• No results found

Which ethical production methods are consumers most concerned about? : using survey data to assess consumers’ preferences for improved human rights, animal welfare and environmental standards

N/A
N/A
Protected

Academic year: 2021

Share "Which ethical production methods are consumers most concerned about? : using survey data to assess consumers’ preferences for improved human rights, animal welfare and environmental standards"

Copied!
51
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Which ethical production

methods are consumers

most concerned about?

Using survey data to assess consumers’

preferences for improved human rights, animal

welfare and environmental standards

Megan Roberts

11088117

Supervisor: Andras Kiss

MSc Economics Thesis

Public Economic Policy

15 ECTS

(2)

1

Table of Contents

Table of Contents ... 1 Statement of Originality ... 1 Extract ... 1 List of Figures ... 2 1. Introduction ... 3 2. Literature Review ... 4 3. Theoretical Background ... 7

4. Research Questions and Hypotheses ... 10

5. Methodology ... 12

6. Findings ... 20

7. Conclusion and Discussion ... 35

Appendices ... 37

Bibliography ... 48

Statement of Originality

This document is written by Student Megan Roberts who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Extract

This research used over 300 responses to an online, hypothetical choice experiment to estimate the willingness to pay for improved sustainability of production methods in chicken, eggs, milk and chocolate. Results showed that approximately 35% of respondents care about all variables equally. The choice experiment showed that consumers are prepared to pay an additional €0.61 for a one unit score increase for 500g of chicken, €0.25 for six eggs and €0.10 for 100g of chocolate. The willingness to pay to increase each individual variable score is fairly similar across products, but when looking at choices made animal welfare was the most important for chicken, eggs and milk, with human rights found to be the least important. Of the chocolate brands tested, Ethiquable and Albert Heijn had a significant and positive willingness to pay to increase the overall score by one unit of €0.14 and €0.08 respectively.

(3)

2

List of Figures

Table 1: Explanation of Dummy Demographic Variables Table 2: Example of Choice Experiment Layout

Table 3: Example of Willingness to Pay for Chocolate Layout, Part 1 Table 4: Example of Willingness to Pay for Chocolate Layout, Part 2 Table 5: Demographic Characteristics of Respondents

Table 6: Correlation between Age, Income and Education Table 7: Self-Reported Importance of Variables

Table 8: Willingness to Pay Values, Excluding Demographics

Table 9: Results from t-test of Sum of WTP Compared to Total WTP Table 10: Price Levels Offered

Table 11: Willingness to Pay as Percentage of Average Price Table 12: Willingness to Pay Values, including Demographics

Table 13: WTP as Percentage of Average Price, including Demographics

Table 14: Willingness to Pay Values for Chicken – Age and Gender Characteristics Table 15: Self-Reported Preference Willingness to Pay

Table 16: Correlation between Price Importance and Most Important Variable Table 17: Results from Joint Significance Tests

Table 18: Information Criteria for Different Models

Table 19: Comparison of Results from Chocolate Experiments Table 20: Chocolate Brands Willingness to Pay

Table 21: Milka and Albert Heijn Gender Differences Table 22: Comparison of Linear and Quadratic Models Graph 1: Type Chosen by Product

Graph 2: Price Level Chosen by Product Graph 3: Scatterplot of Chocolate Brands

(4)

3

1. Introduction

Are consumers willing to pay higher prices for goods which are produced in a more ethically sustainable way? Which ethical factors most affect a consumer’s decision when purchasing food? I will run a survey to examine consumers’ hypothetical purchasing decisions when faced with different options of human rights, animal welfare and environmental standards involved during the production processes. Using the information collected from this survey, I aim to:

1. Analyse differences in consumers’ views towards human rights, animal welfare and the environment across different food products.

2. Estimate the willingness to pay for an increased value of each category and how this differs with individual demographic characteristics and across different products. Individuals use information available to them differently. Some consumers will go out of their way to seek additional information not readily available on how ethically produced goods are and factor this information into their decision making process. Others will simply choose products based on visible qualities, such as price, packaging or nutritional information printed on the product. When media attention is given to large shocks, such as revelations that certain producers rely on child labour to make their products, a normally unconcerned consumer may consider this information when making their next purchase. When individuals are directly presented with this ethical information, without the normally required searching costs, how do they respond? Existing literature suggests there is a group of consumers who have no desire to change their purchasing habits, whilst others will adapt based on the information provided to them. This research explores the issue by looking at how consumers respond when given information on the sustainability of the production process.

I am also interested in which production factors consumers value the most when choosing their goods. There are often trade-offs when selecting a product – do you prefer a good which is produced in a way which benefits the environment but pays less attention to animal welfare or would you prefer one which provides their animals with plenty of living space but is unable to pay their workers a good wage? Are consumers prepared to pay the same premium for goods which do not involve child labour, provide a fair living wage to workers and operate in a safe working environment as those products which provide good living conditions for their animals? Are there specific demographic characteristics which drive the differences in valuations?

Part of this research uses information produced by QuestionMark, a Dutch organisation with the objective of providing information to consumers about how sustainable their shopping habits are. The organisation uses scientists independent of the producers to calculate scores for different subcategories within human rights, animal welfare and the environment. Recently, they have also added information on how healthy a product is. These scores are

(5)

4

then averaged to give an overall sustainability rating for the product which is available to view and compare online or using one of their mobile apps. Given this independent and robust methodology, consumers can have a high degree of confidence that these scores are an accurate analysis of how foods are produced. The organisation was launched using funding from the Dutch Postcode Lottery in 2012 and a previous database by SuperWijzer. So far they have analysed over 36,400 products and have approximately 400,000 users. Results from the analysis showed that approximately 35% of respondents care about all variables equally. The choice experiment showed that consumers are prepared to pay an additional €0.61 for a one unit score increase for chicken, €0.25 for eggs and €0.10 for chocolate. The willingness to pay to increase each individual variable score is fairly similar across products, but when looking at choices made animal welfare was the most important for chicken, eggs and milk, with human rights found to be the least important. Of the chocolate brands tested – Ethiquable, Tony’s, Milka and Albert Heijn – only Ethiquable and Albert Heijn had a significant and positive willingness to pay, €0.14 and €0.08 respectively. This thesis is organised as follows. Existing literature and the theoretical background to conducting a choice experiment are considered in sections two and three. Section four presents the research questions and hypotheses for this research. In section five the methodology and survey design are explained. Section six presents the results and finally section seven concludes and offers discussion of the results along with suggestions for further research.

2. Literature Review

This section looks at existing literature on the development of ethical consumerism and previous studies which have investigated consumers’ attitudes towards ethical goods and the premium they often carry. Whilst there are numerous papers which look at individual attributes, there is limited research on comparing attributes within a product, which is the focus of this research.

