Bachelor Thesis International Marketing Programme

Carbon Labeling

A quantitative study of what the preferred content, design and layout is among Swedish consumers

Bachelor Thesis, 15 credits

Halmstad 2020-05-19 Eric Sundberg, Edvin Elghag HALMSTAD UNIVERSITY

Abstract Global warming has been a topic of discussion since the discovery that man-made emissions is having an affect on the planet almost 50 years ago. Grocery products stand for roughly one third of all EUs carbon emissions as a result of its high- volume production. Carbon labeling is a tool in which retailers and manufacturers can communicate the amount has caused throughout its whole life cycle or that the they are working towards lowering their GHG emission throughout their organization. However, previous research indicates that carbon labels has not yet had its breakthrough moment yet due to the CO2e data is too complex for the consumers to interpret. This led to our research question “What is the preferred content, design and layout of a carbon label among Swedish consumers? “The purpose of this study is to get a deeper understanding into the preferences of the Swedish consumers and what kind of attributes they are looking for to make a carbon label understandable. We found that the Swedish consumer prefers a more complex design than previous studies have suggested based on researches made from other countries. In order to do our explanatory research, we measured these variables with a quantitative survey and made statistical calculations such as mean values and correlation analysis to see if our hypotheses were supported. The analysis shows that the Swedish consumers prefer all the following attributes that is being presented in an order of priority: The label should be colour coded, made by a well-known organization, presented in terms of scale and have the CO2e data presented on the label.

Keywords Carbon labeling, GHG emission, consumers preferences, consumers attitude, design, layout, content

Table of Content

1. Introduction 1 1.1 Background 1 1.2 Problem 2 1.3 Purpose 3 1.4 Definitions 3

2. Frame of reference 4 2.1 Rising demand for sustainable products 4 2.2 Groceries effect on global warming 4 2.3 Problems with misconception of organic products 5 2.4 Effects of eco labels 6 2.5 Carbon literacy 7 2.6 What is carbon labeling 8 2.7 Implementation of carbon labeling 8 2.8 Advantages of carbon labeling 9 2.9 Challenges of carbon labeling 10 2.10 Carbon labeling in other cultures 13 2.11 Existing and experimental carbon label designs 13 Fig 2.1 14 Fig 2.2 14 Fig 2.3 15 Fig 2.4 15 Fig 2.5 15 Fig 2.6 16 Fig 2.7 16 2.11.1 Colour coded vs. Monochrome 16 2.11.2 With data vs. Without data 17 2.11.3 With scale vs. Without scale 17 2.11.4 Well known organization vs. Lesser known organization 18 2.11.5 Per weight vs. per calorie 18

3. Methodology 20 3.1 Research Approach 20 3.2 Research Design 20 3.3 Credibility of Literature Review 20 3.3.1 Reliability 20 3.3.2 Validity 21 3.4 Data Collection Method 21 3.5 Choice of theories 22 3.6 Research Strategy 22 3.7 Survey Method 22 3.8 Time Horizon 23 3.9 Operationalization 23 Fig 3.1 23 3.10 Designing the questionnaire 25 Fig 3.2 26 3.11 Sample strategy 26 3.11.1 Sample size 27 3.12 Data analysis 27 3.13 Generalization 28 3.14 Limitations 28

4. Empirical results 29 4.1 Questionnaire answers 29 Fig 4.1 29 Fig 4.2 29 Fig 4.3 30 Fig 4.4 30 Fig 4.5 Question 1 31 Fig 4.6 Question 2 32 Fig 4.7 Question 3 32 Fig 4.8 Question 4 33 Fig 4.9 Question 5 34 Fig 4.10 Question 6 35 Fig 4.11 Question 7 35 Fig 4.12 Question 8 36 Fig 4.13 Question 9 36 Fig 4.14 Question 10 37 Fig 4.15 Question 11 37 Fig 4.16 Question 12 38 4.2 Comparisons within sample 38 Fig 4.17 38 Fig 4.18 39 Fig 4.19 39 4.3 Correlation analysis 40 4.3.1 Gender 40 Fig 4.20 40 Fig 4.21 40 4.3.2 Age 41 Fig 4.22 41 4.3.3 Education 41 4.3.4 Perceived Knowledge 42 Fig 4.23 42 Fig 4.24 42

5. Analysis 43 5.1 Label design 43 5.1.1 Colour coded vs. Monochrome 43 5.1.2 With data vs. Without data 43 5.1.3 With scale vs. Without scale 46 5.1.4 Well known organization vs. Lesser known organization 46 5.1.5 Per weight vs. per calorie 46 5.2 Further analysis 47

6. Conclusion 49

7. Further Studies 50

References 50 Books 50 Articles 50 Webpages 51

1. Introduction

1.1 Background Global warming has been a topic of discussion at universities and political arenas since the discovery that man-made greenhouse gas emissions is having an affect on the planet 1972 (Nicholls, 2007). Since then it has been widely accepted that human’s greenhouse gas footprint in earth's atmosphere has a warming effect on the planet which if continued will have big impacts on rising sea levels, extreme weather, mass extinction of ecosystems and expansion of deserts.

There is a growing trend among the younger generations in Sweden to become more vocal about their opinions on how the government are approaching the climate crisis. The term “flygskam” which translate to “The shame of flying” has evolved as a result from the climate debate and may be the cause of the decrease in domestic flights in Sweden (Swedavia, 2019). The term was basically nonexistent before 2018 (Google Trends, 2019). There is also a growing movement of children “School striking” on Fridays for the environment and newspapers writing about the “Greta Thunberg” effect. The global interest of the Swedish climate activist Greta Thunberg reached a global peak in September 2019 as she spoke in front of the UN Climate action summit 2019 (Google Trends, 2019). In 2019 there is a clear break in the global google search trend for “Sustainability” and sustainability related topics (Google Trends, 2019).

United Nations initiated a grand campaign promoting global sustainability goals for 2030 which seeks to abolish poverty, decrease inequalities and injustices, to promote peaceful societies and to solve the climate crisis. They seek to make make businesses, governments and the general public take responsibility when it come to sustainability related issues and they are launching the “Global Goals” campaign all over the world (UN Global Goals, 2019). Goal number 12 focuses on Responsible consumption and production and one of its subgoals are to promote universal understanding of sustainable lifestyles. When it comes to food consumption there are already many companies that distinguish their products that are more environmentally friendly with labels and brands which would show nutrition, origin, additives and production methods (Leach et al., 2016). The demand for organic foods in Sweden has received a distinct increase but are facing challenges when it comes to supply of organic foods, trust in the food labeling system and they also lack understanding of what organic production actually means. (Bosona & Gebresenbet, 2018)

Climate change is caused by greenhouse gases and the way that greenhouse gas emissions is measured is through equivalents (CO2e). There are several greenhouse gases - some much more potent albeit much less common than carbon dioxide. The most important greenhouse gases are carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydroflourocarbons (HFCs), perfluorocarbons (PFCs), and sulfur hexaflouride (SF6). CO2e is all greenhouse gases expressed in CO2 in order to quantify and simplify the combined greenhouse gas emissions of a specific product or service. There are different ways of measuring CO2e, but in a carbon labeling scenario both direct and indirect emissions would have to be taken into consideration. For example, in milk production all emissions related to cattle management, such as heating, machinery, vehicles, animal fodder and even the belching of the cows matters and needs to be taken in consideration (Leach et al., 2016).

CO2 and the other greenhouse gases acts as a protective “blanket” from the heat which is radiating from earth to space in the form of infrared radiation. The greenhouse gases are absorbing the radiation and heating the gases up which is in turn is heating the atmosphere and also the surface of the earth.

Water vapour is the earth's largest contributor of greenhouse gas and is controlling the earth's temperature. If the air is warm, it can hold more water vapour thus causing a bigger global warming effect. However, when the water vapour meets cold air it turns into clouds and eventually turned in to liquid water and is later rained back to the surface of the earth. However, with the increasing heating effect of CO2 and other greenhouse gases in earth's atmosphere it allows for more water vapour to enter the atmosphere which increases the temperature even more (NASA, 2008).

For more than two decades there have been tests and researches done to implement labels that communicates the GHG-emissions. How the labels should be communicated and what the labels should contain differs among the different organisations that stands behind these test and researches.

1.2 Problem There are many problems and challenges when it comes to countering global warming. It is a problem that require global commitment since it affects the whole planet. One of the biggest challenges with countering global warming is to communicate and educate the world’s population about what kind of behaviour that is causing it. Carbon literacy is a term for the awareness of and the effects that our everyday actions has on climate change. It is a term that also includes knowledge about what causes climate change and also of what the resulting effects are (Sharp & Wheeler, 2013).

Grocery products stands for roughly a third of EUs carbon emissions as a result of its high- volume production (Sharp & Wheeler, 2013). In Finland Hartikainen et al., (2014) claim it is 25% of the total emissions (Hartikainen et al., 2014). Carbon labeling is a tool in which retailers and manufacturers can use to communicate the amount of CO2e one product has caused throughout its whole life cycle. Different theories with different methods have been tried with various results. They can see that it can affect the consumer purchases choices even though the carbon labeling concept still does not have had a breakthrough yet. The two key problems it all boils down to is that there are no standardisations, regulations or policies when it comes to measuring and using carbon labeling. The other is that greenhouse gas emission data is information that is hard to make understandable and communicate to consumers. This emphasizes the importance of having the right information, with the right content with the layout. There has been done several surveys in countries such as Australia, Finland, Denmark, Germany and UK related to carbon labels. Both in regard of what layout approach that is preferred by consumers as well as evaluating whether it changes consumer behaviour if carbon labels are presented with the product. The articles in our frame of reference which has researched carbon labeling come to basically the same results with some minor differences regarding carbon literacy, standardized measuring methods, label designs, implementation and effects. We have chosen to focus on the visual aspect of the carbon label. This has not yet been tried in Sweden, which makes it interesting to see what kind of content, design and layout Swedish consumers would prefer to have in a carbon labeling programme.

We want to find out the preferences of Swedish consumers what a carbon label could look like without being too complex or confusing or to simple and non-informative. We were hoping that this research could potentially help to contribute useful information for future decision makers and researchers and help them to overcome the obstacles on creating an effective carbon labeling system for Swedish consumers, retailers and grocery producers.

1.3 Purpose There has been attempts of implementing carbon labeling in grocery stores before but without success or a breakthrough in grocery retail store norms. The purpose of this dissertation is to assess and evaluate the preferred way to communicate information via carbon labeling among Swedish consumers. This includes content, design and layout.

By thoroughly researching through previous journals and studies that has been made we hope to funnel their conclusions into different approaches of carbon labeling with differentiated layouts, designs and information provided. We will then test these approaches with Swedish consumers to see which approach is preferred.

What is the preferred content, design and layout of a carbon label among Swedish consumers?

