MASTER'S THESIS
Consumer WTP for Local and Organic Milk: Evidence from Piteå, Sweden
Andreas Gustafsson Simon Gustafsson
2014
Master of Science in Business and Economics (Civilekonom) Business and Economics
Luleå University of Technology Institutionen för ekonomi, teknik och samhälle
ABSTRACT
The purpose of this study is to estimate a potential willingness to pay a price premium for locally and organically produced milk. The study uses both a revealed and a stated preference method to collect data from consumers in Piteå, Sweden, and analyzes the resemblance in the results between the hypothetical and non-hypothetical method. The results suggest that about 80 per cent of the respondents consume local milk while about 40 per cent of the respondents consume organic milk. This indicates that it exist a WTP a price premium for both local and organic milk. Although, there is no significant difference in the respondents’ demographics when consuming local and organic milk compared to substitutes. The difference in the results between the stated and revealed preferences surveys is small which implies that the hypothetical nature in the stated preference method is limited, and that a combined model performs the best fit.
SAMMANFATTNING
Syftet med denna studie är att uppskatta om det finns en högre betalningsvilja för lokalt och ekologiskt producerad mjölk. Studien använder både en revealed preference metod och en stated preferences metod för att samla in data från konsumenter i Piteå, Sverige, och analyserar likheten i resultaten mellan den hypotetiska och icke-hypotetiska metoden. Resultaten tyder på att ungefär 80 procent av de tillfrågade köper lokal mjölk medan ungefär 40 procent av de tillfrågade köper ekologisk mjölk. Detta tyder på att det finns en extra betalningsvilja för både lokal och ekologisk mjölk. Resultatet visade däremot att det inte finns någon signifikant skillnad i de tillfrågades demografi när man köper lokal och ekologisk mjölk jämfört med substitut. Skillnaden i resultaten mellan stated och revealed preferences metoderna är låg vilket indikerar att det hypotetiska inslaget i stated preferences metoden är begränsad, och att en sammanslagen modell ger det bästa resultatet.
TABLE OF CONTENT
CHAPTER 1 INTRODUCTION 1
1.1. Background 1
1.2. Purpose 3
1.3. Method 3
1.4. Scope 3
1.5. Outline 4
CHAPTER 2 THE MARKET FOR LOCAL AND ORGANIC FOOOD 6
2.1. Market for local food 6
2.1.1. Definition of local products 6
2.1.2. Reasons for consuming local products 8
2.2. Market for organic food 9
2.2.1. Definition of organic products 9
2.2.2. Reasons for consuming organic products 10
2.3. Milk as a research item 12
CHAPTER 3 PREVIOUS STUDIES 14
3.1. Search strategy 14
3.2. WTP a price premium for local and organic products 15 3.3. Comparing revealed preference methods vs. stated
preference methods 20
CHAPTER 4 THEORETICAL FRAMEWORK 24
4.1. Traditional utility theory 24
4.2. Random utility theory 25
4.3. Altruism and warm glow 26
4.4. Theory underlying WTP 28
CHAPTER 5 METHODS 32
5.1. Concepts of revealed preferences and stated preferences 32 5.2. The dichotomous choice contingent valuation method 35
5.3. The revealed preference method 37
5.4. Econometric model 37
5.4.1. Binary model and marginal effects 37
5.4.2. Testing models – t-test 40
5.4.3. Testing models – Log likelihood test 42 5.4.4. Testing models – Log likelihood test for pooled data 42 CHAPTER 6 SURVEY DESIGN AND DATA COLLECTION 44
6.1. Survey design 44
6.2. Data collection 46
6.3. Expectations from the econometric models 47
CHAPTER 7 RESULTS 50
7.1. Descriptive statistics 50
7.1.1. Socio demographic descriptives 50
7.1.2. How do the respondents choose between local
and non-local milk? 54
7.1.3. How do men and women choose between local
and non-local milk? 54
7.1.4. How local do the respondents define milk produced
by Norrmejerier? 55
7.1.5. Do the respondents choose local milk when choosing
lactose-free milk? 56
7.1.6. How do different ages choose between local and
non-local milk? 58
7.1.7. How do the respondents choose between organic
and non-organic milk? 58
7.1.8. How do men and women choose between organic
and non-organic milk? 59
7.1.9. How do different ages choose between organic and
non-organic milk? 59
7.1.10. How many of those who consume local milk aslo
consume organic milk? 60
7.1.11. What are the main reasons for choosing local and
organic milk? 61
7.2. Econometric results 62
7.2.1. Revealed preferences for locally produced milk 62 7.2.2. Stated preferences for locally produced milk 66 7.2.3. Combining revealed and stated preferences for
locally produced milk 69
7.2.4. Revealed preferences for organically produced milk 72 7.2.5. Stated preferences for organically produced milk 75 7.2.6. Combining revealed and stated preferences for
organically produced milk 78
7.3. Testing if individual regressions are preferred over a
combined regression 80
CHAPTER 8 CONCLUSIONS 83
8.1. Is there a WTP a price premium for locally produced milk? 83 8.2. Is there a WTP a price premium for organically produced milk? 85 8.3. Is there a difference in the results between the methods? 87
8.4. Future research 88
REFERENCES 89
APPENDIX A – Prices and price premiums for different types of milk 94 APPENDIX B – Revealed preference questionnaire (Swedish) 95 APPENDIX C - Stated preference questionnaire (Swedish) 98 APPENDIX D - Revealed preference questionnaire (English) 101 APPENDIX E - Stated preference questionnaire (English) 104
LIST OF FIGURES
Figure 1 - Theory of WTP 29
Figure 2 - Monetary Evaluation Methods 34
Figure 3 - Local or non-local milk? (RP) 54
Figure 4 - Local or non-local milk? (SP) 54
Figure 5 - Women’s choice between local and non-local milk. (RP) 55
Figure 6 - Women’s choice between local and non-local milk. (SP) 55
Figure 7 - Men’s choice between local and non-local milk. (RP) 55
Figure 8 - Men’s choice between local and non-local milk. (SP) 55
Figure 9 - How local is milk produced by Norrmejerier? (RP) 56
Figure 10 - How local is milk produced by Norrmejerier? (SP) 56
Figure 11 - How many of those who consumes lactose-free milk
consumes local milk? (RP) 57
Figure 12 - How many of those who consumes lactose-free milk
consumes local milk? (SP) 57
Figure 13 - How do different ages choose between local and
non-local milk? (RP) 58
Figure 14 - How do different ages choose between local and
non-local milk? (SP) 58
Figure 15 - The choice between organic and
non-organic milk. (RP) 58
Figure 16 - The choice between organic and
non-organic milk. (SP) 58
Figure 17 - Women’s choice between organic and
non-organic milk. (RP) 59