2.1 Ethical consumerism

Ethical consumerism can be described as “buyer behaviour that reflects a concern with the problems of the Third World” (Strong, 1996). Examples of such problems are poor living conditions for workers, testing products on animals or the use of genetically modified organisms to increase crop yields. This altruistic interest in helping others is not a new phenomenon. Abolishing slave trade and workhouses, animal welfare concerns leading to the creation of the RSPCA and the 1948 UN Declaration of Human Rights were all driven by demands for change from consumers (Cowe & Williams, 2000).

A combination of pressure groups campaigning for fairer working conditions, media attention and an increased corporate interest in using Fair Trade production methods helped demand for sustainable products grow rapidly in the 1990’s and continues to grow

(6)

5

now. Vermeir and Verbeke (2006) explain that sustainable products are those which contribute through their characteristics and impact to achieving economic, social or environmental sustainable development strategies. A report by the Cooperative Bank (2014) found that the value of ethical spending on such products grew by 9% in 2013, aided by boycotts which followed the horsemeat scandal and the collapse of the Rana Plaza clothes factory. Chains such as the Body Shop became popular through marketing which highlights the lack of animal testing on their products; supermarkets like EcoPlaza and Whole Foods have emerged assuring the ethically concerned consumer that all products stocked meet certain standards; some airlines offer the option to make a donation to charity to counteract the flight’s carbon footprint. These examples suggest there is a demand for premium priced products with an ethical benefit. However, conflicting research suggests that people buy these products because they like the goods offered and not only because of the ethical attributes (Carrigan & Attalla, 2001).

The first research into consumers’ preferences for ethical or more sustainable products was conducted by the European Commission in 1997 (European Commission, 1997). They studied consumers in the then 15 member states to find out if they would be prepared to pay a premium for bananas produced in a Fair Trade method. The results were encouraging. On average 74% of the population responded that they would be prepared to buy Fair Trade bananas if they were available at the same price and over one third stated they would pay a premium. Results in awareness and willingness to purchase varied across member states, reflecting knowledge about Fair Trade products and preferences for home grown produce. Several studies have been conducted to test whether consumers demonstrate a preference for ethically produced goods. Zadek (1997) found that 86% of British consumers were “more likely to buy products positively associated with a social or environmental issue”, and 60% would be prepared to boycott products due to ethical concerns. Cortese (2003) estimated approximately one third of the US adult population could be classified as “lifestyles of health and sustainability consumers” who consider environmental and social issues when making their purchases. Literature suggests the increasing profitability of targeting ethical consumers and the potential for niche markets formed of products carrying certain messages (see for example Loureiro & Lotade, 2005 or Cortese, 2003). Carrigan and Attalla (2001) highlighted that although consumers cannot be proven to take ethical considerations into their purchasing behaviour there are other stakeholders who would, such as governments, shareholders or employees.

Other research suggests only certain consumers are interested in ethical shopping and prepared to pay a premium for such products. Didier and Lucie (2008) identified three groups of consumers when investigating consumers’ attitudes towards Fair Trade and organic chocolate labels. Nearly half of their respondents were insensitive to the labels and based their choice only on the price of the products. Cowe and Williams (2000) support this result, finding half of their respondents were concerned but not sufficiently to change their

(7)

6

behaviour. Among the other half, consumers could be separated into different types from those only concerned about value for money, to a group who are actively interested in such issues and those who use ethical information to form their decisions around how they view the brands.

Existing literature suggests many factors can be important when consumers are making purchasing decisions. Brand of the product, demographic characteristics of the individual such as age, income and education or frequency of purchase are just some of those which have been considered. Aprile et al. (2012) examined consumers’ willingness to pay for different labels commonly found on olive oil in Italy. Their main conclusion was that consumers consider all attributes when making a purchase, but are willing to pay the highest premium for labels specifying origin of the product.

2.2 Comparison of ethical attributes

There is limited research into comparing within a product which attributes consumers place the highest value on. Given that consumers consider many factors when making a purchase, determining which of those attributes is the most important to them requires specific, generally hypothetical questions.

Auger et al. (2003) calculated consumers’ valuations for different attributes in soap and trainers. They found that overall the respondents had a strong dislike of animal abuse in the soap production process and child labour when producing trainers, but did not compare attributes within the same product. They also concluded that in general, respondents gave quite high valuations to the ethical components of products, such as the fact a product was not tested on animals or produced in dangerous working conditions. Their research into socioeconomic characteristics found that although personality traits did not show a strong correlation with willingness to pay for ethical products, there was a link between certain demographic characteristics, such as education and income level.

Loureiro and Lotade (2005) used a face-to-face survey to gather information on consumer’s preferences for three types of coffee labels. They found that consumers were willing to pay the highest premium (21.64 cents/lb) for a Fair Trade label which demonstrates the coffee has been produced according to specific social and environmental agreements. For shade grown labels which show the product was made in an environmentally friendly method, consumers were also prepared to pay a similar premium (20.02 cents/lb). These results suggest that consumers are most concerned about the welfare of other humans, followed by environmental standards. Didier and Lucie (2008) find similar results when looking at chocolate in that some consumers consider a Fair Trade or organic label to increase the value of the product, whilst for others how much they like the product must also be considered.

(8)

7

3. Theoretical Background

In this section we look at the theory behind consumers’ choices and how utility maximisation can explain their decisions. We also consider the search costs involved in obtaining information on a product and how the scores calculated by QuestionMark can reduce this.

3.1 Decision Making Theory

I will analyse consumers’ preferences for different attributes within the choice modelling framework, based on random utility theory as developed by McFadden. He described the study of choice behaviour as:

1. “The objects of choice and sets of alternatives available to decision makers 2. The observed attributes of decision-makers

3. The model of individual choice and behaviour, and distribution of behavioural patterns in the population” (McFadden, 1973, p106)

We assume consumers are utility maximising. When presented with options of different products they will select the one which they expect will maximise the utility they gain from consumption of that product. For each product j, individual i will obtain utility Uij for j = 1, …,

J. The individual wants to maximise their utility and therefore chooses the alternative with the highest utility. Individual i chooses alternative k if and only if:

𝑈𝑖𝑘 > 𝑈𝑖𝑗∀𝑗 = 𝑘

The exact utility cannot be observed but some of the attributes of the product and characteristics of the individual can. Lancaster (1966) stated that goods do not give utility to the consumer, rather that the utility is derived from each of the attributes a good has. These are often shared between different goods. Combining goods may provide different characteristics compared to consuming each good individually.

3.2 Choice Experiment Analysis

From the responses to the choice experiment, a conditional logit model is used. This model takes account of the options not chosen, as well as the one chosen, in order to calculate the utility for each attribute and makes the assumption that the axiom of independence from irrelevant alternatives (IIA) holds. This means that the odds ratios between two alternatives remain constant when a new alternative is added. This is most likely to be violated when the alternatives offered are similar to each other, resulting in unobserved factors affecting more than one alternative (Hoffman & Duncan, 1988). Under the assumption that the error term is independent and extreme value distributed, we can assume that IIA holds (McFadden, 1973).

