brands

Pork, Poultry, Others (Mainly Fish)

Leading Private Labels

Pork, Poultry, Others (Mainly Fish)

Other Private Labels

Pork, Poultry, Others (Mainly Fish)

Brands Schneider Maple Leaf Mitchells High Liner Fletchers Cooks Harvest Sterling Silver Anchor Grimms Burns Olympic Maple Birch Drake Olymel Vegreville Capital Packers Etc... Presidents Choice No Name Safeway Select Butchers Cut Compliments Country Morning Western Family Etc...

Model specification and econometric method

The source of data used in the national brands and store brands analysis is the same balanced panel of sample data that is used in previous analysis in the previous sections of this report. Due to the zero expenditure problem, not all households in Ontario and Alberta have positive expenditures on all twelve meat categories in every year. Each household is assumed to face a two-step hierarchy in decision making: households first make the decision of what brands and what types of meat to purchase (participation step), then they will decide how much they

86

will spend on the given product once they have made the brand choice decision (expenditure step). Therefore a two-step estimation following the Heien and Wessels (1990) Working-Leser demand system procedure is applied in the brand choice demand analysis. In the first step, a probit regression is computed that determines the probability that a given household will purchase a brand (national or store branded). The probability of participation is then used as an instrument in the second-stage estimation of the Working-Leser demand system Participation decision for brand choice (which brand to choose)

The first stage of the demand system is modeled as a participation brand choice problem: the dependent variable is represented by a binary choice variables yiht  1 if household h decides to purchase a branded fully processed meat product i at period t and is yiht  0 if the household

does

not

choose

the

given

brand

at

period

t.

Then

E ( yiht )  1* piht  0*(1  piht )  piht , followed by same method as in Chapter 4 and 5, the

brand choice participation decision is modeled as a function of household demographic variables and total meat consumption in all fully processed meat products..

So the likelihood of household brand choice decision (Pr[ yiht  1]) for a random effects panel can be expressed as:   Pr[ yiht  1]  Pr[ X iht   aiht   ih  0]   ( X iht )

and the likelihood of households that do not choose a given brand is:   Pr[ yiht  0]  Pr[ X iht   aiht   ih  0]  1   ( X iht )

where  X iht   0  1 *Total  2 * hhinc  3 * hage  4 * hages  5 * urban  6 * hhsize  7 *T

87

Expenditure decision for grocery stores (how much to spend)

The second step is the estimation of the store expenditure share equations of the WorkingLeser demand system via seemingly unrelated regression (SUR) of the expenditure share that household h spends in a given grocery store i in time period t. In the Working-Leser model, each store expenditure share is a linear function of the log of the total expenditure in all grocery store chains and household demographic variables, lagged store advertising variables. The general form of the second stage equations of Working-Leser demand function can be expressed as:

i  a0  a1 *log( Mtotal )   ij a2 *ln( pij )  a3 *log[ M i (1)]  a4 * Mills  a5 * hhedu  a6 * hhinc  a7 * KID  a8 * chains  a9 * hhsize  a10 * T  a11 * AD  a12 * ADoth  a13 * urban   it

where (i,j) represents the twelve branded fully processed meat products; wi is the expenditure share of meat product i among the twelve branded meat products; pij is the price of branded meat product ij; Mtotal is the total expenditure of all twelve fully processed meat products; M(-1) is the lagged meat i expenditure which may lead to a habit formation, where past consumption decisions serve as predictors of future purchase decisions. AD is the advertising information of a given branded meat i. ADoth is the total of other branded meat advertising information. HHINC is the household income. Kid is the presence of children in the household.

88

Stores represents the number of grocery store chains where household purchage the twelve meat products. T is the time trend variable. Urban represents household reside in urban area HHECU is the level of household head education; Mills is the inverse mill ratios obtained from the fist Probit model estimations.

Model testing and empirical results

TSP International 5.0 was the econometric software used for the estimation of parameters in this study. Likelihood ratio tests (LRT) were applied to select the best fitting model among a number of models. Definitions of variables used for the analysis are listed below in Table 6.2. Table 6.2 Definition and sample statistics of variables used for brand choice analysis Ontario Variables

Definitions

First stage binary dependent variables 1 if choose other NB pork, 0 otherwise D1NB0 1 if choose leading NB pork, 0 otherwise D1NB1 1 if choose other SB pork, 0 otherwise D1PL0 1 if choose leading SB pork, 0 otherwise D1PL1 1 if choose other NB poultry, 0 otherwise D3NB0 1 if choose leading NB poultry, 0 otherwise D3NB1 1 if choose other SB poultry, 0 otherwise D3PL0 1 if choose leading SB poultry, 0 otherwise D3PL1 1 if choose other NB other meats, 0 otherwise D4NB0 1 if choose leading NB other meats, 0 otherwise D4NB1 1 if choose other SB other meats, 0 otherwise D4PL0 1 if choose leading SB other meats, 0 otherwise D4PL1 Second stage expenditure share dependent variables expenditure share of other NB pork S1NB0 expenditure share of leading NB pork S1NB1

Mean 0.15 0.25 0.05 0.09 0.25 0.08 0.20 0.34 0.41 0.37 0.13 0.31

SD 0.36 0.43 0.23 0.29 0.43 0.27 0.40 0.47 0.49 0.48 0.34 0.46

0.05 0.16 0.07 0.19

Alberta Mean 0.23 0.18 0.11 0.05 0.21 0.13 0.30 0.11 0.44 0.40 0.18 0.12

SD 0.42 0.39 0.31 0.22 0.41 0.33 0.46 0.32 0.50 0.49 0.39 0.32

0.08 0.21 0.06 0.17

89

expenditure share of other SB pork S1PL0 expenditure share of leading SB pork S1PL1 expenditure share of other NB poultry S3NB0 expenditure share of leading NB poultry S3NB1 expenditure share of other SB poultry S3PL0 expenditure share of leading SB poultry S3PL1 expenditure share of other NB other meats S4NB0 expenditure share of leading NB other meats S4NB1 expenditure share of other SB other meats S4PL0 expenditure share of leading SB other meats S4PL1 Logged form of meat price logged price of other NB pork LP1NB0 logged price of leading NB pork LP1NB1 logged price of other SB pork LP1SB0 logged price of leading SB pork LP1SB1 logged price of other NB poultry LP3NB0 logged price of leading NB poultry LP3NB1 logged price of other SB poultry LP3SB0 logged price of leading SB poultry LP3SB1 logged price of other NB other meats LP4NB0 logged price of leading NB other meats LP4NB1 logged price of other SB other meats LP4SB0 logged price of leading SB other meats LP4SB1

0.01 0.02 0.08 0.02 0.06 0.13 0.15 0.13 0.03 0.11

0.08 0.09 0.19 0.09 0.17 0.25 0.26 0.25 0.12 0.22

0.03 0.02 0.08 0.04 0.12 0.04 0.18 0.15 0.05 0.04

0.13 0.09 0.20 0.14 0.24 0.14 0.28 0.26 0.15 0.14

0.78 1.04 0.74 0.88 0.96 0.92 0.90 0.97 0.99 0.96 1.05 1.07

0.09 0.09 0.05 0.13 0.08 0.05 0.09 0.08 0.12 0.09 0.04 0.12

0.83 1.03 0.74 0.73 0.89 0.99 0.97 0.92 1.08 1.00 1.05 1.08

0.10 0.08 0.05 0.05 0.12 0.10 0.10 0.07 0.20 0.09 0.07 0.08

Ontario Variables

Definitions

Advertisng expenditure by meat types AD for other NB pork AD1NB0 AD for leading NB pork AD1NB1 AD for other SB pork AD1PL1 AD for leading SB pork AD1PL0 AD for other NB poultry AD3NB0 AD for leading NB poultry AD3NB1 AD for other SB poultry AD3PL0 AD for leading SB poultry AD3PL1 AD for other NB other meats AD4NB0 AD for leading NB other meats AD4NB1 AD for other SB other meats AD4PL0 AD for leading SB other meats AD4PL1 HH demographic and purchase information T year 1-6

Mean

Alberta SD

Mean

SD

856027 678112 856027 678169 2020332 973264 2020332 973345 0 0 0 0 751563 310674 751563 310700 453368 325369 453368 325396 4300882 2055386 4300882 2055558 261385 357791 261385 357820 142351 318332 142351 318359 1504134 321012 1504134 321039 56927 126988 56927 126999 104409 150071 104409 150084 2701 6041 2701 6041 3.50

1.71

3.50

1.71

90

HHSIZE KID1 KID0 HAGE HAGES HHINC HHEDU1 HHEDU0 URBAN RURAL TOTAL LTE STORES

Number of members in household 1 if HH with children , 0 otherwise 1 if HH without children , 0 otherwise Household head age(midpoint) Squared household head age Annual HH income(C$, midpoint) 1 if higher edu, 0 otherwise 1 if no high school edu, 0 otherwise 1 if in urban area, 0 otherwise 1 if in rural area, 0 otherwise Total expenditure on all types of meat logged total exp on all types of meat Number of grocery chains HH visited

2.40 0.22 0.78 55 3212 52386 0.86 0.14 0.68 0.32 63.89 1.43 0.83

1.21 0.42 0.42 12 1281 22189 0.35 0.35 0.46 0.46 78.25 0.70 1.08

2.34 0.22 0.78 53 3006 51932 0.87 0.13 0.69 0.31 60.29 1.45 1.62

1.21 0.42 0.42 12 1272 21909 0.34 0.34 0.46 0.46 64.37 0.67 1.02

Note:

1 .The source of data is Nielsen Homescan™ Panel, Ontario& Alberta, 2002-07) 2. NB=National Brands, SB=Store branded (or Private labels)

TABLE 6.3. First-Step Probit Estimates for Ontario Variables

other NB pork Coeff.