Figure 18 - Women’s choice between organic and
non-organic milk. (SP) 59
Figure 19 - Men’s choice between organic and
non-organic milk. (RP) 59
Figure 20 - Men’s choice between organic and
non-organic milk. (SP) 59
Figure 21 - How do different ages choose between organic
and non-organic milk? (RP) 60
Figure 22 - How do different ages choose between organic
and non-organic milk? (SP) 60
Figure 23 - How many of those who consumes locally produced
milk also consume organic milk? (RP) 61
Figure 24 - How many of those who consumes locally produced
milk also consume organic milk? (SP) 61
Figure 25 - Reasons for chosen product. (RP) 62
Figure 26 - Reasons for chosen product. (SP) 62
LIST OF TABLES
Table 1 - Perceptions about local and organic food 11
Table 2 - WTP for local and organic products 19
Table 3 - Monetary meassures for environmental changes 31
Table 4 - How to calculate the Z-value 40
Table 5 - Statistics from the revealed preference (RP) survey 52
Table 6 - Statistics from the stated preference (SP) survey 53
Table 7 - Regression results with LOCAL as dependent
variable (RP) 63
Table 8 – Marginal effects for local milk (RP) 65
Table 9 - Regression results with LOCAL as dependent
variable (SP) 67
Table 10 - Marginal effects for local milk (SP) 68
Table 11 - Regression results with LOCAL as dependent
variable (RP+SP) 70
Table 12 - Marginal effects for local milk (RP+SP) 72
Table 13 - Regression results with ORGANIC as dependent
variable (RP) 74
Table 14 - Marginal effects for organic milk (RP) 75
Table 15 - Regression results with ORGANIC as dependent
variable (SP) 76
Table 16 - Marginal effects for organic milk (SP) 77
Table 17 - Regression results with ORGANIC as dependent
variable (RP+SP) 78
Table 18 - Marginal effects for organic milk (RP+SP) 80
Table A1 – Prices for different types of milk 93
Table A2 – Price premiums for local milk 93
Table A3 – Price premiums for organic milk 93
1 CHAPTER 1 INTRODUCTION
1.1. Background
The awareness of how important it is to act environmentally friendly has accelerated during the last decades as a cause of climate changes and global warming. Sea levels are rising, extreme weathers are becoming more frequent, the air is getting polluted, and more than 70.000 excess deaths due to extreme air temperature where recorded in Europe the year 2003 (Robine et al, 2007). It is of interest to solve these problems and to find methods for a sustainable environmental behavior. The transportation sector’s impact on the environment is noticed. It is the major source of several emissions, such as carbon-dioxide (CO2), and a forward-looking perspective shows that the sector will be responsible for about 16 per cent of the total man-made warming during the next coming century as a cause of CO2 and O3 emissions (Fuglestvedt et al, 2008). It is important to take actions to stop this trend, and buying local products is one way for a more environmental behavior since it decreases the transportations and thereby reduces the emissions and the environmental damages.
Local farmers face competition from national and international suppliers offering their own cheaper and non-local alternatives. Due to this situation local farmers seek new possibilities for product differentiation in order to successfully compete on the market.
An estimation of a potential willingness to pay (WTP) a price premium for locally produced goods is of interest both among environmentally conscious consumers and by producers in their marketing decisions. A potential WTP a price premium for locally produced goods provides an opportunity for local producers to set higher prices for profit maximization by differentiate themselves from competitive alternatives as being local. By understanding potential demographic differences in the WTP the producers can apply more effective marketing strategies towards their market segment. The
2
farmers can specify marketing within a consumer segment and increase awareness about environmental effects of local and non-local substitutes. An analysis of how consumers value local alternatives is also interesting for policy makers. By estimating consumer preferences for local food policy makers can apply more efficient policies to reach climate goals by setting optimal emission standards or by applying other policy tools.
As well as consuming local products there is an interest in choosing organic alternatives. While localism is preferred to reduce transportation and carbon emission organically produced goods reduced the usage of chemicals in the production. Previous studies have indicated that besides reducing chemicals in food organic farms also contribute with social, environmental and economic benefits (Lohr, 2010; O’Hara &
Parsons, 2012).
Previous studies indicate that consumers prefer local goods compared to non-local alternatives which implies that consumers value environmental factors in purchasing decisions. A local label also tends to be more efficient by creating a higher price premium than other ethical and environmental labels such as an organic label, although this attribute also returns a WTP a price premium (Hine & Loureiro, 2001). Results from previous studies also indicate that differences in psychographic characteristics influence the consumer’s WTP more than differences in economic characteristics, and that a better knowledge and experience about the product increases the price premium for the product (Akaichi et al, 2012).
For estimating a potential WTP a price premium for locally and organically produced milk this study uses both a revealed preference method and a stated preference method.
With a stated preference approach the respondents are asked to directly state how much they are willing to pay for the organic and local milk. Meanwhile a revealed preference approach uses an actual market situation for receiving the respondents’ WTP for the attributes. Due to a combined investigation approach this study will also measure the difference between the two methods. Previous studies have indicated that a hypothetical approach, such as a stated preference approach, may deviate from what the respondent would in fact do in a real situation (Seip & Strand, 1992; List, 2001).
3 1.2. Purpose
The purpose of this study is to estimate a potential WTP a price premium for locally and organically produced milk in Piteå, Sweden. This study uses both a revealed preference method and a stated preference method. The results from these two methods are thereafter compared to see whether there is a difference to be found, or if the results resemble each other.
