Grocery shopping, food waste, and the retail landscape of cities: The case of Seoul

Grocery shopping, food waste, and the retail landscape of cities: The case of Seoul

Accepted Manuscript Grocery shopping, food waste, and the retail landscape of cities: the case of Seoul Keith C.L. Lee PII: S0959-6526(17)32379-X D...

664KB Sizes 0 Downloads 45 Views

Accepted Manuscript Grocery shopping, food waste, and the retail landscape of cities: the case of Seoul

Keith C.L. Lee PII:

S0959-6526(17)32379-X

DOI:

10.1016/j.jclepro.2017.10.085

Reference:

JCLP 10877

To appear in:

Journal of Cleaner Production

Received Date:

07 March 2017

Revised Date:

31 August 2017

Accepted Date:

09 October 2017

Please cite this article as: Keith C.L. Lee, Grocery shopping, food waste, and the retail landscape of cities: the case of Seoul, Journal of Cleaner Production (2017), doi: 10.1016/j.jclepro. 2017.10.085

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

-

Examined household food waste’s connection to food retailing using survey data

-

Food retail formats were distinguished by over-purchasing and trip characteristics

-

Over-purchasing and food waste reduced by short travel times and frequent trips

-

Hypermarket format implicated in contributing to household food waste

ACCEPTED MANUSCRIPT

Grocery shopping, food waste, and the retail landscape of cities: the case of Seoul Keith C. L. Lee a Department of City and Regional Planning University of California, Berkeley a

University of California, Berkeley 228 Wurster Hall #1850 Berkeley, CA 94720-1850 United States of America [email protected]

ACCEPTED MANUSCRIPT

Abstract The reasons for household food waste are complex and not yet fully understood, especially with regards to how the food retail sector might influence and interact with food management practices. This study explores the causes of household food waste from this perspective, drawing from practice theory and the concept of systems of provision. Survey data from 460 Seoul households is analyzed using bivariate and multivariate analyses to assess whether avoidable food waste per household member is affected by the type of food retailers (e.g. supermarkets, hypermarkets, traditional markets, etc.) a household buys its groceries from. Seoul was chosen as the research site due to its diverse food retail sector and South Korea's distinctive food waste policy environment. The results indicated that households' grocery shopping trips take on different characteristics depending on the type of retailer visited, and that these characteristics, including travel time and buying frequency, are differentially associated with over-purchasing and avoidable food waste. Overall, this study suggests that the influence of food retailers on household food waste is not limited to marketing promotions, but also extends to the ways they may shape households' grocery shopping patterns. As such, waste reduction policies and initiatives need to take urban retail development patterns into greater account in order to more effectively tackle the problem of household food waste.

Keywords: food retail; food waste; grocery shopping; practice theory; South Korea; systems of provision

ACCEPTED MANUSCRIPT

1.1. Introduction Food waste has risen on research and policymaking agendas due to increased recognition of its pervasiveness and its detrimental environmental, social, and economic impacts (Gustavsson et al., 2011; Lundqvist et al., 2008). In industrialized nations, urban households are a major source of food waste (Parfitt et al., 2010), and research has attempted to explain consumers’ wasterelated behaviors to improve food waste-reduction policies and promote source-separation. Quantitative research indicates household size, attitudes towards food waste, and food-related habits (e.g. making shopping lists) as drivers of household food waste (Farr-Wharton et al., 2014; Mondéjar-Jiménez et al., 2016; Parizeau et al., 2015; Ponis et al., 2017; Stefan et al., 2013). Qualitative studies, often employing a practice-based approach, emphasize how food waste emerges from households’ difficulties in coordinating food-related practices, such as grocery shopping and meal preparation, with other priorities in everyday life (Evans, 2014; GrahamRowe et al., 2014). Studies have also examined food retailers’ influence on household food waste, but their findings are limited to the role that marketing promotions play in encouraging excess food purchases (Farr-Wharton et al., 2014; Mondéjar-Jiménez et al., 2016). Drawing from a theoretical framework influenced primarily by the practice-based approach and the related concept of systems of provision, this study investigates the influence of food retailing on household food waste by examining households’ grocery trip characteristics and their relationships to different kinds of food retailers.

To this end, a survey was conducted in Seoul, South Korea. Seoul was chosen as the research site for three reasons. First, South Korea’s volume-pricing system for food waste disposal requires most households to prepay for food waste disposal, raising consumers’ awareness about how

ACCEPTED MANUSCRIPT

much food waste they discard. This was anticipated to reduce recall error in the survey. Second, South Korea’s food retail sector comprises a mix of supermarkets, hypermarkets, convenience stores, family-run stores, and traditional markets. This diversity was theorized to facilitate a comparison of how various food retail types are connected to variations in consumers’ everyday lives, and consequently, household food waste. Third, there remains a relative shortage of household food waste research in non-Western contexts.

Overall, survey data analysis provided evidence that food retail type affects avoidable household food waste, and that this stems from factors connecting land use, transportation, and shopping frequency. These results suggest that consumers’ grocery shopping routines connect with particular food retail formats to influence household food waste generation. The connections uncovered between retail types, grocery shopping routines, and household food waste suggest that consumer habits and the household food waste problem may be inextricable from current modes of retail planning and urban development.

1.2. Theories of practice, systems of provision, and consumption This study employs a practice-based approach, which refers to its grounding in several theories of practice. Increasingly popular in sustainable consumption research, a practice-based approach is distinctive in its identification of practices, not individuals, as the unit of analysis when analyzing social issues. A practice can be defined as a recognizable, “routinized type of behavior” (Reckwitz, 2002, p. 249) that is broadly understood by society as being comprised of specific “doings and sayings” (Schatzki, 1996, p. 89) that could be practices too. For example, cooking as a practice combines simpler practices such as selecting ingredients, chopping, or

ACCEPTED MANUSCRIPT

frying. Individuals are said to perform a practices; this translates a practice from an abstract notion into a concrete action, thereby sustaining its existence over time (Schatzki, 1996). The nature of these performances can vary under influence from individual agency (e.g. personal expression or experience), or according to social, institutional, or technological contexts (Warde, 2005), which contribute to practices’ evolution over time. Practices thus occupy a middle ground between individual agency and social structures (Shove et al., 2012).

The advantages of a practice-based approach for sustainable consumption research are welldocumented (Hargreaves, 2011; Røpke, 2009; Shove and Walker, 2010; Spaargaren, 2011; Warde, 2005). Whereas consumption and waste behaviors are conceptualized by rational actor models as the outcome of individual attitudes and decision-making, the practice-based approach situates consumption and waste as necessary components of practices that comprise everyday consumer life (Evans, 2012; Warde, 2005). Addressing unsustainable consumption patterns thus depends on understanding how practices emerge and evolve and how they are performed and coordinated in everyday life. An important influence on practices are systems of provision, which refer to infrastructures and systems that provide consumers with access to goods and services, thereby enabling them to perform practices. Examples include the electricity grid, transportation, and food retailers. Importantly, systems of provision can constrain or facilitate the ways that consumers can perform practices (Southerton et al., 2005), i.e. they can structure everyday life in specific and potentially unsustainable ways. For example, the emergence of supermarkets and one-stop-shopping convenience, refrigeration, and car-oriented urban development in the West combined to concentrate grocery shopping in fewer venues and occasions (e.g. from daily to weekly shopping) (Shove and Southerton, 2000; Watson, 2012).