(9)

8

The utility individual i gets from consuming product j is defined as: 𝑈𝑖𝑗 = 𝛽′𝑥

𝑖𝑗+ 𝜀𝑖𝑗

Where 𝑥𝑖𝑗 are the alternative-specific regressors, 𝛽′ is the set of coefficients giving the

utility derived from each of the alternative specific regressors and 𝜀𝑖𝑗 is the error term which

is unobservable and assumed to be independent of other variables (Katchova, 2013 and Rodriguez, 2007). Not all aspects of utility can be observed. The individual may incorrectly estimate the utility they derive from the product and the analysts cannot observe the part utility gained from every attribute. The error term captures these aspects and is assumed to be independent and of Type 1 extreme value.

Specifically, in this analysis the utility of an individual is calculated as: 𝑈𝑖𝑗 = 𝛼𝑗+ 𝛽1 𝑇𝑜𝑡𝑎𝑙 𝑠𝑐𝑜𝑟𝑒 𝑗 + 𝛽2 𝑃𝑟𝑖𝑐𝑒𝑗+ 𝛾1 (𝐴𝑔𝑒𝑖× 𝑆𝑐𝑜𝑟𝑒𝑗)

+ 𝛾2 (𝐺𝑒𝑛𝑑𝑒𝑟𝑖× 𝑆𝑐𝑜𝑟𝑒𝑗) + 𝛾3 (𝐼𝑛𝑐𝑜𝑚𝑒𝑖 × 𝑆𝑐𝑜𝑟𝑒𝑗)

+ 𝛾4 (𝐶𝑜𝑢𝑛𝑡𝑟𝑦 𝑅𝑒𝑠𝑖𝑑𝑒𝑛𝑐𝑒𝑖 × 𝑆𝑐𝑜𝑟𝑒𝑗) + 𝛾5 (𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛𝑖 × 𝑆𝑐𝑜𝑟𝑒𝑗)

+ 𝛾6 (𝑄𝑀 𝑢𝑠𝑒𝑖 × 𝑆𝑐𝑜𝑟𝑒𝑗) + 𝛿1(𝑃𝑟𝑖𝑐𝑒 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑐𝑒𝑖 × 𝑆𝑐𝑜𝑟𝑒𝑗)

+ 𝛿2 (𝑀𝑜𝑠𝑡 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑡𝑖 × 𝑆𝑐𝑜𝑟𝑒𝑗) + 𝜀𝑖𝑗

where subscript i is for individual characteristics and j for the product attributes. The β coefficients are related to the product, the γ the demographic characteristics of the individual and δ for characteristics of the individual which demonstrate their stated preferences towards each of the variables. In this case, the alternative-specific regressors are demographic characteristics of the individual i, interacted with the score for the product j. There are also regressors which are only related to product j (price and total score). Instead of the total score, the individual breakdown scores (human rights, animal welfare and environment) could be included. The individual demographics are then interacted with each of the breakdown scores in order to see the impact each demographic has on each breakdown score. This allows for testing whether age has the same impact on the willingness to pay to increase the human rights, animal welfare or environment scores by one unit. We would then include the importance the individual places on each of the variables, as well as the price importance which is included with total score.

The demographic interaction variables are included as a dummy variable. The following table shows the definition of each dummy. From this we can compare if a male and a female gain the same utility from a higher scored product, for example.

(10)

9

Table 1: Explanation of Dummy Demographic Variables

Variable Dummy = 0 Dummy = 1

Age Under 35 Over 35

Gender Male Female

Income Under € 60,000 Over € 60,000

Education No university University educated

Country of residence UK Other

QM previous use No Yes

Each coefficient expresses how that variable affects an individual’s utility. For price, we would expect the coefficient to be negative as paying a higher price reduces the utility an individual gets from consuming that product. For a rational individual who cares only about their own private costs and benefits, the score coefficient would be zero. The sustainability score has no impact on their utility. A non-zero coefficient for the score implies that knowledge of an externality affects the consumer’s utility. In this case, they are partially internalising the external costs by demonstrating a utility gain or loss to the externality. If the coefficients differ between individual scores and for each product studied, it suggests that an individual values differently the same attribute in different products.

Once the coefficients are estimated, we can calculate a willingness to pay (WTP) for an improved score by one unit. This is calculated by dividing the coefficient of the relevant characteristic by the coefficient for price. This gives the additional price an individual is willing to pay to increase the relevant score by one unit. For each characteristic, this is:

𝑊𝑇𝑃 = − 𝛽 𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 𝛽 𝑃𝑟𝑖𝑐𝑒

The probability that an individual chooses a product is given by the probability that the individual will derive higher utility from that product, compared to all the other products also on offer. This can be shown as:

Pr (𝑈𝑖𝑗 𝑐ℎ𝑜𝑠𝑒𝑛) = 𝜋𝑖𝑗 = 𝑃𝑟(𝑈𝑖𝑗 > 𝑈𝑖𝑘)

Where Uik represents the utility the individual would gain from each alternative not chosen.

The probability an individual chooses each product will sum to 1, given that we assume they choose a product.

Assuming the error term is Type 1 extreme value, we can calculate the probability individual i chooses product j using exponentials of the estimated coefficients. This follows the logistic distribution and is calculated as:

𝜋𝑖𝑗 = 𝑒𝑥𝑝 𝛽

𝑥 𝑖𝑗

∑ 𝑒𝑥𝑝 𝛽′ 𝑥 𝑖𝑘

(11)

10 3.3 Search costs

Not all attributes of a product are immediately obvious to the consumer, nor the utility a consumer would derive from them. Some, such as price or nutritional values, may be displayed and can easily be observed. Others may be known from experience, such as taste. However, additional information may require search costs by the consumer. The extent to which a firm cares for the environment, the lifestyle quality of the animals involved in making the product or the workers employed by them is not easy to identify.

For a rational consumer the optimum search time is found by equating the cost of search to the expected marginal return (Stigler, 1961). The cost can be defined in time, money or opportunity cost. The benefit could be a cheaper price or non-monetary benefits such as improved health or knowledge that the products were made sustainably. Therefore, a consumer who is only interested in the price of the product will exert no effort in finding information about other attributes. On the contrary, a consumer who places a high value on purchasing ethically produced goods will exert a large amount of effort to gain information on the ethics of the companies they choose to purchase from.