1.3. Method
This study uses both a revealed and a stated preference method for data collection and thereby creates two different regression models to explain the consumption of local milk and two models to explain the consumption of organic milk. The models will be tested for significance through a log likelihood test meanwhile the parameters’
significances are tested through t-tests. To investigate whether there is a difference between the two methods a combined regression is constructed with a pooled dataset.
Thereby another log likelihood test will be conducted for analyzing the significance of the combined model. This test evaluates whether the combined model creates a better fit than the two separate models.
1.4. Scope
The study estimates a potential WTP a price premium for locally and organically produced milk in Piteå. The reason for only conducting the research in Piteå is that we believe there is a difference in a WTP a price premium between different regions and cities. Therefore the results become more specific for consumers in Piteå and can thus be applied by producers and marketers in Piteå.
The reason for choosing milk as a research item is that it is a commonly consumed commodity whereby it facilitates finding respondents that consumes the product in the revealed preference survey. Since milk is a common good it is also a greater possibility that the respondent in facts knows which kind of milk he or she usually consumes. The milk that can be consumed in supermarkets in Piteå can both be organically and locally
4
produced and therefore both attributes are interesting to valuate since they both can be perceived as environmentally friendly.
1.5. Outline
This paper begins with a brief introduction of the market for locally and organically produced commodities in chapter 2. The chapter discusses the unclear definition of what can be defined as a locally produced good and some guidelines that defines an organic good. It is also brings up some possible benefits from consuming local and organic food compared to conventional food. Chapter 2 then ends with a brief discussion about the research item, milk, how the market for dairies is in Piteå, and why milk is considered a good research item. After introducing the market for local and organic food the paper continues with a presentation of previous studies that have analyzed a potential WTP a price premium for local or organic products in chapter 3. Besides studies about local or organic products this chapter also presents studies that have compared stated preference approaches with revealed preference approaches.
In chapter 4 the theoretical framework of this study is presented. First there is a description of the utility maximization theory and the further developed random utility theory that are stating that a consumer acts according to receive the highest expected utility. The chapter then continues with the concept of altruism and warm glow that an individual might receive utility not only by acting selfish. The chapter concludes by connecting the concepts of utility to the concept WTP, and it is shown graphically and with equations how WTP can be calculated from utility models. After discussing the theoretical framework the paper presents the two methods of the study in chapter 5. This chapter displays some strengths and weaknesses associated with stated and revealed preference methods and declare which type of stated preference method that is being used in this study. Chapter 6 will then explain how the questionnaires for the two methods are designed and why the questions are relevant to analyze the WTP a price premium for local and organic milk. The chapter also goes in detail of how the data was collected for the stated and revealed preference analyses and clarifies some implications of the data collection.
5
The results for both methods are presented in chapter 7. First some descriptive statistics are displayed and thereafter the results from the econometric models are shown. In this chapter marginal effects of the explanatory variables are calculated for each regression model. Thereafter the chapter tests whether a combined regression model is to be preferred over separating the revealed and stated preference models. The results are then further discussed in chapter 8 with a presentation of some interesting ideas for further researches.
6 CHAPTER 2
THE MARKET FOR LOCAL AND ORGANIC FOOD
In this chapter some basic concepts of a local and an organic product are introduced.
The chapter begins with an introduction of the market for local food. Section 2.1.1 explains the vague definition of a local product and how the usage of a local label may differ between regions and retailers. Thereafter section 2.1.2 presents some reasons to why a consumer may prefer and consume local products over non-local alternatives.
There are several benefits when consuming from local farmers and this section discusses some economical benefits and shows statistics on reasons to why consumers choose local products. The chapter continues with introducing the market for organic food in section 2.2 with a definition of an organic product in section 2.2.1. Thereafter some reasons to why consumers might choose organic alternatives over non-organic substitutes are discussed in section 2.2.2. The chapter then ends with a short description of the research item and why this paper analyzes milk in section 2.3.
2.1. Market for local food 2.1.1. Definition of local products
There is no universal definition of what a local good is, although the concept of local goods is somehow related to the distance between the producer and the consumer. This lack of definition has created a situation where different distributors use local labels in their own ways. Wal-Mart, being one of the leading retailers in the U.S. uses locally- produced labels on all products being produced within the same state independent of the size of the state and the transportation time. Meanwhile Whole Foods considers local food as goods being transported in maximum seven hours from their distributor to the
7
stores independently of any geographical boundaries, even though the distance generally does not exceed 200 miles (Schmit, 2008).
A local distance might also vary depending on type of region and socio-structures. What is defined as locally produced can therefore vary depending on the population rate and if the region is sparsely populated or more densely populated. In a sparsely populated area the term local tends to have a wider range than in cases with big cities where a local term refers to a shorter distance from the producer. The definition of locality can also depend on political and geographical boundaries such as regions, states or counties.
This uncertainty with defining what is local is known as “flexible localism” (Ilbery &
Maye, 2006).
Even though there is no clear definition of what a local good is there have been numerous attempts to specify the characteristics for allowing a good to be defined as locally produced. According to the U.S. Farm Bill 2008 a product can be eligible for a locally produced label as long as the total transportation from producer to consumer does not exceed 400 miles (Martinez et al, 2010). However, a study in U.S. showed that 50 per cent of the consumers define local products as been produced within a maximum of 100 miles, 37 per cent define a local product as state-made, 4 per cent define a local product as region-made, and 4 per cent define a local product as made in the U.S (Schmit, 2008).
Another view is that a local good is distinguished not only by where it is produced. Who the producer is and how the product is produced are also important questions to answer before labeling a good as locally produced or not. According to Thompson et al (2008) local food is defined by freshness, better taste and less usage of energy and pollutions. If the production is geographically close to the consumer it is a greater likelihood that the products could be characterized as being local because it is a greater probability that the products are fresher and has been transported with less usage of pollutions. But if nearby farmers use much energy and pollute much emission these products should not be labeled locally produced according to Thompson et al (2008). Therefore other variables than the distance between the producer and the consumer and the time for the transportation have to be considered before classifying a good as being local or not.