ACCEPTED MANUSCRIPT

Efforts to influence consumer lifestyles must take into account practices and systems of provision in addition to more regularly mentioned factors such as consumer attitudes and choices.

1.3. Household food waste As has already been noted, a practice-based approach has been increasingly deployed in the study of household avoidable food waste (AFW)—the portion of organic waste that is, or was previously, edible. This theoretical approach highlights the challenge individuals face when having to coordinate and competently perform multiple practices within each day’s finite 24 hours—a challenge often magnified when practices require individuals to be in particular places and particular times, or challenge priorities (Evans, 2014; Ganglbauer et al., 2013; Watson and Meah, 2012). This may affect whether and how consumers adhere to practices known to mitigate food waste, such as meal planning, making grocery lists, and checking the fridge before shopping (Farr-Wharton et al., 2014), which may explain why households continue to waste food despite their best intentions not to (Evans, 2012; Stefan et al., 2013). A significant portion of household food waste also occurs when prepared food is left uneaten, either because there is too much of it or it does not suit household members’ tastes. Evans (2014) suggests that this happens due to the “good provider” mentality often adopted by individuals tasked with meal provisioning, which exemplifies how reducing food waste may conflict with the meanings embedded in meal provisioning practices.

Households’ performances of practices that mitigate food waste are also influenced by systems of provision that support meal provisioning practices—e.g. food retailers. Prior research has

ACCEPTED MANUSCRIPT

focused on how food retailers contribute to household food waste generation via marketing promotions that prompt unplanned purchases, thereby increasing rates of food spoilage at home (Mondéjar-Jiménez et al., 2016; Ponis et al., 2017), but the influence of food retailers may not end there. In addition to their employment of marketing tactics, retailers may offer food in different kinds of packaging while catering to different lifestyles, e.g. with one-stop-shopping or by offering convenience foods (Lee, 2016). These factors are especially relevant in light of research showing packaging’s contribution to food waste (Wikström et al., 2016; Williams et al., 2012) and the association between “convenience lifestyles” and food waste (Parizeau et al., 2015). In conjunction with interconnected systems of provision such as the transportation system, food retailers’ opening hours, amenities, and physical accessibility can influence when and how often consumers do their grocery shopping, as well as the transportation mode used (Lee, 2016; Lee and Soma, 2016; Sonesson et al., 2005). Tight consumer schedules and lengthy trips could contribute to over-purchasing by raising the opportunity cost of not buying everything needed, while car users might find it easier to make impulse buys because they do not have to carry their groceries onto a crowded bus. Lastly, research has shown how food retailing systems can be intertwined in different ways with factors such as cultural traditions and dietary preferences, thereby differentiating consumption practices and food waste outcomes (Leray et al., 2016). The above possibilities illustrate how systems of provision influence food-related practices, leading to food waste in the process.

Continued research in this direction is critical, not only because it provides further insight into household food waste behavior, but also due to the growing problem of urban food waste in developing nations (Adhikari et al., 2006). This is a problem entwined with urbanization, rising

ACCEPTED MANUSCRIPT

incomes, and the rapid modernization of food retail—the transition from a food retail sector dominated by traditional markets, street vendors, and family-run shops to one in which corporate-owned supermarkets and hypermarkets hold the majority market share (Lee and Soma, 2016; Reardon and Hopkins, 2006). As such, it is important to continue investigating the influence of food retailing on household food waste. To begin addressing this need, this study aims to quantitatively evaluate the influence that different aspects of food retailing have on household food waste generation in Seoul, with a focus on grocery trip characteristics, including shopping frequency, travel time, and transportation mode.

2.1. Research context The economic, cultural, and political capital of South Korea, Seoul has a population of approximately 10 million, or around one-fifth of the country's population, and is one of the densest cities in the world. It has been the primary locus of retail modernization since market liberalization in the mid-1990s (Coe and Lee, 2006). Since then, the South Korean food retail sector has been transformed by the rise of a handful of homegrown corporate retailers. Their success began with the development of hypermarkets,1 which were followed by smaller chain supermarkets, and lastly, expansion into the convenience store segment (Lee, 2016). In the process, traditional markets and family-owned grocery stores have experienced rapidly shrinking market shares (Kim and Hallsworth, 2013). Table 1 summarizes the major South Korean food retail formats and their relevant characteristics.

1

Also known as big box stores, superstores, supercenters, or large-format discount stores

ACCEPTED MANUSCRIPT

Table 1 Typology of South Korean food retailers 2 Relevant characteristics Retail Type Hypermarket

Description / floor area 3 >3,000 m2

Supermarket

330m2 – 3,000 m2

Traditional market

Open air or semiopen air cluster of individual stalls/stores

Family store

<600m2; independently owned

Department store supermarket

Located in the basement of highend department stores

Convenience store

<330m2

Transportation & accessibility - Multi-story car park - Usually located near public transit hub - Located in residential areas - No car park - Located in local commercial areas - Car park normally nonexistent; very small otherwise - Located in residential areas - No car park - Multi-story car park - Usually located near public transit hub - Open 24 hours - Ubiquitous throughout Seoul

Goods & services - Sells groceries, often in bulk, alongside large range of other goods - Offers leisure and dining amenities, e.g. food court

Market share by revenue (2015) 4 - 37%

- Smaller range of groceries and household goods

- 15%

- Offers mainly fresh and dried local food products - Grocery stalls mixed with small eateries and clothing/textile stalls

- 14%

- Smaller range of groceries and household goods

- 14%

- Caters to wealthier customers with mix of local and imported foods - Offers dining amenities, e.g. food court - Mainly convenience foods, limited fresh foods, and basic necessities

- n/a

- 18%

All information based on field observations except floor areas and market share data Hypermarket floor area defined by South Korean law (Distribution Development Industry Act, 2012); floor areas for other retail types from The People’s Institute of Economic and Social Studies (2009) 4 Market share data from Euromonitor (2016) 2 3

ACCEPTED MANUSCRIPT

South Korea’s volume-based food waste pricing system was introduced in 2013 as part of longterm zero-waste policies (Yoo and Yi, 2014); it accompanies mandatory waste segregation and volume-based pricing for general waste, introduced in 1991 and 1995, respectively (Kim and Kim, 2012). Under this system, households and small businesses are required to pay for food waste disposal; their district government determines the payment system and per-unit price. The three payment systems include volume-rate plastic bags, sticker/chip proof-of-payment, and radio frequency identification (RFID) machines. Volume-rate plastic bags are issued by district governments and sold at convenience stores at prices corresponding to their volume. Under the sticker/chip system, stickers or plastic chips must be purchased and attached to food waste bins as proof of payment. Households using these systems must dispose of food waste in volume-rate plastic bags or attach a valid sticker/chip to their food waste bin, otherwise the food waste will not be collected. RFID machines are generally installed in apartment complexes and require households to scan an RFID card before allowing them to deposit their food waste, which is then weighed. These households are then billed accordingly. A fourth group of households remain exempt from volume-based food waste disposal at the household level. Instead, they pay a monthly fee to their apartment complex’s management that is based on the complex’s total monthly food waste. This system is currently being phased out and replaced with RFID machines.