1.4 Definitions

LOHAS - Lifestyle of health and sustainability CO2e - Carbon dioxide equivalents

2. Frame of reference

2.1 Rising demand for sustainable products

There is a consumer group whose name is LOHAS which stands for Lifestyle of Health and Sustainability) which are very positive towards all things related to Sustainability. The group consists mainly of women aged 35 - 60, has an above average income and have a higher level of education. They are also willing to pay more for products if it has a positive sustainable positioning (mainly organic foods). This consumer group has had a big increase in size in the past few years in Sweden - from 27% in 2005 to 38% in 2015. LOHAS in Europe makes out about 20% of the population in Europe which would indicate that Sweden is early on this trend (Bosona & Gebresenbet, 2018).

The rising demand from the consumers are pushing corporations to be more transparent about the affect their is causing. The fact that consumers choices actually affect the environment is creating a need for environmental labeling that is accurate and that makes the process of “green purchasing” easy (Cohen & Vandenbergh, 2012).

Even governments, countries and regions are starting to see the benefits of sustainable products and organic farming. In 2018 as part of the United Nations Trade and Development and event with the East African Organic Policy Forum was held in Tanzania. Representatives from the government, market shareholders and people responsible for the agriculture industry and the tourism industry was present. The event was to identify the barriers with organic farming in eastern Africa and to show the linkage between agriculture and tourism (United Nations Conference on Trade and Development, 2017).

The climate crisis should be seen as a collective issue. Producers, retailers and consumers needs to work together towards a more sustainable lifestyle. It shows that consumers are willing to act more sustainable but have a hard time to see the pattern of how their everyday life is causing harm towards the environment. Consumers count on government and policy makers to help them navigate towards more sustainable choices as well as they count on the retailers to provide sustainable products (Feucht & Zander, 2017).

2.2 Groceries effect on global warming Grocery products stands for roughly a third of EUs carbon emissions as a result of its high- volume production. That also means that if consumer behaviour could be changed just a little, it could have a relatively big impact on overall emissions (Sharp & Wheeler, 2013).

Households in Europe are becoming more aware of the effects that grocery production has on global warming, but it has not resulted in the emersion of carbon emission related food labels. Carbon labeling on grocery products are rare and attempts by Tesco 2007 in the UK resulted in its discontinuation in 2012 as a result of non-cost efficient ways of measuring the carbon footprint on products and that the rest of the market “didn’t catch on” (Leach et al., 2016).

The Swedish fast food hamburger chain “Max hamburgers” has been transparent with their total emission of CO2e the past years and by looking through a breakdown of their total emissions the effect beef production industry has is a substantial part of the overall emissions.

(Wrenfelt et al., 2018)

2.3 Problems with misconception of organic products

Environmental food labeling has been around for decades as an effect of the consumers increasing concerns for the environment. The most prominent of the the green marketing labels are the ones related to organic production or farming. As it stands right now neither conventional or organic farming is sustainable when it comes to greenhouse gas emissions and it is a common misconception in Sweden that an organic producer also would have a lower greenhouse gas emissions rate (Bosona & Gebresenbet, 2018).

In an article where conventional and organic farming of wheat in Italy was studied, they concluded that organic farming of wheat resulted was causing higher emissions of greenhouse gases as opposed to conventional. This is primarily because of the lower yield per hectare that organic farming has in comparison to conventional farming (Chiriacò et al., 2017). However, there are many positive impacts on the environment from organic farming that conventional does not provide such as the fact that pesticides and artificial fertilizers are not used, organic products have proven to be more nutritious and contains less residues of synthetic and artificial pesticides and fertilizers. Organic standards are designed to improve animal welfare and allow the animals to have access to their natural needs (Treu et al., 2017). 2.4 Effects of eco labels In 2004, as a result of eco labeling, Coop Sweden declared that because of the consumers choice of ecological products reduced the pesticides by 14 000 kg and the amount of synthetic fertilisers by 1 000 000 kg in Sweden. Even the use of ecolabeling in household cleaning products, where the surfactants used in the product had been replaced by biodegradable ones and that reduced the use of chemicals by 15% (Peano et al., 2015).

One test that has been tried is the traffic light scheme. By using colours to make it easier for the consumers to identify the level of carbon literacy, where Red has the highest carbon emission, yellow the second highest and where green has the least. They used it to analyse purchases of fish. The result showed that the overall sales of fish dropped by 15%, mostly because there was a 35% drop in seafood sales who displayed a yellow sign. However, there was no change in sales of the red nor the green sign labels (Sharp & Wheeler, 2013).

In a study that was made in European countries in more recent years about the purchasing power different labels would have on the consumers it showed that France, Spain and Italy had a similar preference towards a carbon label and a organic label. The same study also shows that Germany, Norway and the United Kingdom actually preferred the carbon label over organic label (Feucht & Zander, 2017).

To understand consumers familiarity and claimed use of eco labels an Australian survey was made to 455 respondents. In the survey they were shown 11 commonly used eco labels and they were asked to recall if they recognised it and if the label had any influence on their purchase decision. The labels were referring to energy efficiency, organic, sustainably sourced materials and Fair Trade. To find out the carbon footprint relative to the other environmental criteria such as being energy efficient, being organic and being certified Fair Trade they were asked to rank the level of importance in nine different levels of importance in a scale of 0-10, this to see how prioritised the carbon footprint is compared to the other environmental criterias. 99% of the respondents said they had familiarity with at least one of the named labels in the survey where the Energy Rating Label (95%) and the Water Rating label (61%) had the highest claimed familiarity as the grocery labels had 32% familiarity. When it comes to influencing their purchases the Energy and Water ratings had the most claimed influence. The grocery labels on the other hand had 39% claimed influence on the respondents purchasing behavior. The results show that consumers have a higher awareness and are more influenced by Energy and Water ratings, but it also demonstrates that grocery labeling has a high impact in influence in the consumers that has familiarity with the labels. This strengthen the theory about the importance of awareness and indicates that there is a potential role for grocery labels when it comes to consumers purchasing choices (Sharp & Wheeler, 2013).

A German study that was made in 2017 shows that when it comes to purchasing power and eco labels, people are more willing to pay a higher price for a locally produced labeled product than a product with a carbon label. However, consumers were willing to pay a premium price of 20% for a carbon label product and that just the presence of a carbon label would increase purchase probability (Feucht & Zander, 2017).

Van Loo et al., (2015) made a study to compare consumer’s preferences of four different sustainability labels including organic meat, free range, animal welfare and carbon footprint in the Belgian market when it comes to chicken meat. Overall, the largest segment of consumers were positive towards all the sustainability labels but the labels with free range claims had the highest positive impact where 90% of all the consumers had a positive attitude towards the label and 97% of the consumers liked the traditional, well recognized free range label. The study also indicates that the consumers are willing to pay twice as much for the free range products than the products who is having a EU animal welfare label. In the bottom of the list we find Carbon footprint labels along with Organic food. They were disliked by more than 20% of the consumers. According to the study, the reason the low test score of the Carbon Footprint label is that the label is a new concept and that this kind of label does not yet exist in the Belgian food market. The other factor is that consumers are uninformed about the content of the Carbon footprint label. Many consumers are confused about the meaning of the labels. They study shows that 78% of the respondents has a willingness to pay for sustainable premium products so they predict that since sustainable products, and carbon dioxide in particular, is a hot topic they think this could change in the future and that there is a market for carbon footprint labels (Van Loo et al., 2015).

An indirect effect a carbon label could have is that a successful carbon labeling system will affect the retailers. They might demand from their suppliers that their products are carbon labeled or that the suppliers have done a LCA analysis on their products. (Feucht & Zander, 2017) One example of the effect a eco label have had is a German retailers committed to phase out wild fish products that did not have a sustainability label. This effect happened even though the knowledge about the label was limited. This means that if retailers would collaborate and work together, they can exclude high emission products without including the consumers (Zander et al., 2015).

2.5 Carbon literacy Carbon literacy is a term for the awareness of climate change and the effects that our everyday actions has on climate change. It is a term that also includes what causes climate change and also of what the resulting effects are. In many cases consumers are not aware of the effects that their consumption is causing and therefore can not as easily make a decision that reduces the effects of climate change (Bosona & Gebresenbet, 2018).

Carbon literacy was measured in a survey of Australian households where the respondents were asked to appreciate the percentage of total Australian annual CO2e emissions that grocery production was causing and 55% of the respondents didn’t know and out of the 45 % that did estimate only 30% of them were within 10 percentage points of the actual value of 33% (Sharp & Wheeler, 2013).

In the next question the respondents were asked to rank individual products within a grocery product category from the highest emitter of CO2e to the lowest. Their findings were that 1 out of four consumers were unsure of the highest emitter and in general the respondents held only modest knowledge of the highest emitter in every category. And they concluded that two out of three would not be able to make informed carbon emission related choices in a supermarket (Sharp & Wheeler, 2013).

Next they were asked to estimate the amount of CO2e emissions per tonne of several specific grocery products and even though they were guided in terms of a range of emissions and given a reference value only 16% of respondents were able to estimate a reasonable level of carbon emission of the selected products (Sharp & Wheeler, 2013).

It was concluded that Australian households only had basic carbon literacy but lack the advanced knowledge to make carbon conscious choices when in a supermarket or provided the amount of CO2e emissions the specific product has. They conclude that a carbon labeling system may give means to improve carbon literacy in Australian supermarkets as well as support consumers in making carbon conscious choices (Sharp & Wheeler, 2013).

When Finnish consumers were asked which part of their consumption that was responsible for the most carbon dioxide only 9% estimated that food consumption was causing the largest carbon footprint. Additionally, they were asked whether they knew what the term “product carbon footprint” meant and only 7% of the respondents linked the term to greenhouse gas emissions and only 5% associated the term with climate change (Hartikainen et al., 2014).

2.6 What is carbon labeling Carbon labeling describes and informs consumers the carbon dioxide emissions created throughout the whole supply chain when it comes to creating a product. The major variety of carbon labeling can be explained by that there is no regulations or standardizations when it comes to measuring CO2e. Governments, manufacturers and retailers can act individually and that can make the labels misleading and hard for consumers to understand. A manufacturer can either have its label verified by a third party, which in most cases is the most trustworthy but at the same you can use a “self declared” label (Sharp & Wheeler, 2013).

The main purpose of a carbon labeling system is to inform consumers at the point of purchase of the total carbon footprint of a specific products total life cycle. By carbon labeling products it could help consumers to improve their overall carbon literacy and also help even the most “carbon-illiterate” make a climate conscious decision when shopping groceries while it also helps the the “carbon-literate” to reduce their search costs (Sharp & Wheeler, 2013; Hartikainen et al., 2013).

The secondary purpose of a carbon labeling system is to incentivise or put pressure on producers to lower their overall carbon emissions in order to gain a comparative advantage over their competitors (Gadema & Oglethorpe, 2011).

Max Hamburgers in Sweden made an initiative that was well received when they printed their carbon emission on their menu. This led to an increased sales in vegetarian hamburgers and shows that labeling can affect consumers purchasing choices without causing a loss in sales (Leach et al., 2016).