Thompson et al (2008) also imply that other factors such as landscape and attractiveness of the ranch could be included when valuing the products origin. Also ethics and
8
producer personality can be included in the concept of locally produced (Thompson et al, 2008).
However these measures of distance and geographical boundaries are formed in the U.S.
and therefore they are not adequate for measurements in a smaller nation as Sweden.
There is although no clear definition of what a local product is in Sweden. Therefore it is not possible to state a number of miles between the consumer and the producer when defining a product as either local or non-local. Instead it is the consumer who defines the products as local or not according to themselves. A Swedish survey (Coop-report, 2009) investigates the consumers’ perceptions of locality in Sweden. According to this report 78 per cent of the respondents consider that a local good is produced in the same region as it is consumed, 47 per cent consider that a local good is made in the same municipal, 45 per cent consider that a local good is made in Sweden, and 5 per cent consider that a local good is produced in Europe.
The Farmers’ market in Sweden (Bondens egen marknad) has established its own definition of locally produced goods. According to the Farmers’ market (2013) the distribution area cannot exceed 250 kilometers in order to be considered as local. For the products to be allowed to be sold on the Farmer’s market as being a local good the farmer has to have grown or bred everything on his own farm.
2.1.2. Reasons for consuming local products
There are several reasons to choose local alternatives before non-local substitutes. When analyzing consumers’ WTP for local products, a potential premium compared to the non-local alternatives should reflect all these factors. According to Martinez et al (2010) locally produced food may create economic benefits to the local economy. When purchasing local products it creates job opportunities within the area and increase regional incomes. The increased income creates a multiplier effect where the income is invested regionally and helping other regional business and incomes to grow. A study has displayed this multiplier effect when analyzing local farmers in Iowa (Otto &
Varner, 2005). This study of over 4.500 consumers and 780 vendors in Iowa showed an income multiplier of 1.47 implying that each dollar earned at the farmers’ market generates additional 47 cents in indirect and induced income. The result of the same
9
study was a gross sales multiplier of 1.58 implying that each dollar spent at the local farmers’ market resulted in 58 cents in indirect and induced sales within the state. The job multiplier was 1.45 implying that each new job at farmers’ market in Iowa created almost a half full-time job at another sector within the state. The results from this study show the economical benefits from purchasing local alternatives before non-local substitutes. A similar study in Oklahoma found an income multiplier of 1.66, a gross sales multiplier of 1.78, and a job multiplier of 1.41 (Henneberry et al, 2009).
However, there are perhaps more advantages than the potential economical benefits.
Other advantages are that locally produced food are associated with being fresher, has better quality and has fewer additives. According to Coop-report (2009) the most important reasons for consuming locally produced food are the environmental benefits (85%) due to reduced transportations. The respondents in this survey also chose the local alternative to support local producers (79%) and to induce economical benefits in the area. 28 per cent chose the local products with the reason that it brings better quality than non-local alternatives. Becker (1996) and Andreoni (1990) also discuss that altruistic factors may also affect the consumers’ in their decision making. They imply that feeling good and a good self-image can contribute to acting environmental friendly.
Therefore consumers’ might choose local alternative, just for feeling good, without considering economical benefits or better quality. Since there are several reasons to choose local alternatives before non-local substitutes (Coop-report, 2009; Martinez et al, 2010) a potential WTP premium would include all these factors.
2.2. Market for organic food
2.2.1. Definition of organic products
As described previously a local product is often only associated with the distance between the consumer and the producer, and the definition is quite vague since there are no regulations of the term. However, an organic product needs to fulfill several criteria for receiving its label according to the National Organic Program; Final Rule (2000).
This program establishes a list of substances that are approved and prohibited in the production of organic food. An organic product is according to National Organic Program; Final Rule (2000) produced with limited use of chemicals and aims to preserve biodiversity and to utilize nature resources in a sustainable manner.
10
The Swedish National Guidelines for Organic Production (2014) interpret the legislation of organic products from the European Union and states some criteria of what should be considered as an organic product. According to the national guidelines, an organic production must have a sustainable agriculture and aquaculture to respect the ecosystem. Moreover, an organic production must contribute to biodiversity by using nature resources responsibly and to produce food of the highest quality. Finally, the product must serve the consumers’ demand when it comes to producing food that do not harm the environment, people welfare, or animal welfare in order to be considered as organically produced.
2.2.2. Reasons for consuming organic products
As with locally produced food there are several potential advantages when consuming organic food compared to when consuming conventional food. A previous study (Lohr, 2010) has analyzed social, environmental and economic benefits associated with organic goods when comparing counties with organic farms to counties without organic farms in the U.S. In total 36 parameters were investigated and compared between the two types of counties. The results of this study suggest that organic farms provide the local economy with several benefits compared to conventional farms. The property tax paid is significantly higher in states with organic farms which creates a multiplier effect were money can be reinvested in the local area. This multiplier effect resembles the multiplier effect from purchasing local alternatives which was mentioned previously in section 2.1.2. The market values for the organic farms are also significantly higher which is important for the financial stability for the farms. Besides the economic benefits the results also suggest that organic farms provide more wildlife habitat compared to conventional farms. The use of chemicals such as insecticide use and nematicide use also was proven to be smaller for the organic farms compared to the conventional farms.
Another study has investigated the economic benefits from organic dairy farms in Vermont and Minnesota (O’Hara & Parsons, 2012). The study compare how sales revenues affect the state’s output, the labor income, the gross state product, and the employment between organic and conventional farms. According to this survey all these parameters are higher for the organic farms compared to the conventional farms. The results also suggest that an increase in sales revenue of an organic farm return an 83 per
11
cent higher increase in employment and 33 per cent higher state gross product compared to conventional farms in Vermont. In Minnesota an increase in sales returned a 14 per cent higher employment effect and 11 per cent higher increase in gross state product compared to conventional farms (O’Hara & Parsons, 2012).
A study made in Canada (Campbell et al, 2012) shows that the perceived benefits from consuming organic food can be that organic products have no synthetic pesticide use (83%), they taste better than conventional food (33%), and are more nutritious (34%).