C TOTAL HHINC HAGE HAGES URBAN HHSIZE T Variables

-4.2394 0.0023 0.0000 0.0462 -0.0003 -0.1221 0.1549 0.2793

t

*** *** ** ** *** ***

TOTAL HHINC HAGE HAGES URBAN HHSIZE T Variables

-8.0 8.9 -0.9 2.4 -1.6 -2.7 7.4 19.8

leading NB pork Coeff.

C

other NB poultry

-1.2639 0.0031 0.0000 -0.0249 0.0003 -0.1558 0.0070 0.1600

***

-3.3 13.5 3.7 -1.7 2.5 -4.0 0.4 14.7

other SB pork Coeff.

Coeff.

-0.8948 0.0042 0.0000 0.0096 -0.0002 0.0990 0.1166 -0.0445

t

** ***

* ** *** ***

-2.5 17.7 -0.8 0.7 -1.9 2.4 6.9 -4.1

leading NB poultry t

*** *** *** * ** ***

other NB other meats

Coeff.

-1.2747 0.0023 0.0000 0.0047 -0.0001 -0.2979 0.1478 -0.0413

*** *** **

-2.7 8.3 -3.0 0.3 -0.8 -5.8 6.7 -2.8

other SB poultry t

Coeff.

-0.9633 0.0046 0.0000 -0.0077 0.0001 -0.0442 0.1710 0.0757

t

** *** *

*** ***

-2.8 18.7 -1.7 -0.6 0.8 -1.2 10.6 7.6

leading NB other meats t

** *** **

Coeff.

Coeff.

0.3266 0.0042 0.0000 -0.0472 0.0005 -0.0746 0.1207 0.0360

t

*** *** *** *** ** *** ***

1.0 18.0 -5.8 -3.7 4.1 -2.0 7.5 3.6

other SB other meats t

Coeff.

t

91

C TOTAL HHINC HAGE HAGES URBAN HHSIZE T Variables

-4.2182 0.0015 0.0000 0.0485 -0.0003 -0.1682 0.0431 0.1675

*** *** ** * ** ***

-5.7 4.5 2.4 1.8 -1.3 -2.9 1.6 9.4

leading SB pork

TOTAL HHINC HAGE HAGES URBAN HHSIZE T

-2.9634 0.0030 0.0000 0.0217 -0.0001 -0.0785 0.0027 0.1077

t

*** *** **

***

*** ***

* ***

-1.4 12.3 -5.7 -0.3 -0.7 -1.9 6.0 0.8

leading SB poultry

Coeff. C

-0.4990 0.0028 0.0000 -0.0040 -0.0001 -0.0773 0.1041 0.0084 Coeff.

-5.6 11.6 2.8 1.1 -0.6 -1.6 0.1 7.5

-1.6313 0.0033 0.0000 -0.0078 0.0001 0.0862 0.0571 0.0523

*** *** * ** ** *** ***

-3.9 13.4 0.7 -0.5 0.5 1.9 2.9 4.2

* ** ***

leading SB other meats t

-1.2775 0.0081 0.0000 0.0286 -0.0004 -0.0372 0.0730 -0.0949

*** ***

-3.6 27.6 1.8 2.1 -2.9 -1.0 4.3 -8.9

Coeff.

-2.0586 0.0065 0.0000 0.0260 -0.0002 -0.0033 -0.0061 0.0160

t

*** *** *** * *

-5.6 24.6 7.5 1.9 -1.8 -0.1 -0.4 1.5

TABLE 6.3. First-Step Probit Estimates for Alberta Variables

other NB pork Coeff.

C

-0.07023

TOTAL

3.51E-03

HHINC

2.62E-06

HAGE

other NB poultry t

Coeff.

other NB other meats t

Coeff.

t

-0.2

-2.40342

***

-5.3

-1.3488

***

-3.5

***

8.8

6.42E-03

***

15.4

7.38E-03

***

17.4

**

2.1

1.06E-06

0.8

1.45E-06

1.2

-0.03071

*

-1.9

0.066061

***

3.7

1.6

HAGES

3.57E-04

**

2.3

-6.72E-04

***

-3.9

0.023545 -2.47E04

*

-1.7

KID0

-0.04012

-0.6

-0.18918

**

-2.7

-0.21184

***

-3.2

URBAN

-0.34594

***

-6.3

-0.12008

**

-2.1

0.243327

***

4.6

HHEDU1

-0.34601

***

-4.6

-0.1232

-1.5

-0.15108

**

-2.1

T

0.018985

1.2

-0.02236

-1.4

0.083515

***

5.8

Variables

leading NB pork Coeff.

leading NB poultry t

Coeff.

leading NB other meats t

Coeff.

t

C

-1.42271

**

-3.2

-1.56135

**

-3.2

-0.58208

-1.5

TOTAL

3.42E-03

***

8.4

5.08E-03

***

11.8

HHINC

9.47E-07

0.7

-2.90E-06

**

-2.0

4.98E-03 -1.73E06

HAGE

-0.01426

-0.8

0.01197

0.6

HAGES

2.22E-04

1.3

-1.26E-04

-0.7

9.74E-03 -3.45E05

KID0

-0.02003

-0.3

-0.11249

-1.4

-0.19501

**

-3.0

URBAN

0.088473

1.5

-0.0775

-1.2

-0.1297

**

-2.5

HHEDU1

0.01614

0.2

0.046498

0.5

-0.08002

***

13.0 -1.5 0.6 -0.2

-1.1

92

T

0.075644 Variables

***

4.6

0.010543

other SB pork

other SB poultry

Coeff.

t

-3.39912

***

-5.2

0.391194

TOTAL

4.12E-03

***

9.0

6.90E-03

HHINC

-3.75E-07

HAGES

0.05271 -4.38E-04

-3.74E03

-0.3

other SB other meats

Coeff.

C

HAGE

0.6

t

Coeff.

t

1.0

-1.57529

***

-3.6

***

17.2

4.82E-03

***

11.9

*

***

-0.2

-2.39E-06

-1.9

4.95E-06

**

2.1

-0.02327

-1.5

-0.01171

-0.7

3.6

*

-1.9

1.84E-04

1.2

1.52E-04

0.9

-1.0

-0.28286

***

-4.2

0.0131

0.2

KID0

-0.08842

URBAN

-0.16285

**

-2.4

0.084935

1.5

0.088345

1.5

HHEDU1

-0.32746

***

-3.6

-0.20883

**

-2.8

0.134998

1.5

T

0.212839

***

10.0

-0.06787

***

-4.5

0.0124

0.7

Variables

leading SB pork

leading SB poultry

Coeff.

t

Coeff.

C

-3.01968

***

-4.0

-0.29479

TOTAL

2.72E-03

***

5.0

4.69E-03

HHINC

4.31E-06

**

2.2

5.20E-06

HAGE

0.032139

1.1

-0.03682

HAGES

-2.10E-04

-0.8

2.67E-04

KID0 URBAN

-0.40941 -0.05632

-4.0 -0.7

-0.47451 0.011565

HHEDU1

0.06364

0.5 1.8

T

0.04180

***

*

leading SB other meats t

Coeff.

t

-0.6

-1.47132

**

-3.0

***

10.5

4.32E-03

***

9.9

***

3.2

6.19E-06

***

4.0

*

-1.9

-0.0194

-1.0

1.4

1.68E-04

0.9

-6.1 0.2

-0.09963 0.052786

-1.2

-0.07519

-0.7

0.198169

-2.28E-03

-0.1

7.10E-03

***

0.8 *

1.9 0.4

Results from the first stage of the national versus store brand model suggest that the decision to purchase any of the four branded products is significantly affected by demographic characteristics in both Ontario and Alberta. In Ontario higher levels of household income are associated with higher probabilities of purchasing leading national and store brands for pork, poultry and other meats and other store brands of pork and poultry. In Alberta, higher incomes are associated with higher probabilities of purchasing leading store brands for pork, poultry and other meats but reduced probabilities of purchasing other store brands of pork and poultry and leading national brands of poultry (possibly reflecting the regional importance of Lilydale as a poultry processor in Alberta). There are also differences in the trends by meat type – for example over time there is a higher probability to purchase all four brands of pork and other meats in Ontario but opposite signs for poultry products. In Alberta the trend variables over time suggest positive signs on the probability of purchasing leading national brand, leading store brand and

93

other store brand for pork but negative signs for leading store brand and other store brands for poultry.

TABLE6.4. Second-Step Working- Leser Model Estimates for Ontario

Variables

other NB pork

leading NB pork

Coeff. Constant

t

0.156296

other SB pork

Coeff.

0.886052

-1.11E-01

t

leading SB pork

Coeff.

-0.5369

-0.202231

t

Coeff.