2.7 Implementation of carbon labeling A study of Australian householders show that the pre-existing knowledge of carbon emissions related to grocery products life cycle is really low and the natural first step to try and change consumer behaviour is to improve householders carbon literacy. The natural response would be to label the carbon emissions along with the prices in grocery stores. By offering carbon labels in a transparent and objective way in shops it would make it easier even for the most non-carbon literate to make more sustainable shopping decisions. They suggest that carbon labeling could help in the to help improve the CO2e awareness. That it would be a good method to reach out to the consumers (Sharp & Wheeler, 2013).

The biggest difference in CO2e is when a consumer changes their purchasing behaviour from buying a high CO2e product to a low CO2e product. For example, going from meat to vegetarian. But for the labeling system to work the study shows that consumers prefer to see carbon emission compared to other similar products. The study also says that consumers will not spend time on engaging and evaluating different labels. They imply that the labels will have to be easy to understand and help the consumer make a fast decision for the labeling system to work. Other methods that has been suggested and tried is the “Stop light” labeling approach. This approach is used to show its carbon impact in use of colors, where green is good, yellow is alarming and red is bad. The outcome of this experiment showed a small change in the purchasing behaviour. Only when the greenest product also was the cheapest, they could see a major change (Sharp & Wheeler, 2013).

Another study also implies that the carbon labeling system will only work when the consumers are willing to act and substitute the high carbon emission products to a product with low carbon emission. Eco labeling still puts a lot of pressure on the consumers and that need to take action for the system to work, they suggest that the labeling system should be implemented carefully and still have the consumers demand in mind. Just like Sharp and Wheeler’s theory that consumers prefer to choose between similar products, they mean that letting consumers compare/choose between salmon and sea bass, instead of making them choose between beef and vegetarian would have a bigger impact in the reduction of the total emissions (Shewmake et al., 2015).

When Tesco pledged in 2007 to label all their products with a carbon footprint label, they labeled over 1100 products but only showed the exact amount of CO2e emissions in relation to the product weight without using a tier-system. This effort was discontinued in 2012 because of the amount of work that had to go into it at that time and blaming other retailers for not joining in to try to standardize and normalize the labeling system. The fact that a large and committed grocery retailer such as Tesco could not successfully implement a carbon labeling scheme shows that it is probably required that a mandatory government regulated carbon labeling scheme is enforced and controlled in order to successfully implement carbon labeling in supermarkets. By having official governmental policies it could help to minimize consumer confusion by standardizing a carbon label on a national level. Furthermore, a government policy could also help in improving the general mistrust in carbon labels by establishing governmentally approved measuring systems that third party “CO2e measuring” companies would have to adhere to (Thøgersen & Nielsen, 2015).

2.8 Advantages of carbon labeling The main reason for carbon labeling would be to change consumer behaviour and to improve carbon literacy among consumers of grocery products. A carbon labeling system based on a traffic light system where a red light would indicate high carbon emission, yellow would indicate a neutral level of carbon emissions and a green light would indicate a low amount of carbon emission will be a guide for consumers to make quick and “carbon friendly” decisions in a supermarket without having to invest time to learn and quantify how much of an impact that 1 gram CO2e/kg has on the environment. Consumers would have to invest considerable amount of time to improve their carbon literacy to know whether one kilogram of CO2e is a lot or little, since it highly depends on the product in question. However, there are estimates that a successful carbon labeling scheme could lower the annual total consumption of CO2e by 1 tonne per individual (Sharp & Wheeler, 2013).

A study was made of Danish consumers who got to pick between different ground coffee products given different kinds of environmental labeling. The labels helped the consumers to identify high emission products and instead helped them to identify the substitute products with substantially lower emissions. Their conclusion was that carbon labeling done in a traffic light manner proved a significant impact of consumer purchasing patterns in favour in regard to lower carbon emission products. It also helped to improve carbon literacy (Thøgersen & Nielsen, 2015).

In the Finnish study there was a clear majority of 86% of respondents that would like to have a carbon labeling system that could be used to compare products between themselves. However there were different opinions in regards to how much information that should be provided at the point of purchase. A concluding point from the Finnish survey was that 90% of all respondents believed that they would most likely change their purchasing behaviour a little if a carbon labeling system was implemented (Hartikainen et al., 2014).

2.9 Challenges of carbon labeling Carbon labeling should not however be considered a universal cure to change consumer behaviour as there is a high risk of confusion between several different green marketing labels as well as skepticism toward the brand and label itself (Gadema & Oglethorpe, 2011).

A survey was done of UK consumers across a selection of respondents reflecting the UK population regarding eco labels and purchasing behaviour. When the respondents were asked of what they deemed to be the most important product attributes the top three responses was quality/taste, nutrition and price. Carbon labeling resulted in being the second least most important attribute over attractive branding (Gadema & Oglethorpe, 2011).

However, 59% of the respondents recognized that the connection between grocery production and climate change as being very important (4). As well as 91% of the respondents claimed that carbon labels on food would help them in making purchasing decisions. They also agree that carbon labels were confusing and difficult to put in perspective as to what they actually are contributing to (Gadema & Oglethorpe, 2011).

In conclusion Gadema and Oglethorpe (2011) is arguing that there is not high enough trust and understanding of how information in carbon labeling is conveyed. Consumers will only go for a low carbon product if there is a direct substitute product or if the prices of the relatively lower CO2e emission product were the same or less. They also conclude that change in consumer behaviour relying only on consumer guilt and voluntary on-product branding would have little to no effect and in turn would not incentivise or put pressure other companies to lower their own CO2e emissions in order to gain a competitive advantage over their competitors.

Instead of taking a “soft” approach to carbon labeling and implement mandatory carbon footprint reporting by producers it could level the playing field of all companies that could instead incentivise companies to reduce carbon emissions. If the information then is conveyed in the same way everywhere perhaps it could lead to a reduction in consumer confusion (Gadema & Oglethorpe, 2011).

Research that has been made show that the barriers that keep consumers from purchasing carbon labeled products was availability, price, habits and marketing. Other factors that prevent carbon labeling were distrust in the labeling system and that consumers have a lack of perceived personal benefit of buying a low emission product. To reach wanted outcome of the labeling, the message needs to be presented so that it is easily understandable and shows context and values (Leach et al., 2016).

As of November 2019, there are 463 different kinds of ecolabels (Ecolabel index, 2019) in 199 countries and 25 industries. The idea of the eco/sustainability labels is to increase the consumers knowledge and make it easier for them to make sustainable choices and to encourage transparency along the food supply chain. However, though the intention of the labeling is positive, the lack of standardization and regulation of the labeling industry, misleading and even deceptive green marketing by manufacturers has been made to undermine the “sustainable consumers”. This has led to dissatisfaction among the “sustainable consumers” and they have been forced to find information elsewhere, such as websites, newspapers, television programs, education and other advertising (Peano et al., 2015).

Just like in the UK report, a research made in Australia show that consumer understand the basics of carbon labeling, but the information about carbon emission is hard to interpret. Consumers understand that when the product indicates a higher carbon number it is worse for the environment than a product with a lower number. They mean that simple emission reports in labeling and communication is to no good since that means that the consumer need pre-existing knowledge and that it lacks relativity. It is better to simplify and make the interpretation of the label easy for the consumer to understand (Sharp & Wheeler, 2013).

Labeling just one category could lead to backlash according to Shewmake et al., (2015). Meaning that if we would just label one product, pork for instance, people would tend to go from pork to over consume something that could be worse when it comes to CO2e - beef for example. This has to do with lack of knowledge when it comes to what products has the biggest effect on the global warming. They agree that that the most efficient way to reduce carbon emissions is to have carbon tax, cap and trade systems and carbon labeling (Shewmake et al., 2015).

Even though retailers could have an impact by implementing carbon labeled products, Emberger-Klein & Menrad (2017) imply that retailers are afraid of lowering the consumers demand for meat because it is a very profitable market and they do not want to encourage people to consume less food. One other factor is that they may not have tools or knowledge to communicate the message effectively (Emberger-Klein & Menrad, 2017).

One key factor is to find the right social ethos. To get the carbon labeling to take off it need to be communicated to the right target group. Studies shows that the differences in the willingness to respond and purchase low carbon products has a correlation to age, educational level and income level. Just like older people have a hard time keeping up with new technology younger people don’t tend to write letters anymore. The carbon labeling system needs to target and communicate the right group of individuals. The study shows that the people that best respond to the labels are young, well-educated and people with a higher level of income. Since there is no government regulations or policies this can be a great challenge for all the different kind of labels since each individual label has to make its own strategy to reach the right consumers (Li et al., 2017).

Yvonne Feucht and Katrin Zander (2017) made a carbon labeling test in different European countries were the results were surprisingly good. Several countries had similar preferences towards a carbon label and an organic one. Some countries also preferred the Carbon label even higher than the organic label. This study however was a bias study towards highly educated people since they have a better understanding about sustainability labels. This means they leave out the lower educated people in the results and do not have an understanding of their preferences (Feucht & Zander, 2017).

2.10 Carbon labeling in other cultures

The consumer behaviour towards carbon labeled products can differ depending on the different cultures. A research that has been made in China shows that financially strong, young to middle aged men with high education have strong low carbon purchasing power. It also shows that the purchasing power differs more when it comes to different regions than within age and gender groups. Consumers at different educational level show the greatest difference in purchasing power when it comes to carbon labeled products. This is parallel with the different educational level in the different regions in China (Chuanmin et al., 2014).

One other factor that plays a vital role in China when it comes to attitude towards carbon labeling is the government's policy. It indicates that establishing a credible authority that works on publicity when it comes to carbon labeling can have a strong positive effect on the country’s inhabitants (Chuanmin et al., 2014).

In Vietnam a study was made about the relationship between labeling and attitude towards meat. The study showed that the even a good attitude towards the label would translate into a purchase. This is due to a restrained financially ability among the population and the perceived high price on labeled products (Nguyen et al., 2019).

2.11 Existing and experimental carbon label designs

France became the first country with legal requirement for carbon labels when they in 2010 introduced the “Grenelle 2” law. In the beginning it was a voluntary environmental labeling scheme for all consumer gods that was sold in France. In 2012 it became compulsory for certain categories of products. (Liu et al., 2016)

Carbon Trust, the UK based non-government organization was pioneers when it comes to carbon labeling. In 2007 they launched the world's first carbon label and has since made more variants. In 2020 they now have four different labels in which producers can license.

- CO2 Measured: that a product’s carbon footprint has been measured and certified - Reducing CO2: that there is a commitment to reduce a product’s carbon footprint, or there has been a reduction in a product’s carbon footprint, plus a commitment to achieve ongoing footprint reductions. - Lower Carbon: that the certified lifecycle carbon footprint of a product, or group of products, is/are significantly lower than the market dominant product. - Carbon Neutral: that a product’s carbon footprint has been reduced and any outstanding emissions are offset (Carbontrust.com, 2020).

Fig 2.1 shows the slogan “Reducing with the Carbon Trust” meaning that the manufacturer’s commitment to reduce the greenhouse gas emission throughout its whole value chain.