The volume-based pricing system is not without limitations. Research on volume-based pricing system for general waste found that raising disposal prices may prompt increased illegal disposal (Kim et al., 2008). This possibility cannot be ruled out for food waste, and households could avoid fees by concealing it in the general waste stream, flushing it down the toilet, or dumping it

ACCEPTED MANUSCRIPT

in public spaces. Additionally, reductions in household food waste associated with the pricing system may partially stem from households’ efforts to dry their food waste prior to disposal, rather than any lasting change in their consumption practices. These limitations reinforce the importance of better understanding household food consumption and waste practices in order to improve the effectiveness of waste reduction strategies.

2.2. Material and methods The majority of household food waste research employing a practice-based approach has been qualitative. Whereas such research is invaluable for explaining how practices contribute to food waste, it mostly cannot identify whether certain practices or their variations generate more food waste than another. Being able to quantify their relative contributions to unsustainable outcomes such as food waste is useful for developing policy strategy. Some studies have demonstrated this by employing quantitative methods with a practice-based approach (Southerton and Yates, 2014; Yates and Evans, 2016). For instance, analyzing survey data from 2,784 British households, Southerton and Yates (2014) established that surplus food is more than twice as likely to occur from meal occasions with others than from eating alone. In light of such utility, a quantitative survey-based methodology was chosen for this study with the aim of being able to quantify the effects of various food retail formats on household food waste.

The data were collected in June 2015 using an online survey of households sampled from a research firm’s proprietary online panel. As panel recruitment is conducted via online channels, panel members are likely to be internet-savvy, particularly among more elderly respondents. The target population under study was Seoul households, as represented by the household member

ACCEPTED MANUSCRIPT

primarily responsible for food provisioning. As such, the sampling frame used for recruitment consisted only of panel members who met the eligibility criteria of being 18 or over, a Seoul resident, and the primary grocery shopper in the household. These criteria were enforced again at the start of the survey using the relevant questions to screen out ineligible participants. Quotas were employed to ensure that a range of genders, ages and household sizes were represented in the sample. Survey participants received a small financial incentive in return for completing the survey.

The survey instrument was programmed using Qualtrics5 and included questions about household characteristics, diet and food consumption patterns, food waste-related attitudes, foodrelated competencies, grocery shopping practices, shopping trip characteristics, and food waste management practices. In accordance with the study’s practice-based approach, survey questions were phrased where relevant to assess how frequently participants performed food-related practices, as opposed to asking about their opinions or willingness to perform such practices. Survey participants were also asked how they paid for food waste and to estimate the amount of AFW they discarded. The survey questions were written in English and translated into Korean; the translations were then checked for cultural relevance and accuracy by a native Korean speaker. In order to test the survey questions, a pilot study was conducted in October, 2014. Separate from the survey, food waste disposal prices in each of Seoul’s 25 administrative districts were collected from local government websites for use in the survey data analysis.

5

Available at http://www.qualtrics.com

ACCEPTED MANUSCRIPT

A total of 3,833 individuals were invited to participate in the survey by email; 1,287 clicked on the invitation link, and 535 valid responses6 were recorded. A total of 111 participants were ejected from the survey for quality control reasons and 48 did not finish the survey within the allotted 72 hours. The remainder of the participants who clicked on the link but did not finish the survey were either not eligible to participate or were screened out due to full quotas. After the removal of outliers and responses that did not provide an estimate of their food waste, the total sample size was 460, representing a 12% completion rate.

The main dependent variable analyzed was the quantity of AFW that households said they discarded on average over the last three months, normalized to a per-week and per-household member basis. This variable was calculated by multiplying households’ estimates of their total food waste volume by their estimates of the percentage of their food waste that was edible. To reduce recall error, participants were asked to estimate their total food waste using South Korea’s volume-based waste pricing system as their frame of reference. Respondents who indicated using the volume-rate plastic bag or sticker/chip systems were asked to recollect the number and volume of bags or stickers/chips they paid for and used to dispose of food waste. The question also emphasized that participants should base their estimates as much as possible on the number of full plastic bags or containers of food waste discarded. Respondents who indicated paying for and discarding their food waste using the RFID system were asked to report the fees they paid per month for food waste disposal. Based on their reported fees, it was possible to calculate the amount of food waste discarded using their district’s disposal price, listed online. The last group paid a fixed fee to their apartment complex and so it was not possible to use a similar tactic.

6

This includes 22 participants who were recruited separately from the online panel.

ACCEPTED MANUSCRIPT

Nevertheless, to maintain consistency in how the question was asked, these households were asked to estimate the volume of the plastic bags they used to dispose of food waste and the number they used. In all cases, participants indicated the average quantity they discarded over the last three months on a per-month or per-week basis. This option was provided to avoid confusion for households with small quantities of food waste, who may make their discards less often than once a week.

The dependent variable exhibited extreme positive skew that violated normality assumptions and the assumption of homoscedasticity for ordinary least-squares regression. As a result, nonparametric tests that do not require the data to be normally-distributed were used to test for bivariate relationships in the data. For multivariate regression analysis, a generalized linear model using a Gamma distribution and log-link function was determined to be the most appropriate for the data. All analyses were performed using SPSS (IBM Corp., 2012).

All aspects of this study were approved by the Institutional Review Board at the University of California, Berkeley for compliance with ethical treatment of human research subjects.

3.1. Results This section begins with an overview of the sample’s demographic characteristics. It then examines participants’ grocery shopping patterns, using bivariate analyses to assess whether grocery trips to different retail formats (as listed in Table 1) can be distinguished by parameters such as trip time, transportation mode, and shopping frequency. Next, it examines the extent to which these factors are connected to participants’ reported over-purchasing behaviors and AFW

ACCEPTED MANUSCRIPT

patterns. Finally, the outcomes from a multivariate regression model that predicts households’ AFW, focusing on grocery trip characteristics while controlling for other influences, are presented.

3.1.1. Sample demographics To more accurately represent the Seoul population, the sample was weighted according to household size using Seoul census data (Seoul Metropolitan Government, 2014). Household size was chosen as the basis for weighting as the demographic variable most consistently identified as an important predictor of household food waste. The sample distribution according to household size, age, education, and monthly income is shown in Table 2 below. Weighting the sample corrected for the general underrepresentation of one and two-person households and reduced the overrepresentation of households with four and five or more people, but did not significantly alter other the other demographic characteristics of the sample.