-0.786066

2.07E-01

t 1.3311

LTE

-0.028697

***

-11.5558

-2.46E-02

***

-8.90423

-4.54E-02

***

-26.3273

-5.66E-02

***

-27.6331

Mills

0.018714

**

3.13178

0.05205

***

11.2485

-1.60E-01

***

-36.8213

-2.67E-01

***

-30.5681

Stores

-1.14E-03

-0.776652

5.18E-03

**

2.96945

-1.50E-03

-1.15357

4.29E-03

**

2.50998

AD

-3.90E-08

**

-2.58271

1.45E-08

1.38038

2.23E-08

0.568126

-1.25E-06

Adoth

2.67E-08

***

LP1NB0

-2.70E-01

**

4.7486

-9.92E-10

-0.374554

-4.68E-09

-1.35176

-8.61E-09

-2.21283

2.67E-02

0.417099

3.21E-02

0.482369

1.00E-01

LP1NB1

-5.13E-02

-0.986056

0.171559

1.14303

-1.33E-01

**

-2.15901

-2.67E-01

LP1SB0

0.117558

0.774746

-4.51E-02

-0.356042

0.561709

**

2.3419

2.11E-01

LP1SB1

-3.50E-01

LP3NB0

8.60E-03

LP3NB1

0.125056

LP3SB0

**

-2.88655 1.34645

*** *

-5.75712 1.70071

-4.23129

-5.31E-02

-0.876525

-3.13E-02

-0.579907

0.051161

0.928665

0.740099

1.65E-02

1.01137

-2.96E-02

-1.31603

-0.028107

-1.44159

3.24359

-7.20E-02

*

-1.95241

1.33E-02

0.266879

0.028443

0.522514

-1.74E-02

-1.18557

0.036626

**

2.3251

3.63E-02

**

2.74419

2.65E-02

1.40591

LP3SB1

-8.67E-03

-0.604301

4.69E-03

0.318124

0.033657

*

LP4NB0

-7.54E-03

-0.694985

-4.81E-03

-0.430792

-0.023366

LP4NB1

1.28E-02

0.943018

3.42E-02

2.06215

4.08E-02

LP4SB0

4.77E-02

*

1.80721

-1.18E-03

-0.043858

LP4SB1

-0.019836

*

-1.67962

0.015534

KID1

-1.53E-02

**

-2.75579

-0.012906

hhinc

-1.46E-07

-1.49811

hhsize

1.14E-02

4.89671

T

***

***

2.52E-03

**

1.83063

4.41E-02

-1.42115

7.65E-04

**

2.1528

**

2.99886

0.04359

**

2.23579

-8.04E-02

**

-1.99233

-0.100991

**

-1.99095

**

-2.40542

-0.029877

*

-1.73087

1.1399

-5.51E-03

0.047136

1.34809

-3.24E-02

-2.11271

5.30E-03

1.29E-07

1.16353

1.63E-07

**

2.42384

1.99E-07

-3.08E-03

-1.30334

-4.35E-03

**

-2.69671

-1.62E-03

-0.860263

**

-1.03906 **

2.36308

0.360814

2.79E-02

**

2.66131

0.019551

1.56947

-9.97E-05

-0.021873

-2.18713

-1.21E-02

**

-2.62213

-1.50E-03

-0.522537

-1.41E-03

-0.378512

2.18E-04

0.198692

-2.66E-03

*

-1.83185

7.41E-04

0.784209

2.27E-04

0.19374

9.24E-06

0.840515

3.33E-05

**

2.40311

-4.97E-06

-0.558632

-2.94E-06

-0.267114

urban

-9.21E-03

hage hages Variables

***

-1.37201

**

other NB poultry Coeff.

leading NB poultry t

Coeff.

other SB poultry t

Coeff.

leading SB poultry t

Coeff.

t

Constant

2.07E-01

1.14709

7.55E-01

***

3.33372

0.41254

**

2.30453

1.94E-01

LTE

2.17E-02

***

7.98362

-2.30E-02

***

-12.5858

-6.79E-03

**

-2.7959

0.085279

***

1.04857 22.7954

Mills

0.064153

***

14.4563

-4.75E-02

***

-5.64562

5.07E-02

***

10.0254

5.74E-02

***

13.263

Stores

1.90E-03

0.963497

4.81E-03

***

3.57003

5.00E-03

**

3.06896

-1.46E-02

***

-6.47299

94

AD

4.03E-09

0.424116

2.47E-09

0.926036

-1.05E-10

-9.14E-03

7.55E-08

*

1.7485

-9.62E-10

-0.323369

-1.03E-08

***

-3.80005

4.54E-09

1.5275

1.30E-08

**

3.16308

LP1NB0

8.24E-02

1.46999

1.38E-01

**

3.16427

-0.047786

-1.06034

-1.62E-01

**

-2.04221

LP1NB1

0.026875

0.425874

-0.118916

**

-2.10996

9.38E-02

1.79407

-1.26E-01

*

-1.73778

LP1SB0

-1.38E-01

-1.16026

-9.14E-02

LP1SB1

5.63E-02

0.953269

0.060486

LP3NB0

-1.57E-01

-2.87579

2.58E-02

LP3NB1

0.022586

0.410672

-4.54E-01

Adoth

**

* ***

-0.726391

0.17558

1.26974

-3.02E-01

**

-2.61556

1.74779

-5.40E-02

-1.08298

-0.169112

**

-2.01175

1.63127

1.00E-02

0.527297

0.078315

**

2.84673

-3.35136

-1.33E-01

**

-2.95264

1.45E-01

**

2.39224

***

LP3SB0

0.028583

1.296

-0.015355

-1.03092

-1.67E-01

LP3SB1

-2.84E-04

-0.013568

-7.19E-04

-0.044516

2.93E-03

LP4NB0

-3.74E-02

-2.473

-8.46E-03

-0.967033

-0.02379

LP4NB1

-1.59E-02

-0.777522

8.58E-03

0.666689

LP4SB0

-0.016543

-0.446913

-5.00E-02

LP4SB1

-7.41E-04

-0.051926

-8.16E-03

KID1

8.39E-03

1.05232

7.65E-05

0.015882

-7.27E-04

hhinc

2.52E-08

0.257223

-1.11E-07

-1.56271

-2.93E-07

hhsize

**

1.44E-02

0.559181

-7.07E-03

-0.143768

**

-1.97645

8.02E-03

0.4562

-0.069944

***

-3.82497

5.11E-02

**

-1.66089

-1.17E-01

**

-2.71454

0.161895

***

3.40774

-0.836221

2.56E-02

**

2.8556

-1.19E-01

***

-5.02421

-0.107259

3.57E-03

-3.13646

2.63E-07

**

2.03012

0.332447 *

1.92552

1.01648

3.10E-03

1.79789

-1.66E-03

-0.700944

3.12E-03

**

-2.02325

-2.79E-02

***

-3.35665

2.84E-03

0.458417

-0.033387

urban

1.96E-02

***

4.75842

-0.010748

***

-3.30461

-3.05E-03

-0.758334

-2.85E-03

hage

1.84E-03

1.04192

-1.44E-03

-1.05922

-1.34E-03

-0.822587

5.49E-03

**

2.66911

hages

-3.18E-05

-1.998

9.18E-06

0.737236

-5.72E-07

-0.038775

-6.53E-05

***

-3.4885

Variables

**

other NB other meats Coeff.

Constant

1.79E-01

LTE

2.21E-02

Mills

7.30E-02

Stores

t

*

-3.44895 0.174608

-0.010136

T

2.66E-03

*

*

leading NB other meats Coeff.

0.849819

7.09E-02

***

5.8303

7.70E-03

***

17.6335

t

other SB other meats Coeff.

0.368926

0.090033

**

2.07536

-6.52E-03

6.34E-02

***

15.3432

***

-5.98143 -0.511675

leading SB other meats Coeff.

t

0.565486

-0.958398

***

-5.08049

***

-3.52342

0.054737

***

16.4427

0.016579

**

2.93222

0.0785

***

18.0504

3.74748

-2.71E-03

**

-2.14878

-0.011935

***

-1.0478

3.08E-08

0.760206

1.19E-06

1.13176

-4.48E-09

-3.87504

0.14358

*** **

1.38E-03

0.523638

9.37E-03

AD

-1.75E-08

-1.41404

-3.45E-08

Adoth

-9.12E-09

-2.77559

-7.59E-09

**

-1.97362

2.57E-09

**

2.35046

-0.159734

**

t

0.867383 ***

*

-1.72503

LP1NB0

1.28E-01

1.51079

0.195965

LP1NB1

-4.48E-02

-0.637609

-0.041057

-0.58852

-0.077522

-1.38543

0.140251

LP1SB0

-3.78E-02

-0.309324

0.067172

0.514654

0.023639

0.160373

0.095697

1.05704

LP1SB1

1.17E-01

1.79288

0.159074

2.35628

-0.053981

-1.4478

0.045682

0.310547

LP3NB0

-1.94E-02

-0.719337

-7.03E-03

-0.270001

-0.012172

-0.88022

0.136917

**

LP3NB1

4.55E-03

7.07E-02

3.48E-03

0.057505

0.021589

0.607809

0.037478

*

1.7834

LP3SB0

0.032378

1.30656

-0.05079

-1.87145

0.036265

3.04358

0.081417

**

2.11547

LP3SB1

-9.87E-02

-3.65982

0.02194

0.857647

-0.012917

-1.2765

0.013481

LP4NB0

5.39E-02

1.25014

-4.61E-03

-0.26029

4.67E-03

0.531255

0.093319

-2.06196

-0.090344

-1.80788

-7.56E-03

-0.728872

0.019905

-1.23263

0.068791

1.62974

0.164313

*

1.85723

0.034572

*

1.87223

*

1.66999

*

*** **

**

*

*

***

-6.4501 1.30863

**

LP4NB1

-5.53E-02

LP4SB0

-0.065658

LP4SB1

4.62E-02

**

2.53003

0.011245

0.629645

-0.016098

*

-1.82702

0.062291

KID1

1.72E-02

*

1.81E+00

9.12E-04

0.104611

-0.011707

**

-2.86966

0.010683

hhinc

-1.89E-07

-1.26102

-1.26E-06

-8.18267

2.21E-07

**

3.12845

1.00E-06

***

3.43034 2.0124

1.98271

0.762626 ***

3.64847 1.29383

1.32281 ***

95

8.12903

hhsize

6.86E-03

1.96949

-3.08E-04

-0.089183

-8.56E-04

-0.564419

-0.015297

T

6.97E-03

**

1.22645

3.98E-03

0.696324

-1.37E-03

-0.236554

9.14E-03

1.438

urban

2.40E-03

0.383786

5.19E-03

0.842611

6.00E-03

2.12371

7.69E-03

1.59093

hage

-1.61E-03

-0.809765

-5.58E-03

**

-3.00747

-1.16E-03

-0.963125

5.28E-03

**

2.97875

hages

2.31E-05

1.22042

6.75E-05

***

3.79092

1.08E-05

0.979924

-4.77E-05

**

-2.91179

**

***

-5.54968

TABLE 6.5 Second-Step Working- Leser Model Estimates for Alberta Variables

other NB pork

leading NB pork

Coeff. Constant

t

Coeff.