Fig 2.1

The second label, Fig 2.2, shows a “CO2 measured” label and has the slogan “Working with Carbon Trust” and has data shown on the footprint. This label has the numeric value of greenhouse gas emission on it. The data is based on the GHG protocol standard (Carbontrust.com, 2020).

Fig 2.2

Fig 2.3 shows the Carbon trust label which guarantees that the company is offsetting all the emissions related to the products complete life cycle and has no impact on climate change (Carbontrust.com, 2020).

Fig 2.3

The labels have different criterions which need to be met in order for a company to be able the license them. Some have more complex requirements than others. (Carbontrust.com, 2020).

Fig 2.4

Leclerc, which is a French retailer, made their own private carbon labelling standard in France 2008 called “Bilan Carbone” where they calculate the carbon footprint of the product it sells. (Feucht & Zander, 2017)

Fig 2.5

“Fig 2.5” is based on a French retailers carbon index called Casino ('l'indice carbone'). This is an experimental label used in a study by Feucht and Zander, 2017.

Fig 2.6

“Fig 2.6” Is also an experimental label, based on the European energy label which is widely recognized and also used in the study by Feucht & Zander, 2017

Fig 2.7

“Fig 2.7” is a model taken from the Finnish company Raisio which were used on their products to indicate the amount of CO2e per 100 grams of product. (Raiso.com, 2020)

2.11.1 Colour coded vs. Monochrome In almost all previous studies regarding carbon labels that has been done the idea of a traffic- light based approach has been tested in comparison to a standard monochrome or single coloured design. In those specific cases the result of a coloured label design has almost always been the most popular layout both in terms of how to simplify complex data but also when it comes to compare products between each other. The traffic light label design is a well-known way to label a product from good to bad and has been used in other labels previously and is a natural tool to use in a carbon label. The EU energy label is a good example of this layout. The EU energy label has got a 7 step scale which could be applicable in a carbon label for groceries as well but since groceries are considered to be purchased with a lot less involvement than e.g. a washing machine a 3 step scale would be more relevant for groceries and easier for consumers to grasp. (Thøgersen & Nielsen, 2015).

Based on this theory following hypotheses was formulated:

Hypothesis 1 (H1): Swedish consumers prefer colour coded labels.

2.11.2 With data vs. Without data In the previous studies there has been different experimental and non-experimental carbon label designs that has been used in their tests. Some of them has included the CO2e emission data and some labels have removed the data from the label itself and instead include a label tool that conveys the information in a simpler way such as a traffic light colour coded label. Examples of the latter would be labels such as the “Reducing with Carbon Trust” (Fig 2.1) or the Finnish raisio (Fig 2.7). To many consumers the CO2e emission expressed as plain data does not give any useful information to make a climate conscious choice since carbon literacy is generally quite low.

However, one of the biggest issues with the use of a label with a scale of colours without data shown and that covers all grocery products is the choice between two products within the same category. If the choice is between beef from a local producer and beef from a national mass producer, then it is likely that the local producer would have lower emissions than the mass producer however their labels would both be red. The labels would still be identical from each other and would likely not give any indication as to which of the choices that is the lesser of two evils. This is an argument for adding data to the label.

Although one of the issues in regard to adding data to a label is that it would bring a whole new dimension of complexity to the label and the data itself is very hard for consumers to grasp and put in context. (Sharp & Wheeler, 2013; Bosona & Gebresenbet, 2018; Hartikainen et al., 2014)

Based on this the following hypothesis was formulated:

Hypothesis 2 (H2): Swedish consumers prefer data on the label.

2.11.3 With scale vs. Without scale

Some of the existing carbon labels as well as the experimental ones found in previous studies has got a scale to indicate how much emissions the specific products have in relation to all other products that are being measured. This creates a more complex label which would require a larger area and more space on the shelves or the product itself. However, the information itself could be easier to interpret.

A model which doesn’t include a scale could be substantially smaller and would be easier to fit on shelf price tags without the information being to small to be able to interpret it. It also decreases the amount of complexity of the label which could minimize confusion and the sense of being overwhelmed with information. (Thøgersen & Nielsen, 2015; Hartikainen et al., 2014)

Based on this the following hypothesis was formulated:

Hypothesis 3 (H3): Swedish consumers prefer the label being presented using a scale.

2.11.4 Well known organization vs. Lesser known organization One of the biggest issues with carbon labeling would be the distrust to the data and the data collection methods used. There are organizations which are more known and also have more credibility when it comes to logotype-based labels such as KRAV in Sweden for organic products. When creating a carbon label, a new organization would likely be required unless KRAV or a similar organization would be expanded to include carbon footprinting in their services. KRAV could then be the controlling agent that guarantees that the data on the labels are correct.

It takes time to build credibility and it is likely that a new company or unknown company would have little effect on consumption unless that label was somehow acknowledged by the government or government controlled entirely. Carbon trust is an example of an organization that controls and guarantees that the producers are following certain criteria that the organization has decided for their different levels of logotypes. This organization is however not widely known or used in Sweden and it has existed since 2001. (Thøgersen & Nielsen, 2015; Carbontrust.com, 2020)

Based on this the following hypothesis was formulated:

Hypothesis 4 (H4): Swedish consumers prefer that the label is being produced by a well-known organization

2.11.5 Per weight vs. per calorie

One of the main issues with carbon labeling is the spread of opinions among consumers of whether the scale should be within a specific product type or simply a scale for all types of food products. There is a division among consumers whether they want to compare within a product type such as different types of “ground coffee” or whether they want to compare ground coffee with substitutes such as tea. According to Hartikainen et. al, 86% of Finnish consumers wish to be able to compare products in a wider sense than just within a specific food category (Hartikainen et al., 2014).

There are different approaches to take when it comes to communicating carbon data relative to the quantity and volume of the product in order to get a relevant and comparable value. One approach is to take the total emissions of CO2e and express it in a standardized weight such as 100 grams or 1 kilogram. This approach is the most common among previous studies that has been made. The second approach would be to communicate the amount of CO2e emissions in relation to the amount of calories that the product has. The argument for this is that the emissions would be comparable with how much energy the specific product is providing as a kind of measurement of hunger relief. (Heller et al., 2015)

Based on this the following hypothesis was formulated:

Hypothesis 5 (H5): Swedish consumers prefer a label where data is presented in relation to the products weight.

3. Methodology

3.1 Research Approach

There are three different kinds of research approaches. Inductive, deductive and abductive. Inductive approach is based on empirical data that is later linked with a qualitative research. Deductive approach is when the research is based on existing theory and hypotheses are created and tested through empirical findings. Abduction approach is a combination of both induction and deduction. (Saunders et al., 2016)

We have decided to go with a deductive approach since we are doing a quantitative analysis. By going through previous studies we made five hypotheses that we later tested.

3.2 Research Design

The set of methods and procedures used in collecting and analyzing variables to answer the research question is called Research Design. There are 3 different types of research designs: exploratory, descriptive and explanatory. Exploratory research design is when there are no or limited previous that has been made prior. This is useful when you aim to deeply understand a problem and to find new insights. A descriptive research design is when your research is to describe a profile of a person, event or situation. The design enables you to describe or and identify variables in different phenomena. The last one is an explanatory research. This design is to explain a relationship between different variables in a situation or problem. (Saunders et al., 2016)

We are using an explanatory design since we are researching the relationship between preferences and different variables such as design, content and layout when it comes to a carbon labeling scheme and the Swedish consumers.

3.3 Credibility of Literature Review In order to maximize credibility in a study consideration regarding the data collection method need to be taken into account in order to ensure that the data collected is correct and relevant. By using the rules of reliability and validity the data collection method can be refined in order to minimize errors (Saunders et al., 2016).

3.3.1 Reliability Reliability is the method of making sure that a research is measured consistently so that the same research could be made with different participants and still achieve the same results. The goal is therefore to minimize bias in the test and to make sure that the participants reflect the population the survey aims to research (Saunders et al., 2016).

To prevent any biases, we used a snowball sampling approach via social media in which any participants who filled in the form and shared our post on Facebook would be eligible to win a gift card at a Spa-resort called Ästad Vingård. In this way we spread it on different Facebook groups and several different social circles in Sweden.

3.3.2 Validity Validity is the method that aims to make sure that the findings of our survey actually measures what it intends to measure (Saunders et al., 2016).

When designing the questionnaire, we decided to do it completely in Swedish since the topic of the survey contains data and information that we believe could easily be misinterpreted when conducting a survey among only Swedish participants. We also believe that by conducting the survey in Swedish we would have more people actually completing the survey since it would be less tedious to translate it and then complete it.

Another measure we went through before we sent out our questionnaire was to test it among a few chosen “pilots” to make sure they understood what the questions meant and what the survey intends to measure. As a result of these tests we made a few changes in the layout but also in the instructions in order to further increase validity and to ensure we got meaningful data after we’ve gone live with the survey.

In order for us to further increase the validity of our survey, our first questions we used illustrative examples of different kinds of possible logotypes in order to more easily show what we aim to research. We were also hoping that these examples would minimize the tediousness that respondents might experience when a form is too long or complex and with only text-based questions.

3.4 Data Collection Method There are two different kinds of sources used to collect data; primary data, which is data that is collected from the researcher. The Secondary data is when you reanalyze data that has been collected for another purpose. (Saunders et al. 2016)

Our primary data collected by a questionnaire we designed that was answered by random people. The questionnaire was spread through our social media channels such as Facebook, Instagram and LinkedIn. To get a wider spread of our questionnaire and to get more participants we had a gift certificate on Ästad Vingård as a price that will be randomly selected from all the participants. This ensures us that the questionnaire will get a wider spread than just our friends and family.

The secondary data we used was twenty two articles and twelve websites to get an understanding of the subject. The articles were peer reviewed articles gathered from Google Scholar and Halmstad University Library database.

3.5 Choice of theories When using a deductive approach, you start your research project by collecting theories about the subject of the researching. This so the researcher can form an idea and make hypothesis. (Saunders et al., 2016)

To get an introduction and understanding to why the idea and demand for carbon labels arose we had to have a deeper understanding into carbon literacy and what challenges previous and eventual carbon labels could face in the future. We researched the effect groceries have on the environment in general and specifically in terms of CO2e. We also looked into how the demand and challenges for carbon labels varies in different cultures.

To explore the effect of a carbon label and to be able to compare it, we gathered information about the effects other eco labels have had, in terms of success, obstacles and misconceptions. This to get an understanding on how consumers react towards the labeling systems.

Finally, we looked into existing carbon labels. How they were presented, what they contained and how they were designed. By reading preference tests that has been made in different countries around the world and the different approaches that was used when it comes to design, it helped us to get a deeper understanding and to form a questionnaire that could answer our research question.

3.6 Research Strategy Saunders et al., (2016) describes that depending on objectives, research question and existing knowledge in the area and resources there are eight different strategies that can be used: Experiment, survey, case study, action research, grounded theory, ethnography and archival research (Saunders et al., 2016).