ACCEPTED MANUSCRIPT

Table 2 Sample demographic characteristics Variable

Values

Household size

1 2 3 4 5+ 18-30 31-40 41-50 51-60 60+ Male Female High school or lower Bachelor's degree Graduate degree < 1,000,000 won 1,000,000 - 1,999,999 won 2,000,000 - 2,999,999 won 3,000,000 - 3,999,999 won 4,000,000 - 4,999,999 won > 5,000,000 won

Age

Gender Education

Monthly income

Weighted % 23.7 23.3 23.6 22.2 7.2 26.8 26.6 25.1 15.3 6.2 37.8 62.2 14.6 74.4 10.7 3.2 9.6 17.0 18.2 20.2 30.4

Unweighted % 16.9 19.7 22.8 25.2 15.4 26.8 25.7 25.7 16.0 5.9 38.6 61.4 14.7 73.9 11.2 2.6 7.7 15.6 17.5 20.8 34.4

3.1.2. Retail formats and grocery trip characteristics The three most visited retail formats in the sample were hypermarkets, supermarkets, and traditional markets; 83%, 60%, and 52% of participants indicated that they had bought groceries at these retail formats in the last three months, respectively. Participants were also asked how often they bought different food types from different retail formats; the decision was made to focus on fresh fruit and vegetable (FFV) buying frequencies for two reasons. First, this food group is inherently fragile and highly perishable, making it especially vulnerable to spoilage from over-purchasing. Second, the traditionally vegetable-heavy Korean diet largely persisted among South Koreans as the country industrialized (Lee et al., 2002); even when younger generations adopt new diets, they maintain high levels of fruit intake (Lim et al., 2014). The

ACCEPTED MANUSCRIPT

popularity of retail formats for buying FFV was determined based on where participants said they bought FFV most frequently. The majority of participants’ (37%) described FFV buying practices that involved shopping equally frequently at two or more retail formats, which is consistent with prior research showing how consumers often patronize multiple retail formats when buying the same product (Uncles et al., 1995). Otherwise, hypermarkets were the most popular (24%), followed by supermarkets and traditional markets (tied at 13%), and family stores (11%). For this reason, the remainder of the analyses will focus primarily on these four major retail formats.

Across all retail formats, households’ FFV buying practices most commonly involved buying FFV once a week (37%) and 2-3 times a month (27%). FFV buying frequency was a distinguishing characteristic of several retail formats, as can observed in Figure 1, as well as from statistical comparisons. Participants who bought FFV most often at hypermarkets tended to do so less frequently than those who bought FFV most often at family stores (Mann-Whitney test: p=0.001), and traditional markets (Mann-Whitney test: p=0.006).

ACCEPTED MANUSCRIPT

Percentage of sample who bought FFV at each format

35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 Hypermarket

Supermarket

Traditional market

None

Less than once a month

Once a month

Once a week

2-3 times a week

Every day

Family store 2-3 times a month

Figure 1 Sample distribution according to FFV buying frequency and retail format

Seoul’s high-density environment was reflected by how the samples’ grocery shopping practices involved very short travel times. Even excluding near-ubiquitous convenience stores, 55% of participants were less than 10 minutes away from their food retailer of choice, and 28% travelled 10-15 minutes. Travel times to hypermarkets were longer than for all other retail formats (Wilcoxon signed-rank test: p=0.000 for all tests), while traditional markets necessitated longer travel times than only supermarkets (Wilcoxon signed-rank test: p=0.000) and family markets (Wilcoxon signed-rank test: p=0.000). Additionally, grocery shopping practices of participants who lived within 10 minutes from hypermarkets involved more frequent FFV purchases than those who lived further away (Kruskal-Wallis test: p=0.002). However, this relationship was not true for any of the other major retail formats (Kruskal-Wallis test: p=0.000).

ACCEPTED MANUSCRIPT

Unsurprisingly, households’ grocery shopping practices at the four major retail formats could be associated with different transportation modes (Figure 2). Grocery shopping at hypermarkets most commonly involved driving, whereas walking was most often linked with shopping at supermarkets, traditional markets, and family stores. In general, transportation mode was also correlated with travel times, with walking trips being significantly shorter than trips made by car or public transit for all four major retail formats (Kruskal-Wallis test: p≤0.002 for all tests). These findings showing how variations in grocery shopping practices’ transportation mode and travel time were linked with different retail formats were mostly consistent with prior knowledge about their land use characteristics.

1.0 0.9 Percentage of shoppers

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Hypermarkets

Supermarkets Walk

Car

Traditional markets

Public transit

Family stores

Other

Figure 2 Sample distribution by transportation mode and retail format

Participants were also asked about the average amount they spent on food per trip to each of the retail formats they indicated visiting in the last three months as a proxy for the quantity of

ACCEPTED MANUSCRIPT

groceries purchased each trip; average per-trip spending, normalized for household size is shown in Figure 3. Per-trip, per-person spending was positively correlated with travel times for all four major retail formats (Jonckheere-Terpstra test: p≤0.057 for all tests). Similarly, transportation mode could also be linked to spending for all four major retail formats, as drivers reported greater per-trip, per-person spending than non-drivers (Jonckheere-Terpstra test: p≤0.058 for all tests). Unexpectedly however, per-trip, per-person spending was positively correlated with FFV buying frequency for hypermarkets, supermarkets, and family stores (Jonckheere-Terpstra test: p≤0.052 for all tests), but had no significant relationship for traditional markets.

$39

$22 $18 $11

Hypermarket

Supermarket

Traditional Market

Family Store

Figure 3 Average per-trip, per-person spending by retail format, normalized for household size (converted to US$)

3.1.3. Over-purchasing 75% of participants indicated that they had bought more food than they intended (i.e. overpurchased) during the last three months. When asked about the reasons for having done so, “wanting to take advantage of a promotion” scored highest in the sample, followed by “portion/packaging larger than needed,” with 78% and 74% of participants indicating that these reasons applied more than half the time.

ACCEPTED MANUSCRIPT

Both over-purchasing and the reasons for doing had statistical relationships to shopping at particular retail formats. When comparing households’ reported over-purchasing at each of the retail formats they visited in the last three months, over-purchasing was reported more frequently at hypermarkets than at any other format (Wilcoxon signed-rank test, p<0.001 for all tests). Over-purchasing at hypermarkets was positively correlated with reasons including “portion/packaging larger than needed” and “wanted to take advantage of promotion” (Jonckheere Terpstra test: p=0.000 for both). This was noteworthy because neither of these reasons were correlated with over-purchasing at family stores, supermarkets, and traditional markets, whereas reasons such as “looked/smelled really tasty,” “wanted to try something new,” and “forgot that I actually needed it” were correlated with over-purchasing at all four of these formats (Jonckheere Terpstra test: p≤0.023 for all tests).