Besides analyzing the perceptions for organic food this study also compares how individuals characterize the organic label to a local label. As discussed in section 2.1.1 a local good is characterized by being produced geographically close and reduces the transportation between the producer and the consumer. In Campbell et al (2012) 73 per cent of the respondent characterized a local product as reducing transportations.
Although 73 per cent thought that a locally produced good reduced transportation, only 28 per cent characterized the product as lowering the carbon footprint. This percentage was just a bit higher than the result for the organically produced food (27%) and implies that the respondents value an organic and a local product equally in reducing the carbon.
In table 1 some results from Campbell et al (2012) is displayed.
Table 1: Perceptions about local and organic food Perception Local Organic No synthetic pesticide use 11% 83%
No natural pesticide use 5% 29%
Better for the environment 28% 52%
Better taste 38% 33%
More nutritious 22% 34%
Non genetically modified 13% 55%
Lower carbon footprint 28% 27%
Lower greenhouse gas emissions 22% 24%
Decreased transportation 73% 15%
Source: Campbell et al (2012)
12 2.3. Milk as a research item
In this study milk is used as a research item, and the purpose is to analyze whether the consumers are willing to pay more for organically produced milk and milk produced from Norrmejerier, a product that potentially can be classified as locally produced by consumers in Piteå. Norrmejerier is a Swedish economic association whose business idea is to offer fresh and attractive dairies of high quality and to provide Northern Sweden with economic benefits. Milk produced by Norrmejerier is produced both in Luleå, which is located 54 kilometers from Piteå, and Umeå, which is located 217 kilometers from Piteå. The company also has activities in other cities producing cheese and other diaries (Norrmejerier Annual Report, 2012).
In this study data is collected from respondents after they have consumed at Coop Forum in Piteå, with the hope that they had purchased milk at the time of the survey. If the respondent had consumed milk then he or she filled a questionnaire to state which product he or she had bought. If the respondent had not consumed milk then the respondent instead filled a questionnaire to state which product he or she usually will buy.
Besides milk produced by Norrmejerier there also exist substitutes produced by Arla, Valio and Coop Forum’s own label. These three other brands are considered as non- local compared to Norrmejerier and are cheaper than Norrmejerier assuming that attributes such as level of fat and content of lactose are the same (see appendix A).
When studying the price levels of the products in Coop Forum, in Piteå, Norrmejerier is about 15 per cent more expensive to consume compared to the non-local alternatives with the same attributes.
Another attribute of the milk produced by Norrmejerier is that it can be bought organically produced or not organically produced. Therefore the consumers decide whether they are willing to pay a bit more for organically produced milk or not (see appendix A). The organic attribute also exist on some products of the other brands whereby it is possible to consume organic and not local, as the same time it is possible to consume local and not organic. When studying the prices for organic milk in Coop Forum organic milk has generally a price premium of 14-15 per cent compared to inorganic substitutes. This premium is similar to the price premium for local milk. The fact that milk can be purchased with a local and an organic label was important for this
13
study since the purpose of this study is to analyze a potential WTP a price premium for both the local and the organic attributes. Therefore different combinations of organically and locally produced milk made this product a proper research item.
Another reason for choosing milk as a research item is that the product is quite homogenous and that taste and quality will not differ so much between brands compared to other products. Other studies who are investigating a potential WTP a price premium for different labels often uses common and homogenous products such as potatoes, tomatoes and milk (see Carroll et al 2013; Hine & Loureiro, 2001). The homogeneity of the products simplifies when only being interested in altruistic, economical and environmental factors of a local label and an organic label, and not difference in taste or quality that may be determinate factors in studies using heterogeneous products such as wine or meat (see Gracia et al, 2012; Thiene et al, 2013).
Since this thesis also examines the difference between a revealed preference analysis and a stated preference analysis we believe that milk is a good research item. Several consumers buy milk very often which expects to facilitate the data collection for the revealed preference study. If instead analyzing a product that is not consumed on a regular basis it would have been more difficult to receive enough data for the revealed preference analysis. A commodity that is consumed on a regular basis is also an advantage for the stated preference study, since the possibility that the respondent knows which products he or she in fact consumes is relatively high.
14 CHAPTER 3 PREVIOUS STUDIES
This chapter presents previous studies that are relevant for this study. First there is an explanation of how the papers were found and what keywords that were being used. The second section presents previous studies that have analyzed a potential WTP a price premium for local or organic attributes for different types of produce and in different geographical areas. To reduce the wide extent of previous studies only previous studies regarding locally or ecologically produced food, and not other goods, are discussed. The purpose of the literature overview is to get an overview of previous results and conclusions, something which is also presented in a table at the end of the section. The third part of this chapter contains previous studies that have compared the two different methodologies, a stated preference method and a revealed preference method. This section discusses the reliability of the two methodologies and how different attempts have been made to reduce biases when conducting stated preference approaches. The chapter then ends with a conclusion of the main results.
3.1. Search strategy
The literature used in this chapter is collected from LTU databases, using Scopus and Proquest most frequently. Google Scholar was also applied in order to find articles about the risk of hypothetical bias when conducting a stated preference approach.
Besides these databases the main source of relevant papers were found in the reference list of previous studies.
Studies estimating consumer’s WTP for different local products were first collected.
These studies were found by searching (WTP “locally produced”) and (WTP “local food”). Some of these studies were also investigating a potential price premium for an
15
organic attribute. However, for additional studies for organically produced food there were also searches for (WTP “organic food”) and (WTP “organically produced”). The second part of the literature overview displays previous studies that have compared stated and revealed preference methods. Studies about the difference between stated and revealed preference methods were found by using the same databases as for the WTP for local and organic products but with searches on (“contingent valuation”), (“contingent valuation” bias) and (“hypothetical bias”).
3.2. WTP a price premium for local and organic products
There have been several studies estimating a potential WTP a price premium for local, organic or other environmentally favorable products using different methods and in different geographical areas. Akaichi et al (2012) analyzes the market potential for the locally produced white bean Mongeta Ganxet in Catalonia, Spain. To estimate the consumers’ WTP for this local product the authors apply a non-hypothetical economic experiment where the participants were offered real products and were allowed an exchange of real money in a simulated market situation. With a non-hypothetical approach the authors exclude methods such as contingent valuation which otherwise is commonly used in these types of surveys since it takes into account both use value and non-use values. However, by using a non-hypothetical approach this study reduces the risk for overestimation which otherwise can be the result of a stated preference approach.