Our research is to investigate the Swedish consumers preference towards carbon label designs our research strategy will be using experimental strategy. This is because we want to see the cause and effect relationships in the respondent’s choices when we present different designs. Since we have a deductive approach, we decided to use a survey, which is most common. By using a survey strategy, it will help us to get a “what do you prefer” answer from a large number of respondents. By collecting a large amount of data, it will be easier for us to see a relationship between the answers, preferred designed features and later find out if the research is representative for the whole population.

3.7 Survey Method Since we are looking for answers in a population's attitude towards a design in an explanatory research, we decided to use a questionnaire. We believe it would be hard to base our populations preference towards a label through interviews with fewer participants. We decided to use a web-based questionnaire since that gives the respondent’s flexibility to answer when the respondent has time, they would be exposed to less pressure and it would cause less of a distraction (Bryman & Bell, 2019).

By using Google Forms questionnaire tool, it gave us the opportunity to adapt the questionnaire so the respondents can answer it one their phones as well as to make it more aesthetically pleasing and user friendly. This was done in order to try and get as many as possible to answer the questionnaire.

3.8 Time Horizon

There are two different kinds of time horizons that need to be considered when planning a research. Saunders et al. explains this if the research should be a “snapshot” of a particular time frame or if the research should be more like a diary that represents events over a given time. The “Snapshot” is called cross-sectional time horizon while the “diary” perspective is called longitudinal (Saunders et al., 2016).

The most suitable option for us is to do a cross-sectional study. Even though Carbon Labels has been around for a longer period of time it is still a relatively new phenomenon. The perception and priorities of different climate threat-factor varies over time which is why this can be seen as a snapshot of what the preferences of a carbon label can be seen that we are researching this question within a certain time frame. Cross-sectional study also works better with our research strategy which is a survey.

3.9 Operationalization In our frame of reference, we identified the arguments, challenges and reasoning for different designs, layouts and content of carbon labeling. We also identified different layouts and designs of carbon labels that has been used in practice as well as experimental carbon labels which has been used in previous studies. By combining what would be a plausible and realistic approach to carbon labeling with limitations accounted for, we created our own experimental labels with different relevant attributes and characteristics on them on them. (Fig 3.1)

Fig 3.1 This is our own approach to an experimental Carbon label that is a suggestion from us that summarize some of the conclusions from previous studies that has been made. However, it is not well known and lacks an organizational backup.

By combining our own contribution with previous theories, we created items from our five hypotheses with added general items and attribute variables:

Hypothesis 1 (H1): Swedish consumers prefer colour coded labels. - Illustrative example of two labels where one is monochrome and one is traffic light colour coded - How highly would you value that a carbon label would be colour coded in a traffic light system?

Hypothesis 2 (H2): Swedish consumers prefer data on the label. - Illustrative example of two labels where one contains emission data and the other don’t - How highly would you value that the emission data was presented on the carbon label?

Hypothesis 3 (H3): Swedish consumers prefer the label being presented using a scale. - Illustrative example of two labels where one is presented on a horizontal scale and the other is not - How highly would you value that emission data would be presented on a scale on the carbon label?

Hypothesis 4 (H4): Swedish consumers prefer that the label is being produced by a well known organization - Illustrative example of two logotype based labels where one is a real carbon label but unknown to Swedish consumers and one which is an experimental extension of a organic label. - How highly would you value that the organization behind the carbon label is from a previously known organization?

Hypothesis 5 (H5): Swedish consumers prefer a label where data is presented in relation to the products weight. - Illustrative example of two labels where both contain the same date but in one case the data is presented in relation to weight of product and the other is presented in relation to amount of calories.

General questions: - How much would you value having access to the measuring process of a carbon label? - How much of an influence would a carbon label have on your grocery purchases? - How much influence does KRAV have on your grocery purchases? - How much influence does MSC have on your grocery purchases? - How much influence does Fair trade have on your grocery purchases? - How much influence does Nyckelhålet have on your grocery purchases? - How much influence does the EU-Organic label have on your grocery purchases?

The questionnaire also had four attribute variables to assess the respondents characteristics. The attributes were:

- Age - Gender - Education - Perceived knowledge about greenhouse gas emissions from grocery products

These attributes were used in order for us to see whether the sample is representative of our population and also to see if there are any correlation or differences between the different attribute variables (Saunders et al, 2016). The general questions were added to be able to diversify and see patterns in our sample. They could also give an indication as to what content that is preferred on the labels.

3.10 Designing the questionnaire By creating an anonymous survey, it contributes to the minimization of bias in the survey responses (Saunders et al., 2016). The questionnaire was self-administered via Google Forms as mentioned above and the respondents were anonymous unless they wanted to participate in the lottery for a gift card at Ästad Vingård by leaving their email addresses. Names were not required to be filled in the form itself, neither was their email addresses. We believe that the survey was “anonymous enough” to not cause any bias in the survey responses.

We chose a closed question approach to our survey since closed questions do not require any writing and it makes the data easier to compare as the answers are mutually exclusive and prewritten (Saunders et al., 2016).

We decided to divide the form in two parts where in the first part we deliberately forced the respondents to make a choice of which one they prefer between two different kinds of labels. The layouts were presented next to an apple, a carton of milk and beef in order to show the differences the label would have within its own range (See Fig 3.2 below). The same grocery products were used in every example to fix the variable so that the products would not affect the respondents choice and create potential bias.

Fig 3.2

In the second part of the questionnaire we asked questions with a 5-point Likert scale where the respondents got to choose how highly they would value certain given attributes on a carbon label. In this part we had the possibility to clarify some of our questions that might not have been clear in the illustrative part of the questionnaire. But most importantly we got to see how much they value certain attributes of the label compared to others. Since a label most likely would not be able to cater all attributes without being oversized we could derive which attributes that the respondents feel are the most valued and arguably the one that should be used. We also tried to streamline the questionnaire from a user interface point of view in order to lower the average time as it could lower the response rate. (Saunders et al., 2016)

The data used for the illustrated questions were selected from the open list of the climate database for grocery products by Research Institutes of Sweden version 1.6, 2019 (Rise, 2020).

3.11 Sample strategy

There are two techniques you can use when sampling data from a population. A non-probability sample is where the respondents are chosen from the population by the researcher. This is more commonly used in a qualitative research. The second technique, the Probability sample, everyone in the population has an equal change on participating in the research. Since we are doing a quantitative research this is the technique we used. Within the probability sample there is another five techniques that can be used in you sample strategy: Stratified random sample, which means you divide the population into smaller groups based on a number of attributes. Systematic sampling is when you select your samples within an interval of your population. Cluster, which is similar to Stratified sampling is when you divide the population into smaller groups, an geographical area for instance. Multi-Stage sample is a further development of Cluster sample. It is a technique used when you use multiple of cluster samples. The fifth sample strategy and the one we are going to use is Simple random sampling. That is when you select your samples randomly using a computer. (Saunders et al., 2016)

We chose Simple random sampling because our study is geographically concentrated to Sweden and it contains periodic patterns. It enables that our result can be representative towards the whole population because it allows us to select our sample without bias.

The quintessential population for our study would be every person in Sweden that has anything to do with grocery shopping. Our strategy to make the questionnaire in Swedish even though there are Swedish citizens that does not speak the language was because we wanted to limit some of the eventual cultural differences that could occur and we believed that we would get a wider spread and more respondents if the questionnaire was in Swedish. To even get a more accurate sized population to this study we excluded people under the age of 20 and over the age of 85.

We made the questionnaire available on LinkedIn and Facebook via our personal profiles. To increase the chance that people not only would participate but also share the questionnaire on their personal profiles we offered the people who participated and shared the questionnaire the chance of winning a gift certificate at Ästad Vingård. We also posted the questionnaire in two different Facebook groups. Since sharing on Facebook can be age and geographically limited, we also shared the questionnaire via email to certain workplaces to get a more accurate representation to the population.

3.11.1 Sample size

According to Statistiska Centralbyrån (scb.se, 2020) the population in Sweden over 20 years of age is approximately 7 900 000 people in 2019. Within the population there are approximately 250 000 people that are over the age of 85. Considering that there are group of individuals that can not do their own grocery shopping by own means, either because of age or other circumstances that makes the process difficult we decided to exclude 250 000 from the population.

We used Calculator.net, Sample Size Calculator to get an understanding of the number of respondents that we needed for our research. Generally, researcher use a 95% confidence level (Saunders et al., 2009). With a population of 7 650 000 people and a 95% confidence level we will need to have a sample size of 385 respondents (calculator.net, 2020). We managed to get 201 respondents which means that our margin of error went from 5% to 6.91%. This means, in this case, there is a 95% chance that the real value is within ±6.91% of the measured/surveyed value (calculator.net, 2020).

3.12 Data analysis We exported the data from google forms into Microsoft Excel as well as SPSS in order for us to be able to create relevant graphs and tables. We could then start analysing the data and dividing the sample into different subcategories in order to see and notice patterns. We could code all the answers in our form in a scale format by changing gender variables to numbers such as “female=1” and “male=2” and in our Likert scale based questions the selected value in a scale from 1 to 5 was simply transferred and did not need to recode any of those variables.

Excel was used in order to create clear graphs and SPSS was used in order to analyse the for correlation relationships within our data.

3.13 Generalization In order to be able to generalize our results as a social human behaviour it is important that the samples are of adequate size. This is so that the result can be applied to people who were not participating in the survey. (Saunders et al., 2016)

To be able to generalize our results we researched the amount of Swedish citizens that are over 20 years of age and the used a calculation tool to estimate the number of respondents that was needed to be able to generalize the results, as mentioned in the Sample strategy. The right sample size to represent the population was 385 respondents. We managed to get 201 respondents which increases our margin of error to ±6.91%.

3.14 Limitations The limitation that were taken under consideration during this study is that we only examine the preferences of Swedish consumers when it comes to groceries. The result can be hard to justify as a Swedish behaviour when it comes to other product categories. Another limitation is that this study was made without any research agencies so the distribution of the survey was made through out our personal network. This limits the spread when it comes to number of participants and can also affect the geographical areas within Sweden and social status situations among the respondents.

After analysing the age spread of our 201 respondents, we could see a clear pattern of younger participants which will limit the way the sample represent the population. The same goes for the education variable when out of the 201 respondents only 2 said say only had “Grundskola” education.

4. Empirical results Our questionnaire was based on 12 questions about carbon label design and effects and 3 questions about our respondents demographics. The first 5 questions are presented as “what do you prefer” and the respondents are presented with two different kinds of label designs which they could only choose one alternative. The questions 6-12 are based on the Likert-scale from 1- 5 where 1 is “Low valued”, “Uninteresting” or “minimal impact” and where 5 is “high value”, “Very Interesting” or “big impact”. We had 201 respondents. The questionnaire was in Swedish and all the respondents was Swedish citizens.

4.1 Questionnaire answers

Fig 4.1 In this pie chart Red is the amount of female respondents and blue is the amount of male respondents. There were in other words 58,2 % Women and 41,8% Men

Fig 4.2 This chart shows the age interval of the respondents.