t

1.34232

0.559658

LTE

-0.018325

***

-3.97456

-9.13E-03

*

Mills

0.071523

***

9.69711

0.052721

***

Stores

2.74E-03

0.7633

1.60E-03

-9.12E-09

-0.426896

-2.00E-08

9.29E-10

0.152556

-4.29E-09

LP1NB0

-0.021544

-0.217877

LP1NB1

-0.084029

LP1SB0 LP1SB1

leading SB pork

Coeff.

t

Coeff.

t

0.741768

-0.56249

-0.824574

-0.054502

-1.86932

-0.092911

***

-16.6405

-0.064616

***

-16.0256

7.22627

-0.352478

***

-23.2035

-0.18476

***

-30.4135

0.473588

0.014118

***

3.61448

5.94E-03

**

2.10349

-0.99817

8.14E-09

0.131748

5.15E-06

*

-1.77287

1.88E-09

0.399104

-2.52E-09

-1.24652

0.018755

0.448584

8.38E-03

0.135165

0.027918

0.746685

-0.991161

-0.205058

-1.0701

-0.130394

-1.27595

-0.183048

-0.62313

-0.551188

-0.928064

-0.901098

-0.431081

-0.59296

-0.290374

-0.305317

-0.51597

-0.563093

0.739631

0.661444

1.15669

1.12245

0.670415

0.891583

LP3NB0

6.43E-03

0.438556

0.011418

0.882269

-0.018981

-0.772841

-5.02E-03

-0.320257

LP3NB1

3.52E-03

0.375289

0.019981

2.02369

8.41E-03

0.486331

0.011745

0.919566

LP3SB0

-2.24E-03

-0.103552

-0.020189

-0.703032

0.072319

2.63138

0.03508

LP3SB1

0.013189

0.548806

0.018882

1.51534

0.034674

0.707669

-2.62E-03

-0.102538

LP4NB0

-4.20E-03

-0.252298

1.66E-03

0.111602

0.011419

0.62919

0.014514

1.2516

LP4NB1

-0.020418

-0.810564

-0.031973

-1.50516

0.106679

3.5168

8.94E-03

0.394482

LP4SB0

0.012251

0.264015

0.017615

LP4SB1

0.012163

0.508719

0.040888

-3.41149 -0.69515

AD Adoth

1.28003

other SB pork

KID1

-0.038666

hhinc

-1.06E-07

hhsize

0.016241

***

*

1.78192

0.669534

0.076176

1.21087

-0.013696

-0.373873

2.59348

-0.172509

**

-2.07235

-0.055532

-1.54403

3.20E-03

0.381543

-0.032084

**

-3.17077

-3.45E-03

-0.454866

1.30E-07

0.952741

-1.18E-07

-0.783922

7.37E-08

0.688618

**

-8.05E-03

-2.62699

9.18E-03

2.4568

3.94E-03

1.44287

-0.010267

-0.40903

0.033828

1.43604

9.08E-03

0.456529

***

-6.63482

8.90E-03

1.59492

-4.58E-03

-0.716492

4.76E-03

1.07782

-5.13E-03

**

-2.52364

-4.77E-03

**

-2.58672

1.86E-03

1.28319

1.08E-03

0.97804

5.87E-05

**

2.99793

5.31E-05

**

2.98159

-1.68E-05

-1.13434

-8.78E-06

-0.805175

urban

-0.052367

hage hages

other NB poultry Coeff.

Constant

**

-2.70858

3.6176

-0.041338

***

**

**

1.79427

-1.22268

T

Variables

***

*

-0.094016

0.686861

**

leading NB poultry t

0.768059

Coeff. -0.527666

**

other SB poultry t

-1.13042

Coeff. -0.382855

leading SB poultry t

-0.448735

Coeff.

t

-0.530283

-0.788184

96

LTE

0.011254

*

1.78377

5.39E-03

Mills

0.055397

***

7.06554

0.019822

Stores

0.011434

**

2.64768

AD

5.35E-09

0.260958

Adoth

1.23828

0.083139

***

11.0568

7.03E-03

2.1286

0.081321

***

13.3048

0.016286

*

1.65752

-1.45E-03

-0.449554

-0.021776

***

-4.68378

-7.44E-03

**

-2.02351

-7.69E-10

-0.394145

-1.35E-08

-0.751601

-2.16E-09

**

4.13E-10

0.163463

5.09E-09

1.48545

4.49E-09

*

LP1NB0

-0.057707

-1.23462

-0.01344

-0.587392

0.116593

**

LP1NB1

0.027768

0.325639

0.184601

3.35765

LP1SB0

-1.13646

-1.85969

-1.03732

-1.27486

LP1SB1

0.322642

LP3NB0

-0.123994

LP3NB1 LP3SB0

*

***

-0.15837

1.908

4.18E-09

2.30692

0.021514

**

0.567457

2.14909

0.120422

1.08889

0.094508

1.38473

-0.353215

-0.42989

1.0039

0.838602

0.344612

1.43271

2.39494

1.13719

1.42734

0.146585

0.182125

-1.92965

3.45E-03

0.200754

-0.057973

-1.19773

-0.015428

-0.813113

-0.01221

-0.459261

-6.07E-03

-0.190496

-0.060319

-0.96845

-4.16E-03

-0.407403

-0.037782

-1.37028

0.020592

1.35923

-0.09483

-1.56566

0.040127

**

2.93135

*

LP3SB1

0.076521

3.04156

-6.15E-04

-0.04717

0.028793

0.955955

-0.215219

*

-1.75533

LP4NB0

-0.012875

-0.509705

-4.02E-03

-0.464776

-0.044727

**

-2.30618

-0.030243

**

-2.69831

LP4NB1

-6.88E-03

-0.25477

5.25E-03

0.295378

-0.097922

**

-3.07147

-0.020032

LP4SB0

8.77E-03

0.277807

6.04E-03

0.217464

-0.024116

-0.541763

0.068819

**

2.57828

LP4SB1

-0.015885

-0.464979

-0.017986

-0.847737

0.036131

0.991219

-0.211901

***

-4.30584

KID1

-0.028896

-2.24337

-5.93E-03

-0.707068

0.033803

**

2.80835

0.020719

**

2.06837

hhinc

-4.51E-08

-0.290258

-3.03E-07

-2.63888

-6.92E-07

***

-3.67887

3.65E-07

***

3.19629

hhsize

0.012741

2.83088

1.86E-03

0.633678

-5.93E-03

-1.31296

5.55E-04

T

**

**

1.42483

** **

-0.815066

0.026355

1.65208

0.013771

0.444177

0.031823

-0.02056

**

-2.84346

-1.37E-04

-0.028232

0.014087

**

1.97602

7.86E-03

*

1.68296

hage

7.98E-03

***

4.22537

-2.45E-04

-0.176594

-4.12E-03

*

-1.74424

-6.28E-03

**

-2.95352

hages

-8.52E-05

***

-4.63214

-1.14E-06

-0.088197

3.06E-05

1.35515

4.87E-05

**

2.56462

other NB other meats Coeff.

Constant

t

leading NB other meats Coeff.

t

other SB other meats Coeff.

t

Coeff.