The results show that the local white bean has a relatively small market in Barcelona.
Around half of the participants (53%) were not willing to pay a premium for the Mongeta Ganxet despite receiving information about the organoleptic1 quality of the product. Furthermore, there were only 7 per cent of the participants that were willing to pay more than the minimum market price for Mongeta Ganxet that can be bought from the producer or a local market. The results also indicate that individual differences in psychological characteristics influence the consumer’s WTP more than differences in economic characteristics and that a better knowledge and experience about the product increases the WTP a price premium for the local product.
1 Aspects of food experienced by senses, e.g. smell, sight and taste (Oxford Dictionaries)
16
One drawback of the experiment is that it only included 90 respondents from an area with a population of four million inhabitants. Thereby it is questionable whether the results are representative for the population in Barcelona and whether it is manageable to draw any conclusions about the demographic and economic factors for the WTP a price premium.
Another study (Yue & Tong, 2009) analyzes the WTP for organically grown and locally grown tomatoes in Minnesota, U.S. The aim is to analyze whether there is a difference in the consumer’s WTP a price premium and to see whether there are any differences in market segmentations between those who consume these two types of produce.
Yue & Tong (2009) collects data from both a hypothetical experiment and a non- hypothetical choice experiment and tests for similarities in a joint estimation. In the hypothetical choice experiment tomatoes are placed on a table and the respondents are asked to evaluate different scenarios of conventional, organic and local attributes.
Totally 365 respondents participated in the experiment, 115 participated in the non- hypothetical experiment and 250 participated in the hypothetical experiment and each individual that participated in the study were only participating in one of the two experiment types to reduce the risk of biases.
The results imply that freshness is the most important factor behind the decision of buying locally produced tomatoes and there was an average WTP a price premium for both organically and locally produced tomatoes by $0.67 compared to conventional tomatoes. The result thereby suggests that consumers value a locally produced label equally high as an organically produced label. Moreover the results showed that older women with a lower education should be the main target for marketing locally produced tomatoes, and that it might exist a hypothetical bias when performing hypothetical experiments. For the local attribute there was a difference by 9 per cent between the stated preference analysis and the revealed preference analysis. Since this study uses real products in the hypothetical experiment Yue & Tong (2009) discuss that this difference between the revealed and stated value probably is lower than what can be expected when using pictures instead. A previous study (Yue et al, 2009) indicates that the usage of real products instead of pictures reduces the size of a hypothetical bias. Yue et al (2009) uses real products for some respondents and pictures for the rest of the respondents in a hypothetical treatment and compares the results to a non-hypothetical
17
approach. According to the results the hypothetical values were significantly higher compared to the revealed values. And there was found a significant variation in WTP between the picture treatments and the real products treatment. The picture treatment generated higher WTP compared to the real product treatment. Therefore these results indicate that a picture treatment might be a reason for a larger hypothetical bias compared to real product treatment in a stated preference methodology.
Many studies have investigated the market potential for local farmers and whether a local label increase the consumer WTP for locally produced specialty food. A study in northern New England investigates whether the residents generally express a favor for locally produced goods over imported substitutes, and if so, what the price premium is for such a good (Bond et al, 2005). The study applies a dichotomous contingent valuation method and collects data through a mail survey. In the contingent valuation study the respondents were asked if they were willing to pay a price premium for either a $5-good or a $20-good to investigate how a WTP a price premium affects with the base price of the product.
The results of the study indicate that residents in New Hampshire, Vermont and Main are generally willing to pay an extra premium for local specialty goods and this premium increases as the base price of the good increases. The results suggest that the usage of a local state logo has the potential to differentiate locally produced specialty goods from imported substitutes which may allow producers to set higher prices without a loss in sales.
Results from Yue & Tong (2009), Akaichi et al (2012) and Bond et al (2005) indicate that consumers prefer local products before non-local alternatives. But how a local label is perceived compared to other environmental and ethical labels is studied by Hine &
Loureiro (2001). This is a study of how the Colorado potato farmers could niche their products as a local (Colorado Grown) product, as an organic product, or as a GMO-free product, which implies that the potatoes are not manipulated in their growing process.
The objective of this research is to reveal and compare consumers’ WTP for Colorado grown potatoes marked as either organic, GMO-free, or Colorado grown.
The results show that the Colorado-grown label carry a higher price premium than to niche the product as either organically grown or GMO-free. According to the results, locally grown potatoes carry a 10 per cent price premium. The premium for organic
18
potatoes on the other hand turned out to be about 7 per cent, and the GMO-free potatoes had a price premium around 6 per cent. This result is of interest since it shows that a locally produced good can receive a higher WTP premium compared to GMO-free and organic foodstuff, ceteris paribus. It also shows that locality is important for the consumers in their purchasing decisions. Additional results imply that respondents of the upper class were willing to pay more for organically grown potatoes than respondents in a lower class. Furthermore, age had a negative effect on WTP for the organic attribute which implies that respondents lowered their WTP for the local alternative as they became older. All these factors were statistically significant at 10 per cent significant level. However, the results suggested that there were no significant difference between the socio demographics when it came to the local label.
Overall there have been many researches about WTP for locally and organically produced goods, and a majority of these results are stating that consumers are willing to pay a price premium for different types of environmentally kind products. And when comparing the local label with other labels, such as organic, sugar free or low fat, the local label tends to be valued highest, something which makes our research interesting since it is a growing market. Previous studies also suggest that the demographic differences affect consumption behavior and thus WTP for both local and organic products. Women tend to valuate environmentally goods higher than men, age seems to have a negative effect on WTP, and psychological characteristics seem to have a larger effect on WTP than economical characteristics. In table 2 the results from previous studies investigating the WTP for local and organic products are displayed.