Reported over-purchasing was also positively linked to per-trip spending; this was true for supermarkets, family stores, and traditional markets (Jonckheere Terpstra test: p≤0.001 for all tests), but not for hypermarkets. The same pattern held when testing the relationship between over-purchasing and travel times, with positive correlations identified for the same three retail formats (Jonckheere Terpstra test: p≤0.039 for all tests), but not for hypermarkets. Lastly, overpurchasing was only connected to transportation mode for family stores—households that walked to this retail format reported less over-purchasing than those who did not walk (KruskalWallis test: p=0.039). Finally, and contrary to expectations, bivariate testing found positive correlations between over-purchasing and FFV buying frequencies for supermarkets, traditional

ACCEPTED MANUSCRIPT

markets, and family stores (Jonckheere Terpstra test: p≤0.004 for all tests), but not for hypermarkets.

3.1.4. Causes and disposal of avoidable food waste The average reported volume of AFW per person, per week was 2.4 liters, with a standard deviation of 8.3 liters, reflecting the very wide range of values reported. Within the sample, volume-rate plastic bags were by far the most frequently used system to pay for food waste disposal, with 72% of the total. 16% reported paying a flat rate to their apartment complex management for unlimited disposal, 3% used proof-of-payment stickers/chips, and 9% paid using the RFID system. There were no statistical relationships detected between the use of different payment systems and AFW.

Interestingly, there were no direct statistical relationships found between AFW and reported over-purchasing at any of the retail formats. This could reflect the fact over-purchasing may not always cause more food waste if households adapt accordingly, e.g. by buying less of other items or making extra efforts to use up over-purchased food. Additionally, unused ingredients are just one portion of a households’ waste stream, in addition to leftovers and partially used ingredients, which may be less affected by over-purchasing. However, over-purchasing at all retail formats except traditional markets and family stores was positively correlated with how often participants threw away unused ingredients (Jonckheere Terpstra test: p≤0.043 for all tests). In turn, how often participants discarded unused ingredients was positively correlated with AFW (Jonckheere Terpstra test: p=0.000), providing evidence to connect over-purchasing with food waste.

ACCEPTED MANUSCRIPT

When asked about the reasons for food waste, discards related to spoilage, food safety, and the sell-by date were more prevalent than discards for other reasons (Figure 4). When asked why they thought food waste had been allowed to spoil, pass its sell-by date, or reach a state causing food safety concerns, participants scored “forgot about the food” and “didn’t have time to eat/cook it” most highly. Both of these reasons were correlated with over-purchasing at hypermarkets (Jonckheere Terpstra test: p≤0.058 for both reasons) and supermarkets (Jonckheere Terpstra test: p≤0.043 for both reasons), but not for traditional markets and family stores.

32%

35%

33%

18% 15%

14%

Cooked too much Bought too much Didn't taste good Suspicious about Clearly unsafe to Past sell-by date food safety eat Figure 4 Reasons for food waste - percent of participants citing reason more than half the time

3.1.5. Multivariate regression results This section presents the results from a generalized linear model (gamma distribution, log-link function) that was developed to predict the volume of AFW reported by survey participants, with a focus on the characteristics of trips households made as part of their grocery shopping practices. Other independent variables tested during model development represented the influences of household demographics, attitudes towards food waste, food management practices and competencies, and diet. Three models are presented for comparison in Table 3; Model A

ACCEPTED MANUSCRIPT

includes only demographics and food-related practices and attitudes, Model B introduces variables related to food waste pricing, and Model C introduces variables representing grocery trip characteristics. Ordinal and continuous independent variables were standardized to allow for easier comparison of regression coefficients. The standardized coefficients can be interpreted by calculating their exponents; these represent the multiplicative change in the dependent variable given a one standard deviation change in the independent variable. Table 3 Results from generalized linear models (gamma distribution, log-link) estimating AFW per person per week Model A Variable

Coeff. Female a

Demographic

Age range Household size

Food-related practices & attitudes

Food waste pricing

Meal planning practice freq. Leftover food consumption Fresh fruit consumption Agrees that wasting food is immoral Confident can cook tasty food Pays flat fee to apartment a Per liter price

Model B

Wald χ2

Coeff.

Model C

Wald χ2

2.125

-.195

1.476

-.139

.773

.289

12.585***

.290

12.968***

.174

4.358*

-.390

14.456***

-.339

10.331**

-.296

8.167**

.273

12.063***

.278

12.012***

.256

10.529**

-.157

4.201*

-.171

5.016*

-.198

7.026**

.340

20.785***

.361

23.497***

.344

19.805***

-.352

15.945***

-.330

13.920***

-.282

10.413**

-.244

9.622**

-.232

8.843**

-.198

6.564**

-.447

4.900*

-.547

7.246**

-.143

3.398

-.219

7.857**

.042

.246

-.325

16.219***

.206

5.928**

Summary statistics

a

FFV buying freq. Min. travel time to food retailer b

No. of observations Akaike Information Criterion (AIC) Incremental Δ AIC Log-likelihood Likelihood ratio Chi-square

Wald χ2

-.235

FFV buying freq. Grocery trip characteristics

Coeff.

380

380

380

871.368

868.103

852.671

-3.265

-15.432

-425.128

-422.052

-411.336

105.073***

112.337***

133.769***

Dummy variable, unstandardized; b Quadratic term. * p < 0.05;

**

p < 0.01; *** p < 0.001.

ACCEPTED MANUSCRIPT

The models produced evidence for a statistical relationship between AFW and two grocery shopping characteristics: FFV buying frequency and travel time. The former exhibited a quadratic relationship with AFW in which AFW peaked at buying frequencies 2-3 times a month. For perspective, Model C indicated that, all else being equal, households who bought FFV daily generated 54% less AFW as households who bought FFV 2-3 times a week, 69% less than households who bought FFV once a week, and 70% less than households who bought FFV 2-3 times a month. On the other hand, households who bought FFV 2-3 times a month generated approximately 42% more AFW than households who bought FFV once a month, and nearly three times as much AFW as households who bought FFV less than once a month. AFW was also positively correlated with households’ travel times from home to their food retailer, such that a one-category increase in travel time was accompanied by a 26% increase in AFW.

A variety of other variables were also found to have statistically significant influences on AFW. Among demographic variables, gender was irrelevant, but larger households were negatively associated with AFW and those represented by older individuals were associated with more. Households where meal planning practices and fresh fruit consumption were more frequent generated more AFW, while those households whose members widely consumed leftover food, agreed more strongly with the statement that “wasting food is immoral,” and had confidence in their ability to cook tasty meals were linked with smaller volumes of AFW. Finally, the volumebased pricing system was found to have an influence on AFW. Higher per-liter prices for food waste disposal were correlated with less AFW, and households who only paid a flat fee for food waste disposal reported less AFW than those who paid per unit.

ACCEPTED MANUSCRIPT

Independent variables tested in the regression models that were not found to be statistically significant included education, income, and household composition; attitudes towards food waste (environmental impact, wasted money, social norms); frequency of performing food-related practices (checking inventory, making written and/or mental lists, consciously avoiding impulse and/or large purchases); food-related social interactions (entertaining guests, packing food for others, receiving food from others); storage conditions and fridge size; total meals cooked at home per week; average amount spent per shopping trip; and transportation mode.

For summary statistics, all three likelihood-ratio chi-square values were significant at the 0.001 level, indicating that all three models were statistically significant. However, model fit was best in Model C, as can be seen from comparing Akaike’s Information Criterion (AIC), which measures goodness-of-fit while penalizing additional independent variables.