Russo et al. (1986) studied the impact of nutritional information available to consumers and found the costs could be defined as three types of effort: collection, computation and comprehension. To change purchasing behaviour a modification in the cognitive effects is needed, for example introducing health traffic light ratings on food products to reduce the effort required to understand information. When processing new information the consumer may change their behaviour due to an increase in perceived benefits or a reduction in search costs. Examples include an improved understanding of living on low wages increasing the perceived benefits of buying products which pay fair wages or the ratings provided by QuestionMark reducing search time by allowing consumers to easily compare products.

4. Research Questions and Hypotheses

This section presents the research questions that I intend to answer using this survey. It also states the hypotheses which will be tested using the data collected.

4.1. Research Questions

In this study, I will combine different factors to compare the importance given by consumers to human rights, animal welfare and the environment. To my knowledge, these three areas have not previously been tested together which will give new insight into which areas concern consumers the most. This analysis will allow comparison of the value consumers place on each attribute and their relative importance to one another. This may be of interest to producers who could then be encouraged to improve their production methods in order to attract more consumers or potentially charge a higher price to compensate for higher production costs. It could also influence government policy as more information available to consumers allows them to make better informed decisions and could help shape policy to encourage more sustainable production techniques.

(12)

11 I will answer the following research questions.

1. Do differences exist between the importance consumers place on human rights, animal welfare and environmental factors during production?

2. Will different products receive different valuations from consumers?

3. Are higher income individuals, better educated or consumers previously interested in ethical shopping prepared to pay more for products with more ethical production methods?

4. Is there a difference between chocolate brands in the willingness to pay for a higher overall sustainability score?

4.2 Hypotheses

Aprile et al. (2012) found that consumers consider all attributes when making a purchasing decision. Therefore, my first hypothesis is:

H1. Each variable (human rights, animal welfare and environment) will be considered equally

important.

Consumers view products differently. Products which can be easily related to the variables may receive a higher valuation for that variable. For example, it is easy to see how improved animal welfare can benefit a chicken but the link may be less obvious for a good which uses a lower proportion of animal products. The price of the product also affects the percentage premium consumers are prepared to pay. Elliott and Freeman (2001) found that consumers indicated a willingness to pay an additional 28% for 10$ items but this fell to 15% for 100$ products. Therefore, my second hypothesis is:

H2. Different products will receive different valuations as a percentage of the mean price for

each variable.

Some research (Cortese, 2003) has found that consumers are prepared to pay a premium for products with a higher standard of production. Evidence is mixed around which characteristics this type of consumer is likely to have. Several papers have suggested that these consumers are better educated and with a higher income than average (for example Auger et al., 2003, Blend and Van Ravenswaay, 1999), highly educated and of a high socioeconomic status (Webb and Mohr, 1998) or older (Didier and Lucie, 2008). Due to the multiple references, my third hypothesis is:

H3: Highly educated and those with a higher income will be prepared to pay a larger

premium for more ethically produced goods than other groups of consumers.

Part of the sample will consist of people who have previously expressed an interest in purchasing ethically produced goods. This will allow for comparison between general consumers and those who are already involved in searching for information about the ethics of the products they are purchasing. My fourth hypothesis is:

(13)

12

H4: Consumers who have previously expressed an interest in finding out information about

the ethics of their products will be willing to pay more for products with higher scores.

Chocolate is a less homogeneous product than the others being studied. I will therefore include additional questions looking at whether consumers’ attitudes towards paying more for an improved sustainability score are affected by the brand. My final hypothesis is: H5: The impact on consumers’ willingness to pay for chocolate on learning the actual score

will differ by brand.

5. Methodology

In this section a detailed description of how the dataset was collected is presented. It begins with some well-known limitations of using a hypothetical survey to collect responses and how this may affect our results. The calculation of scores by QuestionMark used in this research is then described. Next, how the survey was conducted is explained and the design of the choice experiment. Finally, the additional questions included on chocolate brands are explained, along with information on each of the brands used.

5.1 Survey Limitations

Hypothetical bias

It is well known that consumers often respond to surveys in a way which differs to how they would actually behave. Carrington (2010) describes three reasons which may prevent consumers “walking their talk” - implementation intentions, actual behavioural control and situational context. Respondents may provide answers they believe are socially desirable or what the researcher wants; they may incorrectly predict the situation they will next find themselves in, such as availability of products or amount of immediate disposable income or consumers may simply forget their previous intention to purchase a certain type of product. This attitude-behaviour gap has been studied by many authors to try to find ways to reduce the differences. Loomis (2013) suggests ex-ante and ex-post solutions which can be applied to the experiment to reduce this bias. He suggests informing respondents of the impact of their decisions, requesting their honesty to prevent biased results or explaining that an attitude-behaviour gap often exists. This method of “cheap talk” reduced or removed the bias in six out of eight papers he studied, with one over-correcting and in the final paper having no effect. Ex-post, he proposes data screening or comparing results to other non-hypothetical research and adjusting by an appropriate factor.

In order to reduce the hypothetical bias in this survey, I included a combination of cheap talk and realism approaches to encourage respondents to answer truthfully. This means that respondents were told that in previous studies participants have been shown to give overestimates and ask them to give realistic answers so as not to bias the results.

(14)

13

Respondents were also informed that all answers are anonymous, therefore reducing the impact of trying to give a socially desirable response (see Appendix 2 for the script included).

Sample selection bias

The online survey was circulated to a distribution list of current QuestionMark users and on their social media platforms. I also contacted friends, previous colleagues and family members to complete and share the survey. The sample is not representative of the general population and may therefore present some bias in the results. Those contacted by QuestionMark have already expressed an interest in ethical and sustainable shopping. Similarly, the contacts I have are likely to fall into certain demographic groups which do not represent the whole population. Therefore results should be interpreted alongside the demographic characteristics recorded and the limited extrinsic validity of the results should be noted. The survey could be repeated across a wider sample which has the same demographic proportions as the total population to increase the external validity of the results.

There may have been a self-selection bias as those more interested in ethical shopping are more likely to opt in to do the survey and complete all the questions. For this research, it was not possible to force people to complete the survey which may result in an overestimation of the importance of the attributes. However, the overall aim of this research is to compare between the attributes which is the most important. I do not consider it likely that people specifically interested in one area are more likely to drop-out than others. Therefore, the main result which may be affected by the sample bias is the comparison of price to other attributes.

Controlling for External Influences

Respondents were informed that the alternatives offered differed only in the attributes for which information was provided, namely the price, total score and breakdown of scores for human rights, animal welfare and environmental standards. It is possible that some respondents considered price to be a proxy for quality, and therefore assumed a higher priced product was also of better quality. Similarly, feedback from some respondents said that they often include how healthy a product is perceived to be in their decision, for example choosing milk which has a better environmental score because it is less processed, and therefore perceived to be better for the individual’s health. It is difficult to test to what extent respondents took other factors into account but is an additional limitation of using this method.