19
Table 2: WTP for local and organic products Article Year Region Product Method Results
Akaichi et
al 2012 Catalonia, Spain
White bean
Non-
hypothetical experiment
Low/ non WTP was found for local product. The
psychographics are more important than economic characteristics in the influence of WTP
Gracia et
al 2012 Zaragoza, Spain
Lamb meat
Experimental auction
WTP was found for local lamb meat. Women have higher WTP compared to men, and people who have higher local embeddedness tend to have higher WTP
Yue &
Tong 2009 Minnesota,
U.S. Tomato
Non-
hypothetical and
hypothetical experiment
WTP was found for locally produced. Older women with lower education have the highest WTP.
Hine &
Loureiro 2001 Colorado,
U.S. Potato Contingent valuation
A locally produced label has a higher WTP than organically- grown or GMO-free.
Carroll et
al 2013
Mid- Atlantic states, U.S.
Tomato Mail-survey
Consumers are more concerned about where the product is produced than production methods
Hu et al 2009 Kentucky, U.S.
Blueberr y
Conjoint experiment
WTP premium for local products exists. Locally grown label also receives better support than organic or sugar- free.
Bond et al 2005
Northern New England, U.S.
Two different goods
Dichotomous contingent valuation
WTP was found for locally produced. WTP was higher for the good that had higher price base.
Thiene et
al 2013 North-East Italy
Prosecco wine
Questionnaire -based survey
Different locally produced labels receive different WTP
Adams &
Salois 2010 U.S. Food
Summarizes local food studies in U.S.
WTP for local food has increased during the last decades, and WTP for local food is higher than WTP for organic food.
James et al
2009 Pennsylva nia, U.S.
Processe d food
Choice experiment
A local label receives higher WTP compared to organic, low fat, or no sugar-labels
20
3.3. Comparing revealed preference methods vs. stated preference methods
There are two different approaches when estimating the consumers’ WTP for locally produced goods; stated preferences and revealed preferences. Although, stated preference approaches have faced criticism since it has been revealed that the values can get over- or underestimated, something which is called hypothetical bias. Several studies display that respondents value environmental factors differently when comparing a stated preference approach and a revealed preference approach. In surveys using a revealed preference approach respondents display actual WTP as they in fact pay for the product. With contingent valuation, a stated preference approach, respondents are asked to directly state their WTP for the good in question. Since the respondent do not actually pay anything these hypothetical studies often generates different types of biases making survey results unreliable.
Seip & Strand (1992) seeks whether evidence of bias could be found when constructing a contingent valuation study with real payments for membership in Norwegian Society for the Conservation of Nature (Norges Naturvernforbund, NNV). In this study 101 Norwegians were interviewed whether they would be willing to pay an annual fee to enter NNV at 200 Norwegian kronor (NOK). From this sample 64 respondents (64%) displayed a maximum WTP that exceeded 200 NOK and would according to their statements accept a membership in NNV. Therefore these respondents where then received a membership application to see if their displayed WTP from the stated preference analysis would resemble a real life situation. However, only 6 out of 64 (9%) became members in the organization. This implies that a stated preference approach where respondent are asked questions to display WTP is not a realistic valuation tool.
Through telephone interviews with 25 out of those 58 who did not pay the fee Seip &
Strand (1992) analyzed why there might be an upward bias from the contingent valuation study. From this study 17 respondents changed their decisions downwards and 24 respondents answered that their previous statement would have been a WTP for a wider environmental aspect than the organization in question. This is an example of a so-called part-whole bias that can be found in stated preference analyzes when respondent value a more comprehensive environmental good than the narrowly defined good in the survey (Perman et al, 2011).
21
The result from Seip & Strand (1992) indicates that respondents would change their stated WTP if they had longer time to think about their decision. This is also studied by Cook et al (2011) who seeks how “time to think” affects WTP for a vaccine. The result suggest that the average WTP decreased with 41 per cent when the respondents had time to interact with their spouse and became more certain about their decision.
Certainty is also important to reduce hypothetical bias in stated preference analyses.
Blumenschein et al (2008) investigates how follow-up questions about the respondents’
certainty affect their responses. In this study a stated preference analysis and reveal preference analysis was compared with each other. In the stated preference approach the respondents were asked whether they are willing to pay $40 for a diabetes management programme. Then, they had to answer if they were “definitely sure” or “probably sure”
about their previous statement. When analyzing the results between the two methods it was found a significant difference for the whole sample, but when dropping the respondents who answered “probably sure” the difference and thus the hypothetical bias more or less disappeared. This indicates that stated preference methods do not display an intention to undertake a real action unless the respondents are definitely sure about their statement in a contingent valuation survey. List (2001) also criticizes the nature of the traditional contingent valuation analysis. According to List (2001) the common type of WTP study with a hypothetical approach will tend to cause a slight upward bias in relation to the actual values.
However, despite the criticism it is sometimes essential to perform a stated preference approach and therefore it is crucial to create a hypothetical model that is reliable. A hypothetical model is cheaper to perform, is easier to receive data from, it captures non- use values and does not take as much time as a revealed preference approach. Also it is necessary to use a stated preference approach when there is no market for the good or environmental factor in question. Yadav (2007) addresses the issue of a hypothetical bias in contingent valuation studies. This essay contributes to creating a reliable contingent valuation technique by investigating the cause and nature of hypothetical bias. The hypothesis of this analysis is that a hypothetical bias is derived from various uncertainties.
In order to examine a potential hypothetical bias as a result of different uncertainties the author uses both real and hypothetical approaches with different levels of uncertainty.
22
Yadav (2007) measures this by a 10 point certainty scale. The stated hypothesis of the paper is that those that will report an uncertain “yes” in a stated preference approach will be answering “no” in an actual payment situation. This potential difference is the explanation for hypothetical bias.
The results revealed no significant difference between the hypothetical test and the real market situation between the different levels of uncertainty. Therefore, there is not valid to state that uncertainty causes hypothetical bias according to these results. However, there was found a statistically significance at a 1 per cent level when it comes to the difference between stated and revealed values as a cause of the size of the value. A difference between stated and revealed values was less likely to be found when there was a higher amount at stake. The results from Yadav (2007) therefore suggest that a hypothetical bias can be a greater problem when the value of the product is lower.