3.2. Discussion Between the bivariate analyses and the regression results, there was evidence that shopping at different food retail formats is related to household food waste, even when controlling for relevant demographic, attitudinal, and behavioral variables. The most interesting individual regression result was the squared relationship between FFV buying frequency and AFW. This pattern may distinguish between households who eat FFV regularly and those who do not— households who bought FFV once a week or more reported more frequent consumption of leafy vegetables, root vegetables, and fresh fruit than those who did not (Kruskal-Wallis test: p≤0.001 for all three food types). For the latter group, it is plausible that their total FFV purchase volume in a given time period increases with FFV buying frequency, thereby increasing FFV inventory

ACCEPTED MANUSCRIPT

and spoilage rates accordingly. On the other hand, it seemed likely that because the former group reported eating FFV regularly, they bought similar amounts of FFV regardless of how often they did so, but instead their per-trip purchase volumes decreased as buying frequencies increased. Assuming that smaller per-trip purchase volumes result in lower spoilage rates due to the greater ease of managing smaller inventories, this would explain the relationship observed in the model.

The regression results highlighting buying frequency and travel time as drivers of AFW were noteworthy in light of the bivariate analyses that also connected these factors to specific retail formats, households’ reported over-purchasing, and per-trip, per-person spending. Hypermarkets were implicated as being connected to higher AFW via two effects on grocery shopping practices: 1) lower modal FFV buying frequency and longer travel times, and 2) they were most frequently linked to over-purchasing as a result of bulk packaging and marketing promotions. However, no bivariate relationship was found between travel time to hypermarkets and overpurchasing, and so it was difficult to say for certain whether the first effect was actually a cause of over-purchasing or a spurious correlation involving the second effect.

It was also interesting that travel time was significantly correlated with AFW in the model even when controlling for buying frequency, a finding consistent with how travel time was not related to buying frequency for supermarkets, traditional markets, and family stores, but was correlated with over-purchasing for these formats. As such, the relationship between travel time, overpurchasing, and AFW for these three retail formats may be driven by a dynamic other than buying frequency. Transportation mode was ruled out here; although car users can carry larger quantities home than others, their travel times were statistically indistinguishable from public

ACCEPTED MANUSCRIPT

transit users. There was also no statistical evidence to distinguish car users from others in terms of over-purchasing or AFW. Instead, the answer may be related to the opportunity cost of not buying everything needed, which would increase with travel time—in other words, even if households facing longer travel times shop often, e.g. every day, they may still risk having to make a second trip if they do not buy everything needed. This would be particularly true if frequent grocery shopping is dictated more by the need for immediate use and freshness rather than time constraints or the convenience of buying everything at once in fewer trips, as might be the case for FFV. This possibility merits further exploration, since it implies that travel time may mediate the relationship between purchase planning practices and household food waste.

A finding related to this possibility was that among practices related to grocery shopping and food management, only meal planning was statistically significant in the regression results. Even then, its positive coefficient was counterintuitive and contrasted with findings from prior research about the importance of meal planning routines for mitigating food waste (Ponis et al., 2017; Stefan et al., 2013). However, consumer advice in this regard is often accompanied by recommendations that meal planning be undertaken in conjunction with other practices, such as using up leftover ingredients and leaving some meals unplanned. As such, one explanation for the counterintuitive result is that households in the survey may have attempted meal planning without performing the necessary accompanying practices, rendering it a counterproductive means of reducing food waste. It was unclear why other practices, such as making lists and checking the fridge, were insignificant in the model, although there were negative, but not quite statistically significant, bivariate relationships between AFW and households who reported making mental lists all the time (Kruskal-Wallis test: p=0.062) or checking the fridge all the time

ACCEPTED MANUSCRIPT

(Kruskal-Wallis test: p=0.092) before going grocery shopping. This finding contributes further to the possibility that grocery trip characteristics may somehow mediate the effect of such practices on food waste.

It is important to note the exclusion of households’ reported over-purchasing and per-trip, perperson spending from the model due to statistical insignificance. This was expected given the bivariate relationships uncovered between these factors and grocery trip characteristics. Outside of the regression analysis, it was striking that over-purchasing at hypermarkets and supermarkets, but not traditional markets or family stores, was correlated with discards of unused ingredients and reasons for discard including “forgot about the food” and “didn’t have time to cook/eat it.” This finding indicates that the transformation of excess purchases into food waste may be related to where the food was purchased—a plausible notion given the greater prevalence of sell-by dates on food sold in corporate-owned supermarkets and hypermarkets in comparison to venues such as traditional markets. It may also reflect lifestyle differences between shoppers at different retail formats, particularly in relation to time-scarcity.

With regards to volume-based waste pricing, the negative coefficient for the per-liter food waste disposal price reflected the effectiveness of the volume-based waste pricing system for reducing AFW, underscoring the importance of the “polluter pays” principle for managing environmental issues. Even so, the findings could not rule out the possibility that reductions in food waste are associated with illegal dumping, as described in section 1 Next, it was surprising, at first glance, that households who paid a flat monthly fee for food waste disposal were associated with less food waste, since they are not directly charged for their food waste. The explanation involves the

ACCEPTED MANUSCRIPT

survey’s use of the volume-based waste pricing system as a frame of reference when asking participants to estimate food waste quantities. Participants paying a flat fee were not required to prepay for the food waste they throw away and as such, would not have been as conscious of their food waste volumes as other households. Given that people tend to underestimate their food waste (Quested et al., 2011), households using the apartment flat fee system may have reported systematically less food waste than other households—a bias reflected in the regression results.

The remaining regression results were intuitive and variously reinforced the relevance of diet, food management patterns, and particular attitudes for reducing food waste. First, the negative coefficient for households’ willingness to eat leftover food indicated that discards of leftover food may be an important driver of AFW in South Korean households. The positive correlation between fresh fruit consumption and AFW was probably attributable to the increased fragility and perishability of fresh fruit, which would contribute to AFW via increased spoilage rates. The negative relationship between participants’ stated confidence in their ability to cook tasty foods and AFW illustrated the value of consumer education that facilitates the preparation of tasty meals for reducing household food waste, but it was also possible that this variable indirectly measured broader experience with food management that helps to mitigate food waste, e.g. knowledge of how to store food properly or use up leftover ingredients. Lastly, the finding connecting the perception of food waste as immoral with less AFW was intriguing given contrasting theories about the effectiveness of appealing to moral values for promoting waste reduction (Gjerris and Gaiani, 2013; Stern, 2000). This finding may reflect cultural norms about wasting food, as well as the lasting effects of extensive public education campaigns that have

ACCEPTED MANUSCRIPT

accompanied the introduction of mandatory waste segregation and volume-based pricing policies in South Korea.