5.2 Calculation of QuestionMark Scores

QuestionMark uses information publicly available on the product, such as ingredient lists and labels, as well as requesting additional information from the manufacturers of the products it is researching. The score calculations also use databases compiled by other organisations, such as the OECD, International Labour Organisation and scientific experts to

(15)

14

calculate the risks in each area. Their method is publicly available and they work with specialists to improve their approach.

The scores are calculated using information on the ingredients, manufacturing method and any quality marks1 the product has been awarded. The product is then assessed across its whole life cycle from cultivation of the ingredients to the point of sale. This includes manufacturing of the ingredients, transportation of the interim and final goods, processing the product, the packaging it is sold in and the storage conditions that the good requires. Each variable is assessed individually and the overall product risks are then combined. Users of the app can assign a weighting to each variable based on their preferences. For other purposes (including this research), the total score is the rounded average of each of the three variables. A summary of how each variable is calculated is given below. See Appendix 1 for the specific formulae used and for further detail consult the “Method” pages on QuestionMark’s website (TheQuestionMark.org).

Human Rights

Given the difficulty in determining the specific human rights situations at each point in the manufacturing process, the QuestionMark methodology instead focuses on the risk of violations of human rights that occur and how these have been mitigated. They use databases from respected organisations to assess the risk on a scale of one to four. Various formulae are then used which take account of efficiency of a measure, focus placed on that measure and the risk rating for each subtheme.

The ingredients are weighted according to the proportion they are used in the product. Products receive a bonus if the suppliers are demonstrating “positive change”, for example having a quality mark in an area where it represents acknowledgment of reducing the risk in high risk areas.

Animal Welfare

The calculation method for animal welfare is based on the Better Life Quality Mark system which is used for Dutch producers. The welfare of the animals is considered at different points during the production process and has been grouped into five themes: physical impact, housing, feed, transport and slaughter. They are then weighted at the final calculation stage, with physical impact and housing receiving the highest weighting. Each of these themes then has smaller sub-themes, for which there are measurable indicators and a level is given.

The criteria vary depending on the animal being considered. As an example, housing can be split into living space, free range and other. A chicken holding would receive the minimum score if there are nine or more animals per metre squared, an average score for seven to

1

A quality mark guarantees a special characteristic of a product which gives added value, such as produce from better kept animals, organically produced or using Fair Trade methods. (FoodHolland.nl, 2016)

(16)

15

nine animals and the maximum score if there are fewer than seven. These levels are then given a score of two for the highest level, one point for the intermediate and none for the lowest level.

The final score is calculated as a relative score from one to ten. It is assumed that products containing no animal products score a ten, and consequently the highest a product which contains livestock can score is nine.

Environment

Information on each ingredient is collected from a variety of sources – literature, experts and the companies directly. This is complemented by data from Ecoinvent, a not-for-profit organisation which supports companies to manufacture products in an environmentally friendly way and consumers and policy makers to make informed decisions. Their database offers information on the processes used in different products (Ecoinvent.org). The information is input into a specific software system, SimaPro, which can reduce many variables into a few specific indicators using the ReCiPe method, available in the software. This software looks at the causes and effects which combine to impact on three areas: damage to human health, ecosystems and the availability of raw materials. These final indicators are used to calculate an overall environment score for each ingredient which is then used by QuestionMark. Scores are adjusted for any quality marks awarded and weighted by the percentage of the ingredient used in the product.

5.3 Conducting the survey

The survey contained background information explaining the independence of ratings produced by QuestionMark and a brief outline of what each category represents. There was also some “cheap talk” encouraging respondents to answer truthfully. At the end of the survey, there were demographic background questions on age, gender, income, education obtained and country of residence. Respondents were asked to state how much they care about human rights, animal welfare, the environment and price when they are making purchasing decisions, which they consider to be most influential on their purchasing behaviour and also the extent to which they have previously used QuestionMark. An abridged version of the survey can be found in Appendix 2. Before conducting the full survey I carried out a few pilot versions. This helped determine how many questions could be answered in ten minutes to reduce the drop-out rate, decide the price and score intervals between options and provide some initial feedback on the research method.

The survey was conducted using the online platform Qualtrics. This platform is regularly used by students, researchers and companies conducting market research. It allows for unlimited questions, surveys and responses, and can export data in a useable format (csv file). Randomisation of which questions are presented and their order allows for a wider data set to be collected and reduces any potential bias from order or learning effects. Qualtrics also has a “question skip” option which means the survey was designed so that

(17)

16

respondents only answer questions on products they regularly purchase, for example a vegetarian would not buy chicken. However, this means that their value for that item will not be captured in the survey. Both Dutch and English versions were circulated where the questions and design were identical except for the language.

5.4 Choice experiment

When consumers purchase a good they consider several characteristics of the products offered and hence a multi-attribute approach is the most appropriate. I used a choice experiment approach where respondents were presented information on 21 products with fictional attribute combinations and then asked to select which one they would buy in a supermarket scenario. Respondents were presented with information on price, overall score, and each of the individual rankings in a table similar to the one below. They were then asked to consider the information carefully and select their preferred product.

This approach represents a realistic scenario which consumers face when purchasing food products where there are several varieties of products on offer. There may have been some information overload as respondents had to understand the differences between scores and decide how they value them. However, presenting the information in a simple table which has the same format for each question should have reduced the effort and time required for respondents to make their choices. Feedback from pilot surveys suggested this was easier to understand than repeated questions with fewer options.

The table below shows an example of the options presented to a respondent. As can be seen, there are seven choices for each price level and overall score. It is assumed that a respondent makes their decision in two stages. First, they choose the price and overall score of the product they wish to buy. Second, they choose which combination of the three attributes human rights, animal welfare and environment they prefer most. For this reason, the information is presented with the price and overall score in the left-hand columns, followed by the individual attributes. It is possible that not all respondents make their decision like this and the order which information is presented has an effect and should be randomised to avoid people only reading the first columns. However, I think this would have increased the cognitive burden placed on respondents and caused unnecessary additional confusion. Presenting each question in the same informational layout allowed respondents to become familiar with the format and spend more time thinking about their decisions instead of on understanding the information presented to them.

(18)

17

Table 2: Example of Choice Experiment Layout

Chicken Price Overall score Human rights Environment Animal welfare

1 €3.50 4 10 1 1 2 €3.50 4 1 10 1 3 €3.50 4 1 1 10 4 €3.50 4 4 4 4 5 €3.50 4 5 5 2 6 €3.50 4 5 2 5 7 €3.50 4 2 5 5 8 €4.50 6 8 5 5 9 €4.50 6 5 8 5 10 €4.50 6 5 5 8 11 €4.50 6 6 6 6 12 €4.50 6 7 7 4 13 €4.50 6 7 4 7 14 €4.50 6 4 7 7 15 €6.50 8 10 7 7 16 €6.50 8 7 10 7 17 €6.50 8 7 7 10 18 €6.50 8 8 8 8 19 €6.50 8 9 9 6 20 €6.50 8 9 6 9 21 €6.50 8 6 9 9 5.5 Variables

I tested the variables human rights score, animal welfare score, environment score and price. Each score is a value between 1 and 10, and the price is numerical. I tested four different products which have different monetary values: chicken, eggs, milk and chocolate. This allowed for estimations of how consumers view different products and also how much they are prepared to pay when the price of items differs. They are commonly purchased, so will be fairly familiar to respondents and, with the exception of chocolate, are fairly homogeneous. Products which differ greatly in taste, quality or brand impact may be an important factor in consumers’ preferences but would not be captured in this survey.