Yadav (2007) also tests a prediction format2 to see whether prediction estimates are effective against hypothetical bias in a new survey. In this research the respondents were asked how they believe others would respond, as well as how they self respond, in a contingent valuation framework. The test showed a possible hypothetical bias when conducting a traditional contingent valuation methodology. The real value was $8.46, compared the hypothetical value of $10.40. The prediction test also showed a significant difference between the real values ($5.21) and the prediction estimates ($6.91). From the results there was not found that a prediction model is better than a traditional contingent valuation model.
Another method to eliminate hypothetical bias in contingent valuation surveys is to use a cheap talk script (Cummings & Taylor, 1999). A cheap talk script has three general points to eliminate hypothetical bias. First, the script tells the respondents the concept of hypothetical bias. Secondly, it explains why the respondents might answer differently in a stated preference analysis than in a revealed preference analysis. And as a third point it requests the respondents to consider this problem and answer as it was a real situation with actual payment. According to Murphy (2004) the cheap talk script had no significant impact on the hypothetical responses when they analyzed the WTP for the Nature Conservancy in Massachusetts, U.S. Although, they recognized that the cheap
2 Respondents are asked how they believe other respondents would answer the same question (Yadav, 2007)
23
talk script was effective when they eliminated the responses of the lower values and only considered when the payments were higher.
Overall there have been numerous studies investigating the difference between stated and revealed preference surveys. There are advantages and disadvantages of conducting both methods, and there have been several attempts to reduce the hypothetical nature in stated preference methods.
24 CHAPTER 4
THEORETICAL FRAMEWORK
Chapter 4 introduces the theoretical framework behind the analyzes for local and organic milk. The chapter begins with an explanation of the concept utility and how this is implicitly applied by consumers when making their consumption decisions in section 4.1. In section 4.2 some weaknesses of traditional utility theory are discussed and the random utility theory is introduced which is a theory that develops the traditional utility theory by adding a random parameter. Furthermore, the concept of altruism and warm glow is mentioned in section 4.3. These factors might affect a consumer to be willing to pay more for one product before a similar product just to receive a good feeling. The discussion about utility then ends up in an introduction about the theory behind WTP in section 4.4. This section graphically displays the definition of WTP a price premium for locally or organically produced milk.
4.1. Traditional utility theory
As displayed in chapter 3.2 several studies have indicated that it often exists a WTP a price premium for both organic and local food. In chapter 2 some of the underlying reasons to why a consumer might be willing to pay more for local or organic products compared to nearby substitutes are explained. However, these reasons for consuming local or organic products are individual and different consumers act in different ways when choosing between a local or an organic product and a substitute. So what are the factors behind which product we end up consuming?
The utility maximization theory is implicitly used by the consumers when they make their purchasing decisions and is displayed in which product a consumer ends up consuming. The theory states that an individual consumes commodities according to
25
what brings the highest perceived utility given his or her own preferences and budget constraint. What generate utility are all the aspects of the good, the individual, the purchasing situation and every other aspects that have any impact on the individual’s choice. Often price (p) and quality (q) are mentioned as critical variables behind an individual’s choice. However, it should be noticed that aspects like personal experiences and cultural environment can also have an impact on the individual’s choice. Since there are several parameters to consider when analyzing an individual’s utility not all parameters can be considered and included in a utility function. Therefore the ceteris paribus3 assumption is invoked in most economic analyses in order to make the analysis manageable.
If an individual consumes local products instead of non-local products, then the reason is that the locally produced food generates a higher perceived utility, given that both products are feasible given within the budget constraint. By studying the consumption behavior and observing which product the consumer purchases the products can be ranked and interpreted in how an individual perceive utility from different alternatives, like in this study ranking a local or organic good compared to substitutes. However, one drawback with ranking alternatives according to utilities is that we cannot attach any numbers to these utility rankings. Thereby it is not possible to state how much more one good is preferred compared to another good. The values received from the utility function simply record the relative desirability of the commodities (Snyder &
Nicholson, 2011).
4.2. Random utility theory
In traditional utility theory people are assumed to make their decisions through a utility maximization approach. In random utility theory these thoughts are further developed.
In random utility theory it is assumed that the underlying reasons for the chosen alternative are only partially observable and it is acknowledged that everything affecting the individual’s purchasing behavior cannot be observed or measured. Even if the individual always acts rationally and thereby acts according to what brings the highest perceived utility, it is not possible to observe every characteristic of the product,
3 Ceteris paribus = other things being equal (Snyder & Nicholson, 2011)
26
purchasing situation, or characteristics of the individual. Therefore random utility models presume that a countable deterministic variable (V), based on observations, has to be combined with a random unobservable component ( ). When analyzing an individual’s choice between different alternatives this random component captures the factors that cannot be observed and that are not included explicitly in the model but have an impact on the choice. For example, we can observe that an individual consume their preferred good A before good B in nine out of ten times. However, occasionally the individual consumes good B, and still acts according to what brings the highest utility. This deviation from the “regular” purchase cannot be measured through a traditional utility function, assuming that all observable variables keep constant.
However, by introducing an unobservable component ( ) in random utility models the deviation from the “regular” purchase can be considered (Cascetta, 2009).
The true utility function (u) for individual i for good A will therefore be expressed as:
Therefore this random component makes it impossible for an observer to fully describe the individual’s utility function and to predict the consumer’s purchasing choice.
However, the probability that one good generates higher utility compared to another good can be calculated through observations. The concept of introducing a random component as a result of the impossibility to observe all variables that affect an individual’s choice is the basic fundament of random utility theory (Cascetta, 2009).
4.3. Altruism and warm glow
This study investigates a potential WTP a price premium for both a local and an organic label for milk. Both the local and the organic attribute may contribute to make the product taste better or being of better quality (see section 2.1.2 and section 2.2.2), but even if not, consumers might be willing to pay more for these attributes.
An individual’s utility from purchasing local or organic food is usually expressed as a function of both price (p) and quality/taste (q) in regular utility theory. But according to Becker (1996) the respondents also value products through altruistic factors (a). Becker assumes that individual’s are forward-looking and wants to maximize welfare as they