4.1. Conclusion The findings presented several potential implications for policymaking and future research. It was clear that food retailers’ influence on food waste and over-purchasing behavior may not be restricted to marketing promotions. Instead, there is evidence to suggest that travel time and buying frequency are potential determinants of over-purchasing behavior as well, and that their effects may be varied depending on the retail format visited. Consequently, while existing efforts that teach households to better manage food at home and avoid over-purchasing are necessary and could be complemented with regulation of retailers’ marketing tactics, policymaking efforts to reduce food waste should also look beyond individual decision-making and consider whether patterns of food retail development are shaping consumer lifestyles in ways that compound the issue of food waste. This may be particularly pertinent in developing nations, where planners and policymakers may still have the opportunity to avoid development patterns that lock consumers into less sustainable lifestyles. For example, ensuring the continued presence of small food retailers in dense urban areas may help slow predicted growth in household food waste by preserving high-levels of physical access to food retail.

That is not to imply that land use and retail development patterns are the sole drivers of how, when, and how often households buy groceries; other structural factors, such as culture or technology, may interact with systems of provision to influence individual decision-making and lifestyles, with different outcomes for food waste. For instance, this paper did not examine the

ACCEPTED MANUSCRIPT

effects of internet grocery shopping, which is growing in South Korea due to the promise and convenience of free same-day delivery provided that consumers spend a minimum amount. Given this study’s counterintuitive findings about the effectiveness of pre-shopping planning behavior for reducing food waste, further research that investigates adherence to such plans in relation to grocery trip characteristics would be interesting. Further research should also examine how purchasing patterns for food groups other than FFV are related to household food waste.

This study had several limitations. First, the survey relied on participants to estimate their food waste—both its volume and the percentage that is avoidable. This may be a source of error because people often lack awareness about the food they discard (Lazell, 2016), possibly preventing them from accurately estimating how much food waste they generate. However, this study attempted to overcome this challenge by having participants use the volume-based food waste pricing system as a frame of reference to estimate their food waste. By requiring households to actively separate and pay for their food waste, the waste pricing system raises South Korean householders’ consciousness of food waste, potentially improving their ability to more accurately quantify their food waste. Next, the 12% completion rate indicates that nonresponse bias may have affected representativeness. The use of quotas reduced non-response bias with respect to demographic variables, but non-participants may still differ from participants in their grocery shopping practices. For example, participants who completed the survey before quotas filled could be less busy than non-participants, which may influence their grocery shopping and food-related practices. Similarly, participants already interested in food waste may have been more motivated to finish the survey than others. Lastly, this study’s findings may be limited in applicability to other geographic contexts, not only because it specifically addressed

ACCEPTED MANUSCRIPT

the South Korean food retail and waste management systems, but because of the socially and culturally-specific nature of food consumption and waste. On the other hand, this limitation also underscores the need for household food waste research in more diverse geographic contexts.

Acknowledgments The author would like to acknowledge Jennifer Wolch for her feedback on the research design and analysis and Jiyoung Lee for her assistance with the Korean language.

Funding: This work was partially supported by the Korea Foundation.

References Adhikari, B.K., Barrington, S., Martinez, J., 2006. Predicted growth of world urban food waste and methane production. Waste Manag. Res. 24, 421–433. doi:10.1177/0734242X06067767 Coe, N.M., Lee, Y.-S., 2006. The Strategic Localization of Transnational Retailers: The Case of Samsung-Tesco in South Korea. Econ. Geogr. 82, 61–88. doi:10.1111/j.19448287.2006.tb00288.x Distribution Development Industry Act, 2012. Euromonitor, 2016. Grocery Retailers in South Korea, Passport. Euromonitor International. Evans, D., 2014. Food Waste: Home Consumption, Material Culture and Everyday Life. Bloomsbury, London. Evans, D., 2012. Beyond the Throwaway Society: Ordinary Domestic Practice and a Sociological Approach to Household Food Waste. Sociology 46, 41–56. doi:10.1177/0038038511416150 Farr-Wharton, G., Foth, M., Choi, J.H., 2014. Identifying factors that promote consumer behaviours causing expired domestic food waste. J. Consum. Behav.

ACCEPTED MANUSCRIPT

Ganglbauer, E., Fitzpatrick, G., Comber, R., 2013. Negotiating food waste: Using a practice lens to inform design. ACM Trans Comput-Hum Interact 20, 11:1–11:25. doi:10.1145/2463579.2463582 Gjerris, M., Gaiani, S., 2013. Household food waste in Nordic countries: Estimations and ethical implications. Nord. J. Appl. Ethics 7, 6–23. Graham-Rowe, E., Jessop, D.C., Sparks, P., 2014. Identifying motivations and barriers to minimising household food waste. Resour. Conserv. Recycl. 84, 15–23. doi:10.1016/j.resconrec.2013.12.005 Gustavsson, J., Cederberg, C., Sonesson, U., Van Otterdijk, R., Meybeck, A., 2011. Global food losses and food waste: extent, causes and prevention. United Nations Food and Agriculture Organization, Rome. Hargreaves, T., 2011. Practice-ing behaviour change: Applying social practice theory to proenvironmental behaviour change. J. Consum. Cult. 11, 79–99. doi:10.1177/1469540510390500 IBM Corp., 2012. IBM SPSS Statistics for Windows. IBM Corp., Armonk, NY. Kim, G.-S., Young-Jae Chang, Kelleher, D., 2008. Unit pricing of municipal solid waste and illegal dumping: an empirical analysis of Korean experience. Environ. Econ. Policy Stud. 9, 167–176. doi:10.1007/s10018-008-0144-3 Kim, K.-Y., Kim, Y.J., 2012. 2011 Modularization of Korea’s Development Experience: Volume-based Waste Fee System in Korea. Korea Environmental Institute, Seoul. Kim, W., Hallsworth, A.G., 2013. Large format stores and the introduction of new regulatory controls in South Korea. Int. Rev. Retail Distrib. Consum. Res. 23, 152–173. doi:10.1080/09593969.2012.754779 Lazell, J., 2016. Consumer food waste behaviour in universities: Sharing as a means of prevention. J. Consum. Behav. 15, 430–439. doi:10.1002/cb.1581 Lee, K., 2016. Convenient food, inconvenient waste: Systems of provision meet social practices in Seoul, in: Sahakian, M., Saloma, C., Erkman, S. (Eds.), Food Consumption in the City: Practices and Patterns in Urban Asia and the Pacific. Routledge, Milton Park, Abingdon, Oxon ; New York, NY, pp. 216–237. Lee, K.C.L., Soma, T., 2016. From “Farm to Table” to “Farm to Dump”: Emerging Research on Urban Household Food Waste in the Global South, in: Anderson, C., Brady, J., Levkoe, C.Z. (Eds.), Conversations in Food Studies. University of Manitoba Press.