Fig 4.3 In this pie chart red is Swedish gymnasial education, yellow is a bachelors degree and green is masters or above.

Fig 4.4 We asked the respondents to estimate how much knowledge they have about GHG emissions from groceries. On a scale of 1 - 5 where 1 is the lowest and 5 is the highest. This table (Fig 4.4) shows the respondents estimations. The mean value of this question is: 2.73

Fig 4.5 Question 1 Question 1 was designed to find out if the respondents prefer a label that includes data or not. The answers show that a majority (73,1%) prefers the second label that includes data.

Fig 4.6 Question 2 Question 2 was designed to find out if the respondents preferred a traffic light colour coding system or a monochrome single colour system. The respondents were presented two labels where both included data, but one was presented in a coloured traffic light system and the other one was singe coloured. 89.1% preferred the label that was presented using the traffic light system.

Fig 4.7 Question 3 Question 3 was designed to find out if the respondents preferred to have a scale or a more simple colour coded label with less information. The respondents were presented two labels where both included data and traffic light colours but with one also being presented in terms of a scale. 78,6% of the respondents preferred when the label was presented with a scale.

Fig 4.8 Question 4 Question 4 was designed to find out if the respondents preferred a well known organization who does not work with greenhouse gas emissions compared to a lesser known organization that does work with it. We used a Carbon Trust label as our example of an unknown organization in Sweden and we used our experimental version of the Krav logotype in black and marked it with a “CO2e”. 74% of respondents chose the Krav logotype.

Fig 4.9 Question 5 Question 5 was designed to find out if the respondents preferred to have a label which puts CO2e emissions in relation to the amount of calories the product has as opposed to show the emissions in relation to its weight. The results show that 81.1% chose the label where the data was presented in CO2e/kg.

Fig 4.10 Question 6 Question 6 was designed to find out how much the respondents would value that the label was created and maintained by an already known organization. The mean value of this question is: 3,91

Fig 4.11 Question 7

Question 7 was designed to find out how much the respondents would value that the CO2e data was presented on the label. The mean value of this question is: 3,77

Fig 4.12 Question 8 Question 8 was designed to find out how much the respondents would value that the label was colour coded. The mean value of this question is 4,22

Fig 4.13 Question 9 Question 9 was designed to find out how much the respondents would value that the label was presented on a scale. The mean value of this question is 3,84

Fig 4.14 Question 10 Question 10 This question was designed to see the estimated influence a carbon label would have on the respondents when purchasing groceries. The mean value of this question is 3,78

Fig 4.15 Question 11 Question 11 was designed to see if the respondents would be interested in having access to information about the process of calculating the CO2e values used in carbon labeling. The mean value of this question is 3,40

Fig 4.16 Question 12 Question 12 is divided into 6 parts designed to put the estimated influence of a carbon label in relation to the estimated influence of known sustainability or eco labels. The mean for KRAV is 3,76 The mean for MSC is 3,14 The mean for a carbon label is 3,48 The mean for Fair trade is 3,66 The mean for Nyckelhålet is 3,04 The mean for EU-lövet is 2,60

4.2 Comparisons within sample

Fig 4.17 We created to graphs depicting answers of question 7 (How much they would value having data on the label) into two parts depending on whether they had a bachelors degree or higher or gymnasial degree and lower to see what the differences in the results were.

The average value of the less educated were 3,71 and the average value of the higher educated were 3,79.

Fig 4.18 We divided the sample into a lower half and an upper half in terms of the respondents age in order to see the differences in the estimated influence that a carbon label would have on their grocery shopping.

The average value of the younger half were 3,69 and the average value of the older half were 3,86.

Fig 4.19 We divided the sample into a lower and an upper half in terms of the respondents estimated level of knowledge regarding CO2e emissions of grocery products. We wanted to see if there was a difference in the value for having access to information about the process of calculating the CO2e values used in carbon labeling.

The average value of the half that had lower perceived knowledge was e 3,24 and the average value of the higher educated were 3,55.

4.3 Correlation analysis In order for us to see the relation between our datasets we used the Pearson correlation which is a measurement for linear relationships between two variables. The valid range of a correlation range goes from -1 to 1 where -1 is a perfect negative correlation, 0 is no correlation and 1 is a perfect positive correlation. Correlation provides evidence of association and not causation (Saunders et al., 2016).

In the following headers we have provided correlation matrices which we found significant values of. Divided into our different attribute variables. We did this for all our questions and decided only to show the data which proved the most significant.

4.3.1 Gender The Pearson correlation test proved only a minor if any correlation on question 8, 9 and 10 indicating that female respondents tended to pick a higher value on the respective questions:

Fig 4.20

Question 8. Colour coded Pearson -,169* Correlation Sig. (2-tailed) 0,016

Question 9. Scale Pearson -,192** Correlation Sig. (2-tailed) 0,006

Question 10. Impact of carbon Pearson -,231** label Correlation Sig. (2-tailed) 0,001

Question 6, 7 and 11 proved no correlation between genders.

Fig 4.21 The following Pearson correlation matrix shows the relation between the appreciated influence a given eco label would have on the respondents purchasing behaviour with gender included as a variable were females were recoded into 1 and males recoded into 2.

Carbon Fair Nyckel EU Organic Gender KRAV MSC Label Trade hålet logo Pearson Gender Correlation 1 -,193** -,210** -,248** -,210** -,140* −0,091

Sig. (2-tailed) 0,006 0,003 0,000 0,003 0,048 0,197 Pearson KRAV Correlation -,193** 1 ,442** ,570** ,691** ,523** ,496**

Sig. (2-tailed) 0,006 0,000 0,000 0,000 0,000 0,000 Pearson MSC Correlation -,210** ,442** 1 ,462** ,457** ,239** ,443**

Sig. (2-tailed) 0,003 0,000 0,000 0,000 0,001 0,000 Pearson Carbon Label Correlation -,248** ,570** ,462** 1 ,561** ,403** ,520**

Sig. (2-tailed) 0,000 0,000 0,000 0,000 0,000 0,000 Pearson Fair Trade Correlation -,210** ,691** ,457** ,561** 1 ,549** ,504**

Sig. (2-tailed) 0,003 0,000 0,000 0,000 0,000 0,000 Pearson Nyckelhålet Correlation -,140* ,523** ,239** ,403** ,549** 1 ,585**

Sig. (2-tailed) 0,048 0,000 0,001 0,000 0,000 0,000 EU Organic Pearson logo Correlation −0,091 ,496** ,443** ,520** ,504** ,585** 1

Sig. (2-tailed) 0,197 0,000 0,000 0,000 0,000 0,000

The negative correlation for gender shows a very weak or non existing correlation indicating that women tend to be more influenced by the eco labels highlighted in green. There are medium strength correlations between the labels suggesting that a respondent that chose a high value of one eco-label also chose a high value for another. The correlations are marked in blue.

4.3.2 Age The correlation test proved only a minor if any correlation on question 6 indicating that the older the respondent they prefer a more known organization behind the carbon label. There were also a small if any correlation indicating that the younger the respondent, the less likely he/she is to prefer the colour coded design.

Fig 4.22

Pearson Question 2. Colour Correlation −0,207*

Sig. (2-tailed) 0,003 N 201

Question 6. Existing organization Pearson ,159* Correlation Sig. (2-tailed) 0,025

Question 1 - 5 and 7 - 11 proved no correlation with age differences.

4.3.3 Education No correlation was found related to education and answers in question 1 - 11. 4.3.4 Perceived Knowledge The correlation tests for question 1 - 5 which were more design oriented and was created to show the respondents what the labels could look like proved no significant correlation between perceived knowledge of greenhouse gas emissions from grocery products and the respondents choices regarding layout and content of the carbon label examples. The only number of significance is the 0,15 Pearson correlation in question 1 which indicates there are none or a very small correlation between perceived knowledge and the label which included data.

Fig 4.23

Pearson Question 1. Data or not Data Correlation 0,15*

Sig. (2-tailed) 0,033

Fig 4.24 The correlation test between perceived knowledge and question 6 - 11 proved only a minor if any correlation in question 7 and 11 which is indicating a weak correlation with perceived knowledge of greenhouse gas emissions from grocery products and the appreciated impact a carbon label would have on their purchasing behaviour and the value of having data on the labels. In question 10 there is a stronger but still relatively weak correlation between perceived knowledge and appreciated impact a carbon label would have on their purchasing behaviour.

Question 7. Data in label Pearson 0,227** Correlation Sig. (2-tailed) 0,001

Question 10. Impact of carbon Pearson 0,32** label Correlation Sig. (2-tailed) 0

Question 11. Access to information Pearson 0,226** Correlation Sig. (2-tailed) 0,001

No correlation was found related to perceived knowledge and the answers in question 6, 8 and 9.

5. Analysis

5.1 Label design

5.1.1 Colour coded vs. Monochrome Previous studies claims that the most effective way to produce a carbon label is to make it as simple as possible. If you don’t decide to have a breakpoint at a certain CO2e value where you only use a label as a “brand of approval”, but having a label to help consumers to compare different products a successful method is to use a colour coded label (Sharp & Wheeler, 2013).

This method is also preferred by the Swedish consumers. 89.1% (Fig 4.6) preferred the colour coded label using a traffic light system with the colours green, yellow and red over the label that was presenting the same information but in a simple black colour. On the question on how highly they valued that the label was colour coded, on a scale of 1-5 the mean value was 4,22 (Fig 4.12). This strengthens the theory that this is a preferred way to communicate a label. This also suggest that the consumers would like to be able to compare products with each other with the simplicity of using colours.

After making a correlation analysis we found no correlation between age, education and perceived knowledge related to whether they wanted colour coded or not. Only a minor Pearson Correlation of -,169* with a sig. of 0,016 in favour of female value of colour in a label could be found which is hardly any proof of correlation between gender and value of having colour coding in a carbon label.

Based on our empirical findings we found support for our hypothesis (H1): Swedish consumers prefer colour coded labels.

5.1.2 With data vs. Without data Sharp and Wheeler’s research stated that presenting data on the label will create confusion and that the Australians wanted the label to be easy interpreted and simplified. Our results in this matter shows that the Swedish people prefer to have the CO2e data presented on the label (Fig 4.5, 4.11, 4.17). When divided into different educational level; Gymnasial and below, and Bachelor degree and higher, each mean value suggest that both groups regardless of educational background prefer to have the data presented on the label. This is further strengthened by the fact that there is not any correlation between education and any of the questions asked in the questionnaire. The same goes for age and gender. There is only one very small if any correlation between question 1 and perceived knowledge of 0,15 Pearson correlation with a 0,033 significance (Fig 4.23).