-0.159711

-1.3449

-1.27449

-0.795542

-1.03778

2.82799

0.066183

***

8.48493

0.011603

*

1.64315

7.14E-03

1.5403

-6.76E-03

Mills

0.090854

***

15.6391

0.078915

***

13.6173

0.046411

6.53357

0.023989

Stores

-2.07E-03

-0.406299

2.71E-03

0.588412

-4.70E-03

-1.50391

-1.10E-03

***

1.44814

leading SB other meats

LTE

AD

-0.156294

*

0.162996

urban

Variables

-0.025298

**

-0.908903

t *

1.91429 -1.37189

**

2.35765 -0.316523

9.87E-09

0.494596

6.42E-08

1.51434

-1.96E-08

-0.107168

-5.17E-06

*

-1.88117

Adoth

-1.24E-09

-0.318153

4.18E-09

1.25503

3.63E-09

1.10853

-1.67E-08

*

-1.92308

LP1NB0

0.020894

0.366912

-0.062607

-1.3664

8.96E-03

0.305005

0.037239

LP1NB1

-0.136134

-1.34336

0.188058

2.06804

-0.113882

-1.58903

-0.0708

LP1SB0

1.12493

0.756751

2.59839

1.2283

3.65948

LP1SB1

-1.22245

-1.02532

-1.40773

-0.948252

-2.54112

LP3NB0

0.012668

0.489808

0.046931

2.29941

0.032557

LP3NB1

-0.027666

-0.432344

0.010173

0.728603

-9.69E-03

LP3SB0

-0.030977

-0.586749

0.052251

**

1.96212

-7.95E-03

-0.25778

0.011357

1.44487

LP3SB1

0.042904

1.34821

-0.060976

**

-2.1788

0.010099

0.472011

-0.017612

-0.616148

LP4NB0

0.175401

***

5.80239

-0.055411

**

-2.64249

-2.28E-03

-0.213344

0.080376

1.4647

LP4NB1

-0.086119

**

-2.14809

0.31157

***

4.82299

0.025882

1.20564

5.09E-04

0.055206

LP4SB0

8.64E-04

0.016089

-7.60E-03

-0.133533

-0.096824

-1.4377

-0.018356

-0.920359

**

**

** **

0.408692 **

-2.22687

2.09252

-3.32E-03

-0.054559

-1.20968

0.48323

0.531677

2.42827

-3.7209

-0.638882

-0.010461

**

-2.1514 -0.768404

97

LP4SB1

0.110919

**

2.57537

-0.012606

KID1

0.025532

*

1.76999

0.048547

hhinc

8.76E-09

0.039032

4.30E-08

hhsize

-0.011489

-2.11547

-0.01855

**

*** ***

-0.326468

0.018004

0.794163

0.070575

3.62426

-0.019784

**

-2.04856

-2.99E-03

**

2.54291

0.207515

2.65E-07

**

2.35333

3.79E-07

-3.85427

4.00E-03

1.08626

-4.50E-03

0.823592

0.023588

1.08819

-0.097841

-3.82502

7.63E-03

1.48077

6.14E-03

1.29001

-0.31916 **

3.14547 -1.30744

T

7.40E-03

0.225631

0.028902

urban

0.062721

***

7.27548

-0.034438

hage

6.81E-03

**

2.78231

8.43E-04

0.363065

4.83E-04

0.365876

1.48E-03

1.15394

hages

-6.53E-05

**

-2.73736

8.64E-06

0.374081

-3.93E-06

-0.299018

-1.85E-05

-1.46706

***

**

The estimated results for Ontario and Alberta in the second stage model which establish the impact of demographic and other characteristics on the level of spending for pork, poultry and other meats classified by leading national brand, other national brands, leading store brand, other store brands also highlights differences between the provinces. For one variable, the number of stores visited by each household, there is a strong positive relationship between number of stores and level of spending on the leading national brand of pork poultry and other meats in Ontario. The same is not true in Alberta. This suggests that in Ontario the shopper is more ‗loyal‘ to the leading national brand regardless of store choice. In Ontario higher income levels are associated with higher expenditures for leading national and other store brands for all pork, poultry and other meats with the exception of other store brand poultry products. In Alberta higher incomes are only associated with higher spending on leading store brand poultry and other meats and other store brands of other meats. Over time there is a positive increase in sales of the leading national brand of pork but declining sales of the leading national brand of poultry products in Ontario. In Alberta, there is only a small positive increase in the leading national brand of poultry that is statistically significant out of all of the twelve types of product. In Ontario households with older heads have lower expenditures on the leading national brand for pork and other meats and higher expenditures on the leading store brands for poultry and other meats. In Alberta, households with older heads have lower expenditures on the leading national brand and other national brands of pork, higher expenditures on other national brands of poultry but lower expenditures on leading and other store brands of poultry and higher expenditures on other national brands of other meats.

98

-1.99229

Summary and Conclusions The overall objective of the study is to look at the structure of consumer value added meat purchasing behaviour (value added meat type choices, store choices as well as brand choices) in order to improve the understanding of recent food-at-home consumption patterns and discern new trends in value-added meat demand. Specifically the research objectives for the study are threefold: 1. Using household level meat purchase data over the period 2002-2007 in order to: a. Understand how consumers make purchase decisions around fresh, semiprocessed and fully processed products for four meat type categories: beef, pork, poultry and others (fish, lamb, etc) b. Quantify the impact of demographic and regional characteristics differences on meat consumption behaviour, and these differences in the behaviour across meat types.

2. Using household level meat purchase data from 2002-2007 and store level advertising data(1999-2006) in order to: a. Find out whether Canadian consumers show consistency in meat purchasing patterns by store. Are they loyal to particular stores? Does this vary by region, by demographics, by store availability, is store advertising a factor?

3. Use household level purchase data from 2002-2008 and Nielsen Media Measurement's advertising data(2000-2008) in order to: a. Identify how consumers make the decisions about private label versus national brand products in their fully processed value-added meat category. Is product and brand advertising a factor? Does behaviour vary regionally and by demographics?

99

The aim of all of these individual analyses is to determine whether or not there are characteristics of meat purchasing – by animal species, by level of processing, by store and by branding which could enhance understanding of the potential success of value adding strategies. Future value-added meat product development might be enhanced by understanding whether there are significant differences across any of these descriptors. The analysis was conducted for two subsets of the national Nielsen Homescan™ panel meat purchasing data . First of all households were selected with the aim of having as long a purchase history as possible – allowing the analysis of habit formation and trend as significant determinants of household purchasing behaviour. For the existing data this resulted in selecting households who were part of the Nielsen panel over the period 2002 to 2007/2008. As well rather than analyze the entire national panel, households who were from Ontario and Alberta were selected for further analysis. This resulted in the reduction of the panel to maneageable numbers for analysis and allowed the comparison of two very different regions within the country. These two regions were of interest due to the size of Ontario (largest provincial population) and the fact that Alberta is so significant in livestock production but has not traditionally been as significant in value –added meat processing. Summary

Using a relatively arbitrary method of describing individual meat products, meats divided into four major types (pork, beef, poultry and other) were further divided into three main levels of processing. The first and largest category is fresh meat purchases (on every measure the majority of meat purchased through grocery stores by Canadians continues to be in fresh form) ranging from approximately 70% of meat expenditures in Ontario to 75% plus of total meat expenditures in Alberta. Semi-processed meats were classified as those to which some level of further processing had been applied (sauces, flavourings, for example) but for which cooking would still be required by the purchaser. In Ontario this category represents 11% of meat ependitures while in Alberta it only represents 6% of meat expenditures on average. The final category was classified as fully processed which in some cases means no further cooking is required (ham, for example) but in other cases implies that the product has had more than one type of processing applied (breaded formed chicken nuggets, for example)

100

although cooking is still required. Meal type items would be included in fully processed. In Ontario fully processed meats make up over 20% of meat expenditures while in Alberta they average 19% of meat expenditures. By animal species, pork expenditures range from 20 – 25% in Alberta but 20-22% in Ontario over the period 2002 – 2007. Beef remains the dominant meat ranging from 32-38% in Alberta and from 30-33% of total meat expenditure in Ontario. Poultry expenditures range from 29-32% in Alberta over the period 2002-2007 while in Ontario they level of expenditure is more consistently 33-34% over the same period. In each province semi and fully processed beef expenditures are the smallest of the twelve meat types, reflecting the lower number of semi and fully processed beef products available in the market. In the final analysis of this report – the comparison between national branded and store branded products beef was excluded as a category due to the infrequency of purchases by households in Ontario and Alberta. Three models are reported in this study – in each case the models are represented by a two stage structure. In the first stage (of each of the three models) the probability that a household makes a purchase decision (model 1 – to purchase a particular one of twelve types of meat including fresh, semi-processed and fully-processed beef, pork, poultry and other meat, model 2- to purchase meat at a particular grocery store chain, model 3 to purchase national brand or private label brand pork, poultry or other meat) is modelled as a function of demographic variables using a probit model. In the second stage expenditure shares are modelled as functions of demographic variables, trend, habit formation, where possible average market prices and advertising expenditures and the inverse Mills ratio from the first stage of the model. The results suggest indicators of the actual decision to purchase as distinct from the factors affecting the levels of expenditure on meat types in each model.

Consumer Meat Behaviour and Level of Processing

Estimates can be summarized in terms of sign and significance across the two decisions that are modelled. The first decision which is portrayed below is for the decision of whether or not to purchase each of the twelve fresh, semi-processed and fully processed meat products. In general, household headed by an older person are more likely to purchase all types of pork and fresh and

101

fully processed beef but less likely to purchase semi and fully processed poultry products. Higher levels of education are associated with higher probabilities of purchasing pork products in Ontario and poultry products in both provinces. As household sizes increase there is a greater probability of purchasing semi and fully processed meat products. Over time higher levels of processing have a higher probability of being selected.