Product scores and prices were hypothetical because the aim of the survey is to maximise the amount of information gained which is best done through hypothetical combinations of variables. In reality, it would be difficult to find products with the specific combinations and prices required to test for incremental changes.

5.6 Willingness to Pay for Chocolate

As well as the choice experiment questions, respondents were presented with two questions directly asking for their willingness to pay for four different chocolate brands.

(19)

18

They were asked to estimate the overall sustainability score for the brands Ethiquable, Tony’s, Milka and Albert Heijn’s basic and give the maximum they would be prepared to pay for that chocolate in a table similar to the one shown. Pictures of each packaging were included to give an indication of the company’s marketing incentives and increase the link between branding and changes in scores. By requesting the willingness to pay for each brand both before and after information on the true score is revealed, brand preferences are controlled for. Changes in the willingness to pay should be due to consumers placing a higher value on the higher sustainability score, or vice versa. If a consumer perceives one brand to have a higher value than another, this will be reflected in their initial value. The change in value will show how much the individual values the sustainability score as that is the only change in information between the two questions.

Table 3: Example of Willingness to Pay for Chocolate Layout, Part 1

Chocolate brand Estimated score Price

Ethiquable

Tony’s

Milka

(20)

19

The chocolate brands used have a range of priorities in their production processes. Ethiquable is a certified Fairtrade and organic producer, founded in 2009 as a cooperative enterprise (DiscoverBenelux.com, 2016). Their production methods are environmentally friendly and aim to support small scale farming. Tony’s Chocolonely was founded in 2005 with the ambition of creating a “100% slave free” chocolate industry. This is done through long-term relationships with the cocoa farmers and suppliers, a transparent production process and donations to the Chocolonely Foundation which supports related projects (TonysChocolonely.com, 2016). Milka is a Swiss brand which has been producing since 1826. It is popular across continental Europe and since 2012 has been part of the initiative “Cocoa Life” which fights for a more sustainable future of cocoa production and farming communities in six cocoa producing countries (Milka.es, 2015). Albert Heijn supermarkets offer many products with their own branding as a cheaper alternative. Founded in 1887 (Ahold.com, 2016), the supermarkets launched its BASIC range in 2013 in response to growing demand from consumers focusing on the price-quality ratio (Ahold.com, 2013). Respondents were then informed of the actual score each brand has been given by QuestionMark and again asked for the value they place on each.

Table 4: Example of Willingness to Pay for Chocolate Layout, Part 2 Chocolate brand Score New Price

Ethiquable 8

Tony’s 7

Milka 4

Albert Heijn 7

The order respondents were presented with each chocolate was randomised to prevent order effects impacting results. Whilst respondents were able to refer back to the information they provided in the first set of responses, they were urged not to change them. Allowing them to refer back to their estimated scores prevented respondents having to remember the scores and prices they gave and reduced the need to remember initial responses given.

A linear regression was used to estimate the willingness to pay for the chocolate brand questions included. This was done by using the difference between price estimates as the dependent variable and the difference between estimated score and actual score as the independent variable. The regression was considered both with and without demographic variables. The coefficient for score difference tells us how much consumers are willing to pay to increase the score by one unit.

(21)

20

6. Findings

In this section findings from the analysis are presented. The dataset is described along with some key descriptive statistics. The results from the choice experiment are then studied in three ways: analysis of the choices respondents made, willingness to pay estimates excluding demographic characteristics and the same estimates including demographic characteristics. Finally, the responses from the additional chocolate questions to calculate willingness to pay estimates for each brand studied are analysed.

6.1 Dataset

Survey responses were collected between May 17th and June 6th, 2016. 448 people began the survey and 303 completed it, giving a drop-out rate of approximately one third. Respondents were asked how frequently they buy chicken, eggs, milk and chocolate and for each that was purchased at least once a month, they then answered two choice experiment style questions. Including those who did not complete the full survey, this resulted in 577 observations for chicken, 668 for eggs, 639 for milk and 664 for chocolate. Respondents were also given questions on their willingness to pay for four different chocolate brands. This provided 741 observations across the four brands.

The following table shows the breakdown of demographic characteristics of the respondents.

Table 5: Demographic Characteristics of Respondents

Age Education Income

Under 18 1% Less than high school 1% Under €20,000 6%

18-21 5% High school graduate 4% €20,001-€39,999 25%

22-24 18% Some college 8% €40,000-€59,999 28%

25-34 40% Bachelor's degree 44% €60,000-€79,999 22%

35-44 16% Master's degree 35% €80,000-€99,999 14%

45-54 12% Doctorate 7% €100,000-€124,999 3%

55-64 6% Prefer not to say 1% €125,000 or over 1%

65+ 3% Prefer not to say 1%

Prefer not to say 1% Previous QM use

“Heard of Question Mark” Country

Gender No 83% UK 66%

Male 43% No, but I visited the website 2% Netherlands 25%

Female 56% Yes, but not used 7% Other 9%

Prefer not to say 1% Yes, used info 7%

According to the OECD, in 2012 41% of the UK population aged 25-64 had completed tertiary education (OECD Country Note, 2014) and 32% of the Dutch (OECD Country Note, 2013). In comparison this sample is therefore better educated than the average (in total 86% have university education). The average income in 2015 was approximately €24,000 in the

(22)

21

UK and €25,000 in the Netherlands (OECD Better Life Index, 2015) meaning this sample has a higher than average income.

There may be some correlation between a respondent’s age, income and education level which could affect the results, for example an older person may be further along their career and therefore also have a higher income. This could affect results by making one variable look significant when the decision is actually driven by a different characteristic. The following table suggests, however, low correlation between age, income and education.

Table 6: Correlation between Age, Income and Education

Age Income Education

Age 1

Income 0.324 1

Education 0.1567 0.2426 1

Respondents were asked to rate on a scale of one to ten overall how important they consider each of human rights, animal welfare, environment and price to be when purchasing a product. They were also asked to choose which of these is the most important. Some respondents commented that their preferences may differ between products as they place more importance on each variable for different products. However the responses, shown in the following table, can be used as a general indicator of preferences. Price was considered the most important by approximately half the respondents, in line with research by Didie and Lucie (2008). Price was also given a higher importance value when ranked individually. Among the other variables, the averages were similar but more people considered animal welfare to be the most important factor (21% compared to 13% and 14% respectively for human rights and environment).