Other studies that was made in the English and Belgian markets, both stated that that the carbon labels were hard to interpret and needed to be presented in a simple way. The question that can be asked is what is simple and easily interpreted and what is not? Clearly, a majority of our respondents finds that data presented on the label is the preferred approach. This can be part of other factors such as our respondents also preferred the data to also be presented together with a traffic light system where each value group is colour coded. So the theory that data presented on the label is confusing may be related to how the data is presented. By having the consumers demand in mind, presenting the data the right way it can help the consumers to compare products to each other and make a more sustainable choice. (Loo et al., 2014; Gadema & Oglethorpe, 2011; Sharp & Wheeler, 2013)

A lot of the challenges that has been but forth in theories boils down to the data is too complex to help the consumers make a smart sustainable choice when grocery shopping. The mean value of how much the respondents valued that data is presented on the label was 3.77 (Fig 4.11), while the mean value of the respondents that valued having access to the information about the process of how the CO2e-value is calculated is 3.40 (Fig 4.15).

Previous researchers have been assuming that, for a carbon labeling programme to work, just having data on the label for comparison is not good enough, meaning people want to know more about the data and have more information about the calculation process (Sharp & Wheeler, 2013). This may be partly true due to challenges of trusting a label when there are several different ones and especially if the producers of different products make their own label. However, our results suggest that there is a bigger part of the population that wants the label to be presented with data than the part that want to know how the data is calculated.

The almost unanimous answer that the respondents wanted the label to be colour coded, with the help of data it could substantially benefit the producer/products that are within one code of colour but has a lower emission value presented than other products with higher values but within the same colour code. As more consumers want data on the label, there is also a demand for information about the calculation process (Fig 4.15). This is a layout challenge since this information can not be part of the content on the label simply because it is too much information. A possible solution to this is to have this information accessible elsewhere. Either on the website of the organisation that is maintaining the label or possibly via a QR-code accessible from the grocery stores.

Gadema & Oglethorpe claims that people are only willing to substitute a higher CO2e product to a lower CO2e product when the quality and price are similar or lower. When a product category like meat who will probably always be within the red category, data can help people make the most sustainable choice within the same category (Gadema & Oglethorpe, 2011).

Based on our empirical findings we found support in our hypothesis (H2): Swedish consumers prefers data to be presented on the label.

5.1.3 With scale vs. Without scale To present the label with a scale could either be to simplify the information that was going to be given, or it could be more complex towards the respondents due to it is more information and design to embrace. Our results however states that our respondents prefer the label when it is presented with a scale (Question 3 Fig 4.7). 78.6% claimed that they preferred the label with a scale over the one only presented in one single colour. The mean of question 9 (Fig 4.13) is 3.84 is also an indication of a high value and preference of having a scale on a carbon label. Our interpretation of that is that a majority of our respondents believe they obtain more valuable information when being presented with a scale than without. There is no indication of any correlation between age, education and perceived knowledge and the preference of a scale in a carbon label. There is only a small Pearson correlation of -0,192 with a significance of 0,006 that indicate that females have a weak or no correlation in valuing scale higher than men (Fig 4.20).

The overall results are however a scenario where the majority chose the more complex design (Fig 4.7). This evidence contradicts the theories that the labels needs to be simplified with less information in order for them to have an effect among consumers (Sharp & Wheeler, 2013).

The high percentage of respondents that preferred a label using a scale is something that need to be taken under consideration when designing a label. The almost unanimous choice might have to do with the added level of possibility to compare between products. Based on previous studies people would like to be able to compare products within the same product category. Since the scale adds another comparison level this also enables people to compare products within the same category. (Hartinkainen et al., 2014; Sharp & Wheeler, 2013)

What our respondents did not have to have in mind when they were presented with this question is the placement of the label. A label using a scale will need a bigger amount of space since it will have to contain more elements to the design. To make the label visible and accessible a preferred placement of the label would be next to the price tags in the grocery store. By placing it next to the price tag you also take the people's shopping habits under consideration since they already have an existing pattern of looking at the price tag. (Leach et al., 2016)

The results show that a label using a scale is the preferred by the respondents. However, the challenge of the placement is something that designers need to have in mind when designing a label using a scale.

Based on our empirical findings we found support in our hypothesis (H3): Swedish consumers prefer the label being presented using a scale.

5.1.4 Well known organization vs. Lesser known organization Question 4 (Fig 4.8) and question 6 (Fig 4.10) were designed to gain insight in our respondents thoughts regarding the organization behind the label. In theory it has been concluded that it is highly unlikely that a carbon label would develop simply from entrepreneurship and non obligation or from a retailer collaboration (Gadema & Oglethorpe, 2011). A carbon label programme that would span across all groceries would likely require governmental control or influence. The question then would be, what organization would or should be responsible for the implementation and maintenance of the carbon label.

By looking at the results of question 4 - 74% preferred our own experimental label of an extension of KRAV as opposed to a lesser known but real carbon label from the UK. in Question 6 the mean value was 3,91 which also suggests that it would be a good choice to implement a carbon label through an already existing brand or organization. Although the comparison might not entirely fair since Carbon Trust is not owned or controlled by the government. It was the best we could use since we couldn’t find an actual government controlled carbon label programme so but we chose one of the most commonly used carbon label in the UK. (carbontrust.com, 2020; Peano et al., 2015)

As previously concluded in our theory it is highly unlikely that a carbon label programme could successfully be implemented without government intervention or control (Gadema & Oglethorpe, 2011). The Carbon Trust approach would most likely be the only way possible to do it without government control. However, the obstacle of having only a logotype based carbon label is that the business model is built around the idea that producers and companies themselves actively work towards certifying their products but has no real solution to the producers and companies that do not want to label the products.

Our results shows that the respondents prefer additional features, such as colour, scale and data on the label (Fig 4.5, 4.6, 4.7) to help the consumers compare the different products. This suggest that even if a well known organisation implemented a carbon label, which is preferred by the consumers (Fig 4.6), the sought after effect would be absent if they only included their logo. This is a challenge for the organization, to include all the features the consumers want but still be able to communicate that they are the one maintaining the label.

In other words, this approach would not be possible should a decision of a government controlled carbon label scheme that involves all grocery products be implemented. Which according to Gadema & Oglethorpe is the most likely scenario. However, it is clear from our empirical findings that if KRAV would branch and create a logo with the same structure as Carbon Trust - Swedish people would most likely value it. (Thøgersen & Nielsen, 2015; Gadema & Oglethorpe, 2011)

Arguably that “well known organisation” term is a bit ambiguous. This would be since an unknown organisation can create a label/brand and within a short period of time become a well known organisation especially with government support. Our findings related to this indicates the importance of trust.

Based on our empirical findings we found support in our hypothesis (H4) Swedish consumers prefers that the label is being produced by a well known organization

5.1.5 Per weight vs. per calorie We were doubtful as to if we could make this subject easy to understand and test in a questionnaire, but we decided to try it in question 5. (Fig 4.9) The main argument of having a carbon label where the data is based on the amount of calories the product have is to give an indication of how much hunger relief that specific product has. E.g. it is likely that 100 grams of beef is going to generate more hunger relief than 100 grams of apple which also includes some non edible parts. This could arguably mean that the two are really not comparable at all.

If we compare the percentage differences between milk and beef in the two examples used in question 5 in the example where it is measured in terms of kg of product the beef has got 31,11 times as much emissions as the milk. In the other example when it is based on calories, the beef only has 5,6 times the amount of emissions as milk. However, calories can be misleading as well as it is likely that a bag of 100 kcal potato chips would not be as filling as the equivalent 100 kcal of beef.

The result among our respondents were 81.1% in favour of carbon data in relation to the products weight (Fig 4.9) which was to be expected and is also usually the way carbon data is presented on actual carbon labels if not the total value of the product itself is printed. This gives a clear indication as to what is preferred in our sample.

Based on our empirical findings we found support in our hypothesis (H5) Swedish consumer prefers a label where data is presented in terms of weight.

5.2 Further analysis

About 10% of the population would not change their purchasing behavior if a carbon label was implemented (Hartinkainen et al., 2013). Those are the lost cause group that a carbon labeling system has to avoid. By having the labeling impact question (Fig 4.16) we were able to trace the answers and see if we had any of these people amongst our respondents who claimed no label would have any impact on their purchasing behavior. We could not see a clear pattern so none of the answers were excluded.

We asked our respondents to estimate their perceived knowledge about CO2 emission (Fig 4.4) to see if their preferences would differ. We could see a weak correlation between the people who estimated that they had a high perceived knowledge would more likely want to have information about the calculating process (Fig 4.19, 4.23, 4.24), but not when it came to design attributes. This suggests that even though people have a low pre existing knowledge about greenhouse gas emission they prefer the same design. To increase carbon literacy, consumers need to be educated. A more complex carbon label with more information could arguably help to increase carbon literacy among the people who have low perceived knowledge or low carbon literacy. The result of a more “educating” label might be the increase in carbon literacy among the whole population in the long run. This in turn would most likely increase the interest of having access to information about the calculation process.

One of the main limitations in this study was the fact that the distribution of ages did not reflect the distribution of ages in the population, however when conducting a Pearson correlation test (Fig 4.22) there was no evidence of a correlation indicating that opinions regarding the design, layout and content of a carbon label would be different between the ages.

Overall the impact of a carbon label would have an influential effect (Fig 4.14). It also states that women (Fig 4.20) and people over the age of 29 tended to value a carbon label higher (Fig 4.18) and that a respondent that chose a high value of one eco-label also chose a high value for another (Fig 4.21). This shows that the right social ethos (Li et al, 2017) for a carbon labeling scheme is the LOHAS, which tend to be well educated women over 35 (Bosona & Gebresenbet, 2018). A criteria for the carbon label to work is that consumers embrace the information that is given on the label (Shewmake et al., 2015). Our results states that there is an existing pattern of looking at the information at the different labels which also indicates that implementing a carbon label would not affect their shopping habits (Leach et al., 2016).

As mentioned in the theory, Tescos attempt to implement a carbon label was discontinued due to the expensive and time consuming work and that no other retailer followed in their footsteps (Thøgersen & Nielsen, 2015). This suggest that one retailer alone can not implement a carbon labeling system, but they would need all the retailers support to implement the system. This scenario seems more unlikely to happen since Tesco already have tried. This however only strengthens the theories that implementation would need to be under governmental control to make it achievable (Thøgersen & Nielsen, 2015; Cohen & Vandenbergh, 2012).

The most realistic scenario for a carbon labeling programme to work would then be with a government controlled implementation and maintenance, which would need to cover all grocery products and not just a select few, since labeling only one or a few products could have the negative counter effect of consumers simply picking the products which are not labeled (Shewmake et al., 2015; Gadema & Oglethorpe, 2011). Since all products would need to be labeled one of the challenges that needs to be considered is the placement and the size of the label. A supposable placement could be next to the price tags on the shelves. Our respondents most valued attribute according to its mean value is the traffic light colour coded system with 4,22 (Fig 4.12) and the second most valued attribute was that the carbon label was created from an already existing organization with a mean of 3.91 (Fig 4.10). Scale was valued at the mean of 3.84 (Fig 4.13).

It is highly unlikely that a price tag could fit an organization's logotype and a traffic light colour coded scale on its surface area, which would mean that perhaps both the organization and scale would have to be excluded in its most minimalistic layout.