ACCEPTED MANUSCRIPT

Lee, M.-J., Popkin, B.M., Kim, S., 2002. The unique aspects of the nutrition transition in South Korea: the retention of healthful elements in their traditional diet. Public Health Nutr. 5, 197–203. doi:10.1079/PHN2001294 Leray, L., Sahakian, M., Erkman, S., 2016. Understanding household food metabolism: relating micro-level material flow analysis to consumption practices. J. Clean. Prod. 125, 44–55. doi:10.1016/j.jclepro.2016.03.055 Lim, H., Kim, S.Y., Wang, Y., Lee, S.J., Oh, K., Sohn, C.Y., Moon, Y.M., Jee, S.H., 2014. Preservation of a traditional Korean dietary pattern and emergence of a fruit and dairy dietary pattern among adults in South Korea: secular transitions in dietary patterns of a prospective study from 1998 to 2010. Nutr. Res. 34, 760–770. Lundqvist, J., de Fraiture, C., Molden, D., 2008. Saving water: from field to fork: curbing losses and wastage in the food chain. Mondéjar-Jiménez, J.-A., Ferrari, G., Secondi, L., Principato, L., 2016. From the table to waste: An exploratory study on behaviour towards food waste of Spanish and Italian youths. J. Clean. Prod., Research on sustainable cleaner production and sustainable energy options 138, Part 1, 8–18. doi:10.1016/j.jclepro.2016.06.018 Parfitt, J., Barthel, M., Macnaughton, S., 2010. Food Waste Within Food Supply Chains: Quantification and Potential for Change to 2050. Philos. Trans. R. Soc. B Biol. Sci. 365, 3065–3081. doi:10.1098/rstb.2010.0126 Parizeau, K., von Massow, M., Martin, R., 2015. Household-level dynamics of food waste production and related beliefs, attitudes, and behaviours in Guelph, Ontario. Waste Manag. 35, 207–217. doi:10.1016/j.wasman.2014.09.019 Ponis, S.T., Papanikolaou, P.-A., Katimertzoglou, P., Ntalla, A.C., Xenos, K.I., 2017. Household Food Waste in Greece: A Questionnaire Survey. J. Clean. Prod. doi:10.1016/j.jclepro.2017.02.165 Quested, T.E., Parry, A.D., Easteal, S., Swannell, R., 2011. Food and drink waste from households in the UK. Nutr. Bull. 36, 460–467. doi:10.1111/j.1467-3010.2011.01924.x Reardon, T., Hopkins, R., 2006. The Supermarket Revolution in Developing Countries: Policies to Address Emerging Tensions Among Supermarkets, Suppliers and Traditional Retailers. Eur. J. Dev. Res. 18, 522–545. doi:10.1080/09578810601070613 Reckwitz, A., 2002. Toward a Theory of Social Practices A Development in Culturalist Theorizing. Eur. J. Soc. Theory 5, 243–263. doi:10.1177/13684310222225432 Røpke, I., 2009. Theories of practice—New inspiration for ecological economic studies on consumption. Ecol. Econ. 68, 2490–2497.

ACCEPTED MANUSCRIPT

Schatzki, T.R., 1996. Social Practices: A Wittgensteinian Approach to Human Activity and the Social, 1 edition. ed. Cambridge University Press, New York. Seoul Metropolitan Government, 2014. Seoul Census Data. Shove, E., Pantzar, M., Watson, M., 2012. The Dynamics of Social Practice: Everyday Life and how it Changes. SAGE Publications Ltd, London. Shove, E., Southerton, D., 2000. Defrosting the Freezer: From Novelty to Convenience A Narrative of Normalization. J. Mater. Cult. 5, 301–319. doi:10.1177/135918350000500303 Shove, E., Walker, G., 2010. Governing transitions in the sustainability of everyday life. Res. Policy 39, 471–476. doi:10.1016/j.respol.2010.01.019 Sonesson, U., Anteson, F., Davis, J., Sjödén, P.-O., 2005. Home Transport and Wastage: Environmentally Relevant Household Activities in the Life Cycle of Food. AMBIO J. Hum. Environ. 34, 371–375. doi:10.1579/0044-7447-34.4.371 Southerton, D., Warde, A., Hand, M., 2005. The limited autonomy of the consumer: implications for sustainable consumption, in: Southerton, D., Chappells, H., Vliet, B.V. (Eds.), Sustainable Consumption: The Implications Of Changing Infrastructures Of Provision. Edward Elgar, Cheltenham, pp. 32–48. Southerton, D., Yates, L., 2014. Exploring Food Waste through the lens of Social Practice Theories: Some reflections on eating as a compound practice, in: Ekström, K.M. (Ed.), Waste Management and Sustainable Consumption: Reflections on Consumer Waste. Routledge, pp. 133–149. Spaargaren, G., 2011. Theories of practices: Agency, technology, and culture: Exploring the relevance of practice theories for the governance of sustainable consumption practices in the new world-order. Glob. Environ. Change 21, 813–822. Stefan, V., van Herpen, E., Tudoran, A.A., Lähteenmäki, L., 2013. Avoiding food waste by Romanian consumers: The importance of planning and shopping routines. Food Qual. Prefer. 28, 375–381. doi:10.1016/j.foodqual.2012.11.001 Stern, P.C., 2000. New Environmental Theories: Toward a Coherent Theory of Environmentally Significant Behavior. J. Soc. Issues 56, 407–424. doi:10.1111/0022-4537.00175 The People’s Institute of Economical & Social Studies, 2009. SSM gwa daehyeongmateuui paguiryeok, geurigo yutongeob sangsaengbangan [The destructive power of SSMs and superstores, and how retailers can co-exist]. The People’s Institute of Economical & Social Studies, Seoul.

ACCEPTED MANUSCRIPT

Uncles, M., Ehrenberg, A., Hammond, K., 1995. Patterns of Buyer Behavior: Regularities, Models, and Extensions. Mark. Sci. 14, G71–G78. doi:10.2307/184149 Warde, A., 2005. Consumption and Theories of Practice. J. Consum. Cult. 5, 131–153. doi:10.1177/1469540505053090 Watson, M., 2012. How theories of practice can inform transition to a decarbonised transport system. J. Transp. Geogr. 24, 488–496. doi:10.1016/j.jtrangeo.2012.04.002 Watson, M., Meah, A., 2012. Food, waste and safety: negotiating conflicting social anxieties into the practices of domestic provisioning. Sociol. Rev. 60, 102–120. doi:10.1111/1467954X.12040 Wikström, F., Williams, H., Venkatesh, G., 2016. The influence of packaging attributes on recycling and food waste behaviour – An environmental comparison of two packaging alternatives. J. Clean. Prod. 137, 895–902. doi:10.1016/j.jclepro.2016.07.097 Williams, H., Wikström, F., Otterbring, T., Löfgren, M., Gustafsson, A., 2012. Reasons for household food waste with special attention to packaging. J. Clean. Prod. 24, 141–148. doi:10.1016/j.jclepro.2011.11.044 Yates, L., Evans, D., 2016. Dirtying Linen: Re-evaluating the sustainability of domestic laundry. Environ. Policy Gov. 26, 101–115. doi:10.1002/eet.1704 Yoo, K.-Y., Yi, S., 2014. Evaluation and development of solid waste management plan: a case of Seoul for past and future 10 years. J. Mater. Cycles Waste Manag. 1–17. doi:10.1007/s10163-014-0294-2