Table 7: Self-Reported Importance of Variables

Variable Average (1-10) Most Important

Human rights 6.30 13%

Animal welfare 6.54 21%

Environment 6.54 14%

Price 7.49 51%

6.2 Type Analysis

In each question respondents were offered a choice of 21 products. These can be grouped into seven “types” of product: Care most about human rights (1), animal welfare (2) and environment (3); care equally about all variables (4); care least about environment (5), animal welfare (6) and human rights (7). These seven types of product were offered at three different price levels in each question that the respondent answered, with a higher price corresponding to a higher overall ethical production score. These hypothetical choice

(23)

22

questions forced the respondent to make a trade-off between the levels of importance they place on each of the variables which would not normally occur.

Approximately 35% made choices which suggest they care about all variables equally (type 4). This is in line with research suggesting people consider many different characteristics when making choices (Aprile et al., 2012) and confirms there are a group of consumers for which hypothesis one holds: each variable (human rights, environment and animal welfare)

will be considered equally important.

Among products with a high degree of animal ingredients, the most commonly chosen type had high animal welfare scores compared to human rights and environment (19% chicken, 27% eggs, 14% milk) followed by less concerned about human rights (14% chicken, 17% eggs, 15% milk). For chocolate, which uses less animal products and therefore has a weaker link between a desire to improve animal welfare and consumption of the good, higher human rights score was the most commonly chosen product, after all equal. The following graph shows the distribution of types chosen for each product.

Graph 1: Type Chosen by Product

Graph 1 shows the percentage each “Type” was chosen for each product. The mode was always Type 4 – all variables equally important.

0 10 20 30 40 0 10 20 30 40 Chicken Chocolate Eggs Milk HR highest AW highest E highest

All equal E lowest

AW lowest HR lowest T yp e Pe rce n ta g e

(24)

23

Overall respondents were willing to pay higher prices for higher total scores. This suggests there is a group of consumers who are willing to pay a premium for more sustainably produced goods. In theory, the only difference between the products is the price and the sustainability score, suggesting that consumers who pay the higher price wish to reduce the negative externality produced by consumption and are willing to pay in order to do so. It is noticeable that the mode for each product, except chicken, is for the highest price level. This may be because prices although based on market prices may have been lower in the lower range of what consumers would usually expect to pay. For chocolate, 50% of respondents opted for the highest priced option. This may be because an error in the question stated that the prices offered were for 500g of chocolate, although they were based on the market price for 100g. If respondents noticed this mistake, the prices would seem very good value for 500g and respondents would likely be prepared to pay the highest price of €1.19. The following graph shows the percentage of each product chosen at each price level.

Graph 2: Price Level Chosen by Product

Graph 2 shows the percentage of each price level that was chosen for each product. With the exception of chicken, the most commonly chosen was always the highest price level.

33 39 28 26 24 50 32 32 36 29 31 40 0 20 40 60 80 1 0 0

Chicken Chocolate Eggs Milk

Low Price Medium Price High Price

p

e

rce

n

t

(25)

24 6.3 Relationship between Variables

Each product choice gives information on the price, total score, human rights score, animal welfare score and the environment score. The survey was designed so that the total score is a function of the price. Choices with a higher price have a higher total score compared to the other options for that product. This may not always hold in reality as other attributes which affect price but are not included in the survey may differ, such as quality. However, for the purpose of this analysis where only information on the price and scores were included, we assume a higher score requires higher production costs and therefore commands a higher price.

The rounded average of the three scores gives the total score, following the QuestionMark methodology. The analysis will therefore focus on using either the total score to indicate the overall score level compared to price, or the breakdown of the three. This allows us to test whether consumers are influenced differently by the total score and each individual score. 6.4 Results Excluding Demographic variables

We first consider the results excluding the demographic variables. This makes the assumption that demographic characteristics are not significant in determining an individual’s preferences for sustainable production methods and matches findings by Roberts (1996). He found that preferences were better explained by a scale to measure “responsible consumer behaviour”, such as an individual’s recycling habits. In this survey, it could be measured by questions on the self-reported importance placed on each variable and the variable the respondent considered the most important. This is considered in the next section along with the demographic characteristics.

The following table shows results of the willingness to pay estimations, excluding the demographic variables. For chicken, all the scores are significant except for an increase in the environment score, suggesting that consumers are only prepared to pay to increase the animal welfare and human rights scores. For eggs, all of the values are significant, with the value given to human rights and environment similar to each other but lower than the value for animal welfare. For milk, only human rights and environment are significant, suggesting consumers place less importance on improving animal welfare in dairy farms. This could be because consumers consider dairy farms to already have high animal welfare standards compared to poultry. There has recently been publicity about dairy farmers not receiving a fair price for their milk as supermarkets competing over milk prices have driven down the price paid to farmers (Barret & Shubber, 2015). This could partly explain why human rights are valued the most for milk as consumers are more aware of the low wages dairy farmers are currently earning.

(26)

25

Table 8: Willingness to Pay Values, Excluding Demographics

Significance level: ***1%, **5%, *10%. Red text indicates negative values. Standard errors are shown in parentheses. The total score value represents how much respondents are willing to pay to increase the overall score. The individual breakdowns show how much consumers value an increase in each individual score, also by one unit. To increase the total score by one unit, on average each individual score would also need to increase by one unit2. Therefore, summing the individual breakdowns gives a similar price to the total score. If the value for total score is higher than the sum, this suggests consumers value a higher overall score more than an increase in each variable. If the total is less than the sum, consumers value increases in each specific variable more.

A t-test shows we cannot reject that the difference is significant for each product. 𝐻0: 𝑊𝑇𝑃 (𝑇𝑜𝑡𝑎𝑙 𝑠𝑐𝑜𝑟𝑒) = 𝑊𝑇𝑃 ( 𝐻𝑅 + 𝐴𝑊 + 𝐸 𝑠𝑐𝑜𝑟𝑒𝑠) 𝐻1: 𝑊𝑇𝑃 (𝑇𝑜𝑡𝑎𝑙 𝑠𝑐𝑜𝑟𝑒 ≠ 𝑊𝑇𝑃 (𝐻𝑅 + 𝐴𝑊 + 𝐸 𝑠𝑐𝑜𝑟𝑒𝑠)

Table 9: Results from t-test of Sum of WTP Compared to Total WTP

Total score WTP Sum scores WTP t-statistic Pr( |T|>|t|)

Chicken € 0.61 € 0.66 0.40 0.69

Eggs € 0.25 € 0.25 0.24 0.81

Milk € 0.05 € 0.05 -0.04 0.97

Chocolate € 0.10 € 0.08 -0.97 0.33

Red text indicates negative values.