102

Model 1 – Consumer Behaviour and Level of Processing – first stage decision Variable

Pork

Pork

Pork

Beef

Beef

Beef

Poultry

Poultry

Poultry

Other

Other

Other

Fresh

Semi

Fully

Fresh

Semi

Fully

Fresh

Semi

Fully

Fresh

Semi

Fully

ON AB ON AB ON AB ON AB ON AB ON AB ON

AB

ON

AB

ON

AB

ON

AB

ON AB ON

AB

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

age

+

+

+

+

+

+

+

+

-

+

-

+

+

educ

+

+

+

+

+

+

+

+

-

+

urban

-

-

-

-

Meat

+

+

+

+

+

+

+

+

+

+

-

+

+

Exp

hsize time

-

+ -

+

+ +

+

+

+

+

-

+ -

-

+

-

+

-

+

-

-

+

+

-

+

+

+

+

+

+

+

-

+

+

+

+

+

+

103

+

+

+

+

In terms of factors which explain the level of expenditure on each of the twelve meat types the the consistent explanators appear to be the household size and/or having children in the household (in both provinces). Price responses, when statistically significant, suggest inelastic demands for most of the twelve meat types in both provinces (a 1% decrease (increase) in price results in a less than one percent increase (decrease ) in quantity sold). Advertising effects do not appear to be significant across the twelve meat types in explaining the level of meat expenditure. Households with children are likely to spend less on semi and fully processed pork and beef but likely to spend more on fully processed poultry products. In Ontario households with higher levels of income are likely to spend less on all types of pork, on fresh and semi-processed beef but more on all types of poultry and fresh and semi-processed other meat products. In Alberta the effects of income are negative for fresh pork and beef but positive for all types of poultry products.

Consumer Meat Behaviour and Store Selection

It is worth stating that the vast majority of households do not choose to purchase their meat regularly at the same grocery store. Most households in the Nielsen panel purchase meat at more than one store and can purchase meat at up to 5 stores on a somewhat regular basis. In Ontario, households who spend more on meat have a higher probability of shopping at Empire, Loblaws, Metro and less likely to shop at ‗other‘ grocery stores. Higher incomes, older household head age and larger household size result in a higher probability of shopping at a Loblaws store (including all stores owned by Loblaws). Over time more households are choosing to purchase meat at Metro, Empire and other grocery stores, in Ontario. Households in urban areas have a lower probability of shopping at Empire, Loblaws but a higher probability of shopping at Metro and other grocery stores. In comparison, in Alberta, households who spend more on meat are more likely to spend that money at Co-op, Empire, Loblaws and Safeway grocery stores (reflecting the increasing concentration in the grocery retailing industry in Canada). Urban dwellers are less likely to buy meat at Co-op, Empire, Loblaws but more likely to make meat purchases at JPG and Safeway. Larger household size suggests an increased probability of purchasing meat at Co-op, Empire, Loblaws and other grocery stores. In Alberta higher income households are more likely to make meat purchases

104

at Loblaws, JPG and other grocery stores and less likely to make meat purchases at Co-op stores. Over time probability of purchasing meat is growing at Empire, Loblaws and other grocery stores and declining at Safeway stores. The significant explanatory variables for the decision on level of spending on meat at each of the grocery store chains are also variable across provinces. The number of grocery store chains shopped at by a household is a significant determinant of level of meat spending at all chains except Loblaws in Ontario. In Alberta the number of chains shopped at is positively related to the level of spending at Empire, JPG and other stores but is negatively related to the level of spending at Safeway and Loblaws brand stores. In Alberta, households with higher levels of education spend more on meat at Co-op, Loblaws and JPG and less at Empire and Safeway. In Ontario households with higher levels of education spend more on meat at Metro and less at Empire. Households with larger sizes spend more on meat at Loblaws and other stores in Ontario and at Co-op, Loblaws and others in Alberta. Store advertising has no significant effects in Alberta but has small positive effects on meat spending for Co-op and Safeway stores in Ontario – both non-traditional and small retailers in Ontario. Consumer Behaviour and Choice of National Brand versus Private Label Meat Products

Results from the first stage of the national versus store brand model suggest that the decision to purchase any of the four (leading national brand, other national brands, leading store brand, other store brands) branded products is significantly affected by demographic characteristics in both Ontario and Alberta. In Ontario higher levels of household income are associated with higher probabilities of purchasing leading national and store brands for pork, poultry and other meats and other store brands of pork and poultry. In Alberta, higher incomes are associated with higher probabilities of purchasing leading store brands for pork, poultry and other meats but reduced probabilities of purchasing other store brands of pork and poultry and leading national brands of poultry (possibly reflecting the regional importance of Lilydale as a poultry processor in Alberta). There are also differences in the trends by meat type – for example over time there is a higher probability to purchase all four brands of pork and other meats in Ontario but opposite signs for poultry products. In Alberta the trend variables over time suggest positive signs on the

105

probability of purchasing leading national brand, leading store brand and other store brands for pork but negative signs for leading store brand and other store brands for poultry.

The estimated results for Ontario and Alberta in the second stage model which establish the impact of demographic and other characteristics on the level of spending for pork, poultry and other meats classified by leading national brand, other national brands, leading store brand, other store brands also highlights differences between the provinces. There is a strong positive relationship between number of stores and level of spending on the leading national brand of pork poultry and other meats in Ontario. The same is not true in Alberta. This suggests that in Ontario the shopper is more ‗loyal‘ to the leading national brand regardless of store choice. In Ontario higher income levels are associated with higher expenditures for leading national and other store brands for all pork, poultry and other meats with the exception of other store brand poultry products. In Alberta higher incomes are only associated with higher spending on leading store brand poultry and other meats and other store brands of other meats. Over time there is a positive increase in sales of the leading national brand of pork but declining sales of the leading national brand of poultry products in Ontario. In Alberta, there is only a small positive increase in the leading national brand of poultry that is statistically significant out of all of the twelve types of product. In Ontario households with older heads have lower expenditures on the leading national brand for pork and other meats and higher expenditures on the leading store brands for poultry and other meats. In Alberta, households with older heads have lower expenditures on the leading national brand and other national brands of pork, higher expenditures on other national brands of poultry but lower expenditures on leading and other store brands of poultry and higher expenditures on other national brands of other meats.

106

Conclusion

At a household level there is significant variability in the markets for meat products, by species and by level of processing. There are significant demographic differences in household purchases of meat by grocery store chain across provinces. By products from different animal specieis the market for fully processed meat products is also variable across provinces and by demographic characteristic. It is clear from the results presented that there is no one correct pattern of value added meat product development across animal products from different species. To a certain extent the results presented are generated by the products available in the marketplace. There are clearly much higher numbers of pork and poultry semi-processed and fully processed products available than there are for beef. However the types of further processed products available in the pork and poultry areas are different either in their nature or in their uptake by consuming households. For example, in certain models households with children were less likely to purchase fully processed pork and beef but more likely to purchase poultry fully processed products. Grocery store meat purchases exhibit little store loyalty – most households purchase meat at more than one store. In terms of meat product development the ability to reach a significant number of Canadian consumers is thus attached to the necessity to market through more than one grocery chain. Loblaws is one store chain with national reach that seems to be attractive to certain demographics – with older household heads, with higher incomes and larger household sizes in Ontario, for example. The determinants of meat spending at grocery stores in Alberta is more evenly divided across Co-op, Safeway, Empire and Loblaws, possibly due to traditional store availability in Canada. This significantly increases the logistical difficulties of developing new value-added meat products and delivering them to consumers in Canada. Consumers also differ considerably in their interest in and level of spending on national brand and private label products. For some meat products store brand or private label products seem to be expanding in household preferences while in others they seem to be contracting – these results seem to be animal specific or firm specific since there are relatively few processors for each animal species within Canada. Higher income households seem to prefer both national and store brands of meat products in both provinces. An interesting result in Ontario is the result that households who purchase meat at more stores seem to have higher expenditures on national

107

brands of pork, poultry and other meats, implying that even if they don‘t have store loyalty they may have national brand loyalty.

108

References Agriculture and Agri-Food Canada. Consumer Perceptions of Food Safety and Quality: A Closer Look at the 2006 Results 2007. Available from http://www.aic.ca/conferences/pdf/2007/Presentations/Diane_Jebson.pdf. ———. 2009. Canadian Consumers 2008 [cited Feb. 10th 2009]. Available from http://www4.agr.gc.ca/AAFC-AAC/displayficher.do?id=1170942402619&lang=eng. ———. 2009. Healthy Lifestyles 2008 [cited March, 10th 2009]. Available from http://www4.agr.gc.ca/AAFC-AAC/displayafficher.do?id=1201555501204&lang=eng. ———. 2009. The Canadian Poultry Meat Processing Industry 2009 [cited March, 19th 2009]. Available from http://www4.agr.gc.ca/AAFC-AAC/displayafficher.do?id=1172156161151&lang=eng. ———. 2009. Fact sheet - All about Canada's red meat industry 2009 [cited March, 15th 2009]. Available from http://www4.agr.gc.ca/AAFC-AAC/displayafficher.do?id=1184009758250&lang=eng. ———. 2009. Overview of the Canadian Agriculture and Agri-Food System 2008 2009 [cited March, 15th 2009]. Available from http://www4.agr.gc.ca/AAFCAAC/display-afficher.do?id=1228246364385&lang=eng. Ailawadi, K. L., K. Pauwels, and Jbem Steenkamp. 2008. Private-Label Use and Store Loyalty. Journal of Marketing, Volume 72, Number 6, November 2008 Alston, Julian M., and James A. Chalfant. 1991. Can We Take The Con Out Of Meat Demand Studies? Western Journal of Agricultural Economics 16 (01). Alston, Julian M., James A. Chalfant, and Nicholas E. Piggott. 2000. The Incidence of the Costs and Benefits of Generic Advertising. American Journal of Agricultural Economics 82 (3):665-671. Amrouche, N., G. Martn-Herrn, and G. Zaccour. 2008. Pricing and Advertising of Private and National Brands in a Dynamic Marketing Channel. In Journal of Optimization Theory & Applications., Volume 137, No. 3, page 465-483. Anders, Sven, and Anke Moeser. 2008. Assessing the demand for value-based organic meats in Canada: a combined retail and household scanner-data approach. International Journal of Consumer Studies 32 (5):457-469. Atkins, F. J., W. A. Kerr, and D. B. McGivern. 1989. A Note on Structural Change in Canadian Beef Demand. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 37 (3):513-524.