Our suggestion then would be to create one more advanced logotype for the situations when there is more space to be had and one more minimalistic with perhaps only traffic light and data values on them.

6. Conclusion The purpose of this study was to get a deeper understanding into what the preferences are among Swedish consumers when it comes to carbon labeling in grocery stores, we wanted to explore different approaches to carbon labeling and test them and ultimately answer the following question.

What is the preferred content, design and layout of a carbon label among Swedish consumers?

We created five hypothesis based on the theory and we found support on all five:

Hypothesis 1 (H1): Swedish consumers prefer colour coded labels Hypothesis 2 (H2): Swedish consumers prefer data on the label Hypothesis 3 (H3): Swedish consumers prefer the label being presented using a scale. Hypothesis 4 (H4): Swedish consumers prefer that the label is being produced by a well known organization. Hypothesis 5 (H5): Swedish consumers prefer a label where data is presented in relation to the products weight.

1. The highest valued attribute among our respondents were the traffic light colour code with 89,1% choosing it in the illustrated question and had a mean value of 4,22 when asked how much they value it. 2. The second highest valued attribute were that the organization behind the carbon label was a previously known organization even though carbon footprinting is not their area of services. 74% of respondents picked the KRAV logotype in our illustrated question and it had a mean value of 3,91 when asked how much they value it. 3. The third most valued attribute was the scale with a 78,6% support in the illustrated question and it had a mean value of 3,84 when asked how much they value it. 4. The fourth most valued attribute was to have the data presented on the label with a 73,1% support in the illustrated question and it had a mean value of 3,77 when asked how much they value it.

In previous studies they found that a carbon label when presented with complex information was difficult for consumers to interpret and that the CO2e data was too complex to communicate in a way which could encourage a more climate friendly purchase. On all accounts we found evidence in our empirical findings for the support of all of our hypotheses which suggests that Swedish consumers prefer a more complex design in every scenario. It is likely that it is not Swedish consumers preferences or understanding of the labels that would limit the design, layout and content of the label. It is rather the limitations of how the label itself should or could be implemented in a real life scenario that would be the biggest obstacle of realizing a carbon label programme in Sweden. We found that Swedish consumers were quite unanimous when it came to their preferences and could not see any correlation of importance when dividing our sample within age, education, gender or perceived knowledge. 7. Further Studies Our suggestion for further studies is to test a carbon label that is designed according to the preferences from our study. It would be interesting to see if the outcome would be as our study predicts. Another interesting angle would be to examine how the producers and the Swedish retailers would react and how they would be affected if a labeling system like this was implemented that involves labeling all grocery products.

References

Books

Bell, E., Bryman, A., Bryman, A., & Harley, B. (2019). Business research methods (Fifth edition). Oxford: , Oxford University Press.

Saunders, M., Lewis, P., & Thornhill, A. (2016). Research methods for business students (7. ed.). Harlow: Pearson Education.

Articles

Chiriacò, M., Grossi, G., Castaldi, S., & Valentini, R. (2017). The contribution to climate change of the organic versus conventional wheat farming: A case study on the carbon footprint of wholemeal bread production in Italy. Journal of Cleaner Production, 153(1), 309–319. https://doi.org/10.1016/j.jclepro.2017.03.111

Chuanmin, S., Xiaomin, Y., Yukun, Z., Chuanxi, S., & Penghui, D. (2014). Consumer behaviour on low-carbon agri- food purchase: A carbon labelling experimental study in China. Agricultural Economics (Czech Republic), 60(3), 133– 146. https://doi.org/10.17221/20/2013-AGRICECON

Cohen, M., & Vandenbergh, M. (2012). The potential role of carbon labeling in a green economy. Energy Economics, 34(S1), S53–S63. https://doi.org/10.1016/j.eneco.2012.08.032

Emberger-Klein, A., & Menrad, K. (2018). The effect of information provision on supermarket consumers’ use of and preferences for carbon labels in Germany. Journal of Cleaner Production, 172, 253–263. https://doi.org/10.1016/j.jclepro.2017.10.105

Feucht, Y., & Zander, K. (2018). Consumers’ preferences for carbon labels and the underlying reasoning. A mixed methods approach in 6 European countries. Journal of Cleaner Production, 178, 740–748. https://doi.org/10.1016/j.jclepro.2017.12.236 Gadema, Z., & Oglethorpe, D. (2011). The use and usefulness of carbon labelling food: A policy perspective from a survey of UK supermarket shoppers. Food Policy, 36(6), 815–822. https://doi.org/10.1016/j.foodpol.2011.08.001

Hartikainen, H., Roininen, T., Katajajuuri, J., & Pulkkinen, H. (2014). Finnish consumer perceptions of carbon footprints and carbon labelling of food products. Journal of Cleaner Production, 73, 285–293. https://doi.org/10.1016/j.jclepro.2013.09.018

Heller, M., & Keoleian, G. (2015). Greenhouse Gas Emission Estimates of U.S. Dietary Choices and Food Loss. Journal of Industrial Ecology, 19(3), 391–401. https://doi.org/10.1111/jiec.12174

Hoang Viet Nguyen, Ninh Nguyen, Bach Khoa Nguyen, Antonio Lobo, & Phuong Anh Vu. (2019). Organic Food Purchases in an Emerging Market: The Influence of Consumers’ Personal Factors and Green Marketing Practices of Food Stores. International Journal of Environmental Research and Public Health, 16(6). https://doi.org/10.3390/ijerph16061037

Leach, A., Emery, K., Gephart, J., Davis, K., Erisman, J., Leip, A., … Galloway, J. (2016). Environmental impact food labels combining carbon, nitrogen, and water footprints. Food Policy, 61(C), 213–223. https://doi.org/10.1016/j.foodpol.2016.03.006

Li, Q., Long, R., & Chen, H. (2017). Empirical study of the willingness of consumers to purchase low-carbon products by considering carbon labels: A case study. Journal of Cleaner Production, 161, 1237–1250. https://doi.org/10.1016/j.jclepro.2017.04.154

Liu, T., Wang, Q., & Su, B. (2016). A review of carbon labeling: Standards, implementation, and impact. Renewable and Sustainable Energy Reviews, 53(C), 68–79. https://doi.org/10.1016/j.rser.2015.08.050

Nicholls, N. (2007). Climate: Sawyer predicted rate of warming in 1972 [3]. Nature, 448(7157), 992. https://doi.org/10.1038/448992c

Peano, C., Baudino, C., Tecco, N., & Girgenti, V. (2015). Green marketing tools for fruit growers associated groups: application of the Life Cycle Assessment (LCA) for strawberries and berry fruits ecobranding in northern Italy. Journal of Cleaner Production, 104, 59–67. https://doi.org/10.1016/j.jclepro.2015.04.087

Sharp, A., & Wheeler, M. (2013). Reducing householders’ grocery carbon emissions: Carbon literacy and carbon label preferences. Australasian Marketing Journal (AMJ), 21(4), 240–249. https://doi.org/10.1016/j.ausmj.2013.08.004

Shewmake, S., Okrent, A., Thabrew, L., & Vandenbergh, M. (2015). Predicting consumer demand responses to carbon labels. Ecological Economics, 119, 168–180. https://doi.org/10.1016/j.ecolecon.2015.08.007

Techane Bosona, & Girma Gebresenbet. (2018). Swedish Consumers’ Perception of Food Quality and Sustainability in Relation to Organic Food Production. Foods, 7(4). https://doi.org/10.3390/foods7040054

Thøgersen, J., & Nielsen, K. (2016). A better carbon footprint label. Journal of Cleaner Production, 125, 86–94. https://doi.org/10.1016/j.jclepro.2016.03.098

Treu, H., Nordborg, M., Cederberg, C., Heuer, T., Claupein, E., Hoffmann, H., & Berndes, G. (2017). Carbon footprints and land use of conventional and organic diets in Germany. Journal of Cleaner Production, 161, 127–142. https://doi.org/10.1016/j.jclepro.2017.05.041

Van Loo, E., Caputo, V., Nayga, R., Seo, H., Zhang, B., & Verbeke, W. (2015). Sustainability labels on coffee: Consumer preferences, willingness-to-pay and visual attention to attributes. Ecological Economics, 118, 215–225. https://doi.org/10.1016/j.ecolecon.2015.07.011

Wrenfelt, Peter., Dahlgren, Katrin., Grant, Johanna. (2018). Analys av MAX Burgers AB:s klimatpåverkan år 2017, https://www.max.se/globalassets/download-files/max-metodrapport-och-resultat-klimat-2017- 180604-final.pdf

Zander, K., Brgelt, D., Christoph-Schulz, I., Salamon, P., & Weible, D. (2015). Consumers’ Response to Sustainability Labeling in Wild Caught Fish. IDEAS Working Paper Series from RePEc. Retrieved from http://search.proquest.com/docview/1767757308/

Webpages

Calculator.net (2020), Retrieved 2020 January 28, https://www.calculator.net/sample-size- calculator.html?type=2&cl2=95&ss2=201&pc2=96.8&ps2=7900000&x=80&y=22#findci

Carbon Trust (2019), Retrieved 2019 October 29, https://www.carbontrust.com/what-we-do/assurance-and- certification/product-carbon-footprint-certification-and-labelling

Ecolabel Index (2019), Retrieved 2019 November 02, http://www.ecolabelindex.com/

Google Trends (2019), Retrieved 2019 October 22, https://trends.google.com/trends/explore?date=today%205- y&geo=SE&q=flygskam

Google Trends (2019), Retrieved 2019 October 22, https://trends.google.com/trends/explore?q=Greta%20Thunberg

Google Trends (2019), Retrieved 2019 October 22, https://trends.google.com/trends/explore?date=today%205- y&q=sustainability

NASA (2008), Retrieved 2019 October 22, https://www.nasa.gov/topics/earth/features/vapor_warming.html

Rasio (2020), Retrieved 2020 March 13, https://www.raisio.com/carbon-footprint-label

Rise (2019), Öppna Listan, Retrieved 2020 March 28, https://www.ri.se/sv/media/906/download

Statistiska Centralbyrån, (2020), Sveriges befolkningspyramid, Retrieved 2020 March 27, https://www.scb.se/hitta- statistik/sverige-i-siffror/manniskorna-i-sverige/sveriges-befolkningspyramid/

Swedavia (2019), Swedavia’s passenger statistics for April 2019, Retrieved 2019 October 22, https://www.swedavia.com/about-swedavia/news/swedavias-passenger-statistics-for-april-2019/

United Nations Conference on Trade and Development, (2017), Opportunities for enhancing markets for organic products via tourism sectors discussed at the East African Organic Policy Forum, Retrieved 2019 October 22, https://unctad.org/en/pages/MeetingDetails.aspx?meetingid=1463

UN Global Goals (2019), Retrieved 2019 October 22, https://www.globalgoals.org/

Edvin Elghag

Eric Sundberg

PO Box 823, SE-301 18 Halmstad Phone: +35 46 16 71 00 E-mail: [email protected] www.hh.se