Given the range in the value of the products, it is also worth considering the estimates as a percentage of the total price. In the questions, respondents were given the option of the following price levels:

2

Consider as an example a product with a total score of 7 calculated by the average of 6 for human rights, 8 for animal welfare and 7 for environment. To increase the total score to 8 would require each variable to increase by one unit (the mean of 7, 9 and 8).

Score Total Human Rights Animal Welfare Environment

Product Chicken € 0.61 *** (0.092) € 0.15 ** (0.058) € 0.40 *** (0.073) € 0.11 (0.069) Eggs € 0.25 *** (0.019) € 0.07 *** (0.006) € 0.12 *** (0.019) € 0.06 *** (0.007) Milk € 0.05 (0.049) € 0.03 *** (0.008) -€ 0.00 (0.035) € 0.02 *** (0.009) Chocolate € 0.10 *** (0.015) € 0.01 (0.010) € 0.05 *** (0.003) € 0.03 *** (0.006)

(27)

26

Table 10: Price Levels Offered

Product Low price Medium price High price

Chicken (500g) € 3.50 € 4.50 € 6.50

Eggs (6, free-range) € 1.29 € 1.79 € 1.99

Milk (1l, semi-skimmed) € 0.85 € 0.99 € 1.19

Chocolate (500g)3 € 0.89 € 0.99 € 1.19

The following table shows the willingness to pay for an increased score as a percentage of the mean price level for each product.

Table 11: WTP as Percentage of Average Price, excluding Demographics

Mean price Total WTP Human Rights Animal Welfare Environment

Chicken € 4.83 13% 3% 8% 2%

Eggs € 1.69 15% 4% 7% 4%

Milk € 1.01 5% 3% 0% 2%

Chocolate € 1.02 10% 1% 5% 3%

The red text represents estimates which were not found to be significant.

As can be seen, the percentage value for total willingness to pay is approximately equal for chicken and eggs, slightly lower for chocolate and was not found to be significant for milk. The similarity for chicken and eggs suggests that consumers are prepared to pay a portion of the total price, as opposed to a monetary amount, to improve the sustainability of food produced from chickens. This is contrary to research by Elliott and Freeman (2001) who found that consumers are prepared to pay a lower proportion of the price for items which already hold a high value, although their results used a much wider price range. The percentage respondents were prepared to pay for chocolate was slightly lower, suggesting that consumers take other attributes into account when making their choice and that the sustainability is less important.

For human rights, the breakdown was similar across the products – approximately 3% of the value of each product. Animal welfare was high for chicken, approximately 8% of the average value, but lower for chocolate (5%). This matches findings by the type analysis which suggested options with a high animal welfare score were preferred in products with a more obvious link to animal welfare. However, the percentage increase for animal welfare in milk was 0%. Finally environment scores were also of a similar value for all the products, approximately 3%. This shows as a proportion of the price, the environment is valued equally across products.

From these values, it would appear that for each product the variables are valued in the same order. Animal welfare (when significant) always has the highest percentage of the

3

As previously mentioned, the chocolate prices were market prices for 100g but respondents were misinformed that the price was for 500g.

(28)

27

willingness to pay, followed by human rights and then closely followed by environment. This would suggest we should reject hypothesis two: different products will receive different

valuations as a percentage of the mean price for each variable.

6.5 Results Including Demographic Characteristics

We now consider whether demographic characteristics affect willingness to pay estimates. The model used is the same as previously but now also includes individual specific demographic characteristics. The individual specific characteristics used a dummy variable for each demographic, interacted with the either the total score of the product option or each breakdown score (see table 1 for dummy variable explanations). The baseline is for a male, under 35, earning less than €60,000 per year, with no university education, living in the UK and who has not previously heard of QuestionMark.

Including the demographic variables in the regression produces the results in the following table. The demographic and self-reported preferences that were found to be significant are also shown. Full results tables can be found in Appendix 3.

Table 12: Willingness to Pay Values, including Demographics

Product Total Human

rights

Animal Welfare

Environment Demographics Preferences

Chicken € 0.88 *** (0.142) -€ 0.06 (0.089) € 0.30*** (0.080) € 0.25*** (0.074) Age Gender Most imp. Price imp. Eggs € 0.19 *** (0.034) -€ 0.03 (0.058) € 0.15 *** (0.050) € 0.06 ** (0.032)

Age Most imp.

Price imp. Milk € 0.14 *** (0.025) -€ 0.00 (0.030) € 0.07 * (0.039) € 0.04 * (0.022)

Country Most imp. Price imp. Chocolate -€ 0.47 (1.181) -€ 0.00 (0.038) € 0.01 (0.034) € 0.04 (0.029) None None

Significance level: ***1%, **5%, *10%. Red text indicates negative values. Standard errors are shown in parentheses. As can be seen from the asterisks, none of the variables for increasing the human rights score or the estimates given for chocolate are significant. This suggests when including demographic and preferences for individuals, respondents are not prepared to pay significantly more to increase the scores in these areas. Human rights was also the least frequently answered (13%) to the question of “Which variable do you consider to be the most important”, suggesting that consumers prefer better animal welfare or environmental standards of production compared to human rights for the products studied. This could be because the link is more obvious between production and animal welfare or environmental compared to human rights. If consumers wish to support better human rights they may find other ways to do so, such as campaigning for higher wages for producers, which they may consider to have a more direct impact than purchasing products with a good human rights score. Of the three variables - human rights, animal welfare and environment - human rights is possibly the least well defined. Information included at the beginning of the survey gave

Referenties

GERELATEERDE DOCUMENTEN

1) For each illegal image in the public database D, an extractor uses the helper data h in order to com- pute a hash-extract hext that is then matched (in a matching algorithm) with

televisiereclames een positiever effect heeft op kijkers dan recente populaire muziek, omdat er met gedateerde populaire muziek meer nostalgische herinneringen worden opgeroepen

MalekGhaini, “Effect of friction stir welding speed on the microstructure and mechanical properties of a duplex stainless steel,” Materials Science and

Specifically, the aim of this study was to expand the literature on the topic of trivial product attributes, by investigating consumers’ willingness to pay, including the

A global-local analysis method and, sub-structuring and reduction methods are applied to generate dif- ferent fidelity models from a given coarse meshed global FE model. These

These results indicated a positive effect of feed supplemented with SMS on the meat as higher amounts of the unsaturated fatty acid oleic acid was observed for the experimental

2012. Consumer Knowledge: Effects on Evaluation Strategies Mediating Consumer Judgments. Journal of Consumer Research, Vol. Promoting Sustainable Consumption: Determinants of Green

In addition, the suitability of two specific methods (open- ended contingent valuation and choice-based conjoint analysis) for measuring WTP for a relatively high-priced,