109

Baltagi, Badi. 2008. Econometric Analysis of Panel Data, 4th Edition: Syracuse University. Beaumont, John R. 1988. Store Location Analysis: Problems and Progress. In Store choice, store location and market analysis, edited by N. Wrigley: New York and London: Routledge, Chapman and Hall; Routledge. Besanko, D., J. P. Dube, and S. Gupta. 2003. Competitive price discrimination strategies in a vertical channel using aggregate retail data. Management Science 49 (9):1121-1138. Biesalski, H. K. 2002. Meat and cancer: meat as a component of a healthy diet. European Journal of Clinical Nutrition. 56 Suppl(1), S2-11. Boetel, Brenda L., and Donald J. Liu. 2003. Evaluating the effect of generic advertising and food health information within a meat demand system. Agribusiness 19 (3):345-354. Bonfrer, Andre, and Pradeep K. Chintagunta. 2004. Store Brands: Who Buys Them and What Happens to Retail Prices When They Are Introduced? Review of Industrial Organization 24 (2):195-218. Bontemps, C., V. Orozco, and V. Requillart. 2008. Private Labels, National Brands and Food Prices. Review of Industrial Organization 33 (1):1-22. Boserup, E. 1989. Agricultural growth and population change. In Agricultural growth and population change. Bowler, I. 1985. Some consequences of the industrialization of agriculture in the European Community In Healey, M. and Ilbery, B. (eds) The Industrialization of the Countryside. GeoBooks, Norwich, pp.75-98 Brucks, Merrie. 1988. Search Monitor: An Approach for Computer-Controlled Experiments Involving Consumer Information Search. The Journal of Consumer Research 15 (1):117-121. Carrboro Farmers Markets, Inc. 2009. Meat Guidelines 2007 [cited April, 15th 2009]. Available from . Chalfant, James A., Richard S. Gray, and Kenneth J. White. 1991. Evaluating Prior Beliefs in a Demand System: The Case of Meat Demand in Canada. American Journal of Agricultural Economics 73 (2):476-490.

110

Chen, P. Y., and M. M. Veeman. 1991. An Almost Ideal Demand System Analysis for Meats with Habit Formation and Structural Change. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 39 (2):223-235. Chernev, Alexander. 2006. Decision Focus and Consumer Choice among Assortments. Journal of Consumer Research 33 (1):50-59. Chiang, Jeongwen. 1991. A Simultaneous Approach to the Whether, What and How Much to Buy Questions. Marketing Science 10 (4):297-315. Chintagunta, Pradeep K. 1993. Investigating Purchase Incidence, Brand Choice and Purchase Quantity Decisions of Households. Marketing Science 12 (2):184-208. Chioveanu, Ioana. 2008. Advertising, brand loyalty and pricing. In Games & Economic Behavior. Elsevier, Vol. 64(1), pages 68-80 Connor, John M., and Everett B. Peterson. 1992. Market-Structure Determinants of National Brand-Private Label Price Differences of Manufactured Food Products. The Journal of Industrial Economics 40 (2):157-171. Cozzarin, Brian, and Ellen Goddard. 1992. A preliminary look at advertising beef, pork, chicken, turkey, eggs, milk, butter, cheese and margarine in Canada, Commodity Advertising and Promotion, H. Kinnucan, S. Thompson H. Chang. Ames , Iowa : Iowa State University Press. de Jonge, Janneke, Hans van Trijp, Ellen Goddard, and Lynn Frewer. 2008. Consumer confidence in the safety of food in Canada and the Netherlands: The validation of a generic framework. Food Quality and Preference 19 (5):439-451 . Degeratu, Alexandru M., Arvind Rangaswamy, and Jianan Wu. 2000. Consumer choice behavior in online and traditional supermarkets: The effects of brand name, price, and other search attributes. International Journal of Research in Marketing 17 (1):55-78. Doyle, M. P., and M. C. Erickson. 2006. Emerging microbiological food safety issues related to meat. Meat Science 74 (1):98-112. Eacute, Bustos-Reyes sar Augusto, Oacute, Gonz scar, aacute, and Benito lez. 2008. Store and store format loyalty measures based on budget allocation., Journal of Business Research, Vol 61 (9), page 1015-1025. Eales, James. 1996. A Symmetric Approach To Canadian Meat Demand Estimation. Journal of Agricultural and Resource Economics 21 (02).

111

Enneking, U., C. Neumann, and S. Henneberg. 2007. How important intrinsic and extrinsic product attributes affect purchase decision. Food Quality and Preference 18 (1):133-138. Erdem, T., M. P. Keane, and B. Sun. 2008. A Dynamic Model of Brand Choice When Price and Advertising Signal Product Quality. Marketing Science 27 (6):11111125. Esbjerg, Lars, and Tino Bech-Larsen. 2009. The brand architecture of grocery retailers: Setting material and symbolic boundaries for consumer choice. Journal of Retailing and Consumer Services In Press, Corrected Proof. Fok, D., C. Horvath, R. Paap, and P. H. Franses. 2006. A hierarchical Bayes error correction model to explain dynamic effects of price changes. Journal of Marketing Research 43 (3):443-461. Freebairn, J., E. Goddard and G. Griffith. 2004. Generic Promotion in the Food Marketing Chain. A Report for the Rural Industries Research and Development Corporation No. 04, 35 p. Gaquez-Abad, Juan C., and Manuel Sachez-Perez. 2009. Factors influencing olive oil brand choice in Spain: an empirical analysis using scanner data. Agribusiness 25 (1):36-55. Geeroms, Nele, Wim Verbeke, and Patrick Van Kenhove. 2008. Consumers' healthrelated motive orientations and ready meal consumption behaviour. Appetite 51 (3):704-712. Goddard, E. G. Griffith. 1992. The Impact of Advertising on Meat Consumption in Australia and Canada. Series 2/92. Sydney : New South Wales Department of Agriculture, Economic Services Unit. Gonzalez-Benito, O., M. P. Martinez-Ruiz, and A. Molla-Descals. 2009. Spatial mapping of price competition using logit-type market share models and store-level scannerdata. Journal of the Operational Research Society 60 (1):52-62. Guadagni, P. M., and J. D. C. Little. 1998. When and what to buy: A nested logit model of coffee purchase. Journal of Forecasting 17 (3-4):303-326. Hailu, Getu, Andreas Boecker, Spencer Henson, and John Cranfield. 2009. Consumer valuation of functional foods and nutraceuticals in Canada. A conjoint study using probiotics. Appetite 52 (2):257-265. Halford, J. C. G., E. J. Boyland, G. Hughes, L. P. Oliveira, and T. M. Dovey. 2007. Beyond-brand effect of television (TV) food advertisements/commercials on

112

caloric intake and food choice of 5-7-year-old children [electronic resource]. Appetite 49:263-267. Hansen, Karsten, Vishal Singh, and Pradeep Chintagunta. 2006. Understanding StoreBrand Purchase Behavior across Categories. Marketing Science 25 (1):75-90. Hassan, D., and S. Monier-Dilhan. 2006. National brands and store brands: competition through public quality labels [electronic resource]. Agribusiness : an international journal 22:21-30. Hassan, Zuhair A., and S. R. Johnson. 1983. Quarterly Demands for Meat in Canada with Alternative Seasonality Hypotheses. Canadian Journal of Agricultural Economics 31 (1):77-94. Hassan, Zuhair A., and Stanley R. Johnson. 1979. The Demand for Meat in Canada: an Application of the Transformation of the Variables. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 27 (3):1-12. Hassan, Zuhair A., and L. Katz. 1975. The Demand for Meat in Canada. Canadian Journal of Agricultural Economics 23 (1):53-63. Hinloopen, Jeroen, and Stephen Martin. 1997. Market-Structure Determinants of National Brand-Private Label Price Differences of Manufactured Food Products: Comment. The Journal of Industrial Economics 45 (2):219-223. Holm, Lotte, and Helle Kildevang. 1996. Consumers' Views on Food Quality. A Qualitative Interview Study. Appetite 27 (1):1-14. Howell, Debbie. 2004. NATIONAL BRANDS FEEL THE PRESSURE. DSN Retailing Today 43 (20):22. Hsiao, Cheng. 2003. Analysis of panel data: Second Edition: Cambridge University Press. Huang, C. L., and J. Fu. 1993. Consumer preferences and evaluations of a processed meat product. Journal of food distribution research 24 (1):149-157. Hui, Sam K., Eric T. Bradlow, and Peter S. Fader. 2009. Testing Behavioral Hypotheses Using an Integrated Model of Grocery Store Shopping Path and Purchase Behavior. Journal of Consumer Research April 8, 2009. Juhl, H. J., L. Esbjerg, K. G. Grunert, T. Bech-Larsen, and K. Bruns. 2006. The fight between store brands and national brands--What's the score? Journal of Retailing and Consumer Services 13 (5):331-338. Kahn, Barbara E., and David C. Schmittlein. 1989. Shopping trip behavior: An empirical investigation. Marketing Letters 1 (1):55-69.

113

Kamakura, Wagner A., Byung-Do Kim, and Jonathan Lee. 1996. Modeling Preference and Structural Heterogeneity in Consumer Choice. Marketing Science 15 (2):152172. Kinsey, Jean, Ben Senauer, and Yvonne Jonk. 1993. Desirable Attributes For Value Added Meat Products Survey 1993. University of Minnesota, Center for International Food and Agricultural Policy. Konishi, Hideo. 2005. Concentration of Competing Retail Stores. Journal of Urban Economics 58 (3):488-512. Kulshreshtha, S. N., and A. G. Wilson. 1972. An Open Econometric Model of the Canadian Beef Cattle Sector. American Journal of Agricultural Economics 54 (1):84-91.

Lambert, Remy, Bruno Larue, Clement Yelou, and George Criner. 2006. Fish and meat demand in Canada: Regional differences and weak separability. Agribusiness 22 (2):175-199. Lerohl, Mel L., Ellen W. Goddard, and Jose L. Lomeli. 2004. Effects Of Advertising, Food Safety And Health Concerns On Meat Demand In Canada. Journal of Food Distribution Research 35 (01). Litz, Reginald A., and Gulasekaran Rajaguru. 2008. Does Small Store Location Matter? A Test of Three Classic Theories of Retail Location. Journal of Small Business and Entrepreneurship 21 (4):477-492. Lomeli, J. . 2005. What has been causing the decline in beef consumption in Canada? . Unpublished M.Sc. thesis, Department of Rural Economy, University of Alberta Maynard, Leigh J., Ellen Goddard, and John Conley. 2008. Impact of BSE on Beef Purchases in Alberta and Ontario Quick-Serve Restaurants. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 56 (3):337-351. MLA. 2009. What is a value added product? Meat & Livestock Australia 2008 [cited April, 15th 2009]. Available from . Montgomery, Alan L. 2002. Reflecting uncertainty about economic theory when estimating consumer demand. Advances in Econometrics; BookVolume: 16; 2002. Moon, S., and G. Voss. 2009. How do price range shoppers differ from reference price point shoppers? Journal of Business Research 62 (1):31-38.

114

Moschini, Giancarlo, and Anuradha Vissa. 1993. Flexible Specification of Mixed Demand Systems. American Journal of Agricultural Economics 75 (1):1-9. Nielsen Homescan National All Channels, 52 Weeks Ending December 31, 2002 – 2007. Nielsen Media Measurement, National Advertising Data, 2002-2007. Norat, Teresa and Elio Riboli. 2001. Meat Consumption and Colorectal Cancer: A Review of Epidemiologic Evidence. Nutrition Reviews 59 (2):37-47. Piedra, Mario, Alvin Schupp, and Donna Montgomery. 1995. Factors influencing consumer choice among fresh meats. Louisiana Agriculture 39(1):21-23. Reynolds-Zayak, Leona. Understanding Consumer Trends Can Present New Opportunities. Alberta Agriculture, Food and Rural Development 2004. Available from http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/all/sis8735/$file/8735.pdf? OpenElement. Reynolds, Anderson, and Ellen Goddard. 1991. Structural Change in Canadian Meat Demand. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 39 (2):211-222. Salma, Karray and Guiomar Martin-Herran. 2009. A dynamic model for advertising and pricing competition between national and store brands. European Journal of Operational Research 193 (2): 451-467. Salvanes, K., and D. DeVoretz. 1997. Household Demand for Fish and Meat Products: Separability and Demographic Effects. Department of Economics, Simon Fraser University. Schnettler, Berta, Lisete Vallejos, Ntor Seplveda, Ricardo Vidal, and Roberto Silva. 2009. Consumer willingness to pay for beef meat in a developing country: The effect of information regarding country of origin, price and animal handling prior to slaughter [electronic resource]. Food Quality and Preference 20 (2):156-165. Sethuraman, Raj. 2003. Measuring National Brands' Equity over Store Brands. Review of Marketing Science 1:1-26. Silberstein, R. B., and G. E. Nield. 2008. Brain activity correlates of consumer brand choice shift associated with television advertising. International Journal of Advertising 27 (3):359-+.

115

Sood, Sanjay, Yuval Rottenstreich, and Lyle Brenner. 2004. On Decisions That Lead to Decisions: Direct and Derived Evaluations of Preference. Journal of Consumer Research 31 (1):17-25. Srinivasan, V. 'Seenu', and Anand V. Bodapati. 2006. The Impact of Feature Advertising on Customer Store Choice. Stanford University, Graduate School of Business Research Papers 1935 Statistics Canada. 2008. Census Data 2001 [cited October, 10th 2008]. Available from http://www.statcan.gc.ca/. ———. 2009. Industry profile (Canada's food processing industry) 2007 [cited April. 15th 2009]. Available from http://www.statcan.gc.ca/pub/15-515x/2004001/index-eng.htm. ———. 2009. Overview of Canadians' Eating Habits 2007 [cited March, 15th 2009]. Available from http://www.statcan.gc.ca/pub/82-620-m/2006002/4053669eng.htm. ———. 2009. CANSIM table 002-0011, 2008 [cited March, 10th 2009]. ———. 2009. Retail store sales by selected commodity CANSIM, table 080-0018 2009 [cited April, 10th 2009]. Available from http://www40.statcan.gc.ca/l01/cst01/trad52-eng.htm. Steptoe, Andrew, Tessa M. Pollard, and Jane Wardle. 1995. Development of a Measure of the Motives Underlying the Selection of Food: the Food Choice Questionnaire. Appetite 25 (3):267-284. Stewart, Hayden and Noel Blisard. 2008. Who Pays More for Food? Journal of Agricultural Economics 59 (1):150-168. Stull and Broadway. 2004. Slaughterhouse Blues: The Meat and Poultry Industry in North America, Belmont, Calif.: Thomson/Wadsworth, 2004. xviii + 172 pp. ISBN 0-534-61303-9. Tryfos, P., and N. Tryphonopoulos. 1973. Consumer Demand for Meat in Canada. American Journal of Agricultural Economics 55 (4):647-652. Tyagi, Rajeev K. 2006. Store Brand Strength. Review of Marketing Science 4:1-16. USDA. 2009. USDA Rural Development Value-Added Producer Grants 2009 [cited April, 15th 2009]. Available http://www.rurdev.usda.gov/rbs/coops/vadg.htm.

from

116

Verbeke, Wim, and Ronald W Ward. 2001. A Fresh Meat Almost Ideal Demand System Incorporating Negative TV Press and Advertising Impact. Agricultural Economics. 25 (2-3):359-74. Verbeke, Wim, Ronald W. Ward, and Jacques Viaene. 2000. Probit analysis of fresh meat consumption in Belgium: Exploring BSE and television communication impact. Agribusiness 16 (2):215-234. Wang, Fenghua. 2002. Estimating the effects of advertising on brand and quantity choices. Ph.D., Boston University, United States -- Massachusetts. Wilkie, W. 1994. Consumer Behavior: New York: John Wiley. Wrigley, Neil. 1988. Store choice, store location, and market analysis cation, and market analysis: Economic and Social Research Council (Great Britain). Xu, Xiaosong, and Michele Veeman. 1996. Model choice and structural specification for Canadian meat consumption. Eur Rev Agric Econ 23 (3):301-315 . Yeh, M. H. . 1961. Consumer Demand for Beef and Pork in Canada. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 9 (2):68-78. Yiannaka, Amalia, Konstantinos Giannakas, and Kien C. Tran. 2002. Medium, message and advertising effectiveness in the Greek processed meats industry. Applied Economics 34 (14):1757 - 1763. Young, L. J. . 1987. Canadian meat demand. Policy Branch Working Paper 10/87. Ottawa: Agriculture Canada. June. Yuxin, Chen, Yang Sha, and Zhao Ying. 2008. A Simultaneous Model of Consumer Brand Choice and Negotiated Price. Management Science 54(3): 538-549.

117

Loading...

Project Report - AgEcon Search

RURAL ECONOMY Analysis of Value-Added Meat Product Choice Behaviour by Canadian Households Xu Zhang and Ellen Goddard Project Report # 10-04 Project ...

3MB Sizes 1 Downloads 8 Views

Recommend Documents

Project Report - AgEcon Search
Potential for Alberta's Pork in Selected U.S. Markets. Michele Veeman and James Unterschultz. Project Report # 00-02. AA

Puccinia striiformis - AgEcon Search
the Paci ic Northwest region of the United States; after. 2000, stripe rust became prevalent in eastern Canada .... too

child labor - AgEcon Search
complicating any causal interpretation of correlations that can be observed. However, the existing evidence is strong. F

Open - AgEcon Search
UNESCAP-CAPSA. The Centre for Alleviation of Poverty through Secondary Crops' Development in Asia and the. Pacific (CAPS

by Hamilton - AgEcon Search
He goes on to cover such advanced topics as factor analysis, regression diagnostics, nonlinear relationships, time-serie

by Hamilton - AgEcon Search
He goes on to cover such advanced topics as factor analysis, regression diagnostics, nonlinear relationships, time-serie

Documentos CEDE - AgEcon Search
formato impreso, digital o en cualquier formato conocido o por conocer ... neither they, nor the CEDE at Universidad de

PDF995, Job 7 - AgEcon Search
do homem sobre a língua ainda não tinha se firmado como uma forma de ciência, o que foi concretizado com o surgimento da

Alltech Mini-MBA - AgEcon Search
Known colloquially as the 'Alltech mini-MBA', the program is broken up into four modules, with interstitial 'refreshers'

nota di lavoro - AgEcon Search
NOTA DI. LAVORO. 17.2010. By Lucia Vergano, ECCET, IPTS,. JRC, European Commission,. Spain. Georg Umgiesser, ISMAR-CNR,