Wasting Food, Wasting Resources: Potential Environmental Savings Through Food Waste Reductions

Food is needed to maintain our physical integrity and therefore meets a most basic human need. The food sector got in the focus of environmental policy, because of its environmental implications and its inefficiency in terms of the amount of food lost along the value chain. The European Commission (EC) flagged the food waste issue a few years ago and adopted since then a series of policies that partially address the problem. Among these, the Resource Efficiency Roadmap set the aspirational goal of reducing the resource inputs in the food chain by 20% and halving the disposal of edible food waste by 2020. Focusing on consumer food waste, we tested what a reduction following the Roadmap's food waste target would imply for four environmental categories in EU28 (European Union 28 Member States): greenhouse gas emissions, land use, blue water consumption, and material use. Compared to the 2011 levels, reaching the target would lead to 2% to 7% reductions of the total footprint depending on the environmental category. This equals a 10% to 11% decrease in inputs in the food value chain (i.e., around half of the resource use reductions targeted). The vast majority of potential gains are related to households, rather than the food‐related services. Most likely, the 2020 target will not be met, since there is insufficient action both at Member State and European levels. The Sustainable Development Goals provide a new milestone for reducing edible food waste, but Europe needs to rise up to the challenge of decreasing its per capita food waste generation by 50% by 2030.


Introduction
Food is a key satisfier of one of the most basic human needs, that of subsistence (Max-Neef 1992). Nevertheless, the degree of fulfillment of this need is very uneven around the world. While 805 million people-11% of the world populationwere undernourished in 2014 (FAO 2014), one third of food is wasted globally ). This has relevant economic, ethical, social, and environmental implications.
pressures on the environment (Tukker and Jansen 2006), while at the same time large amounts of food are thrown away (Monier et al. 2010;. The Resource Efficiency Roadmap adopted by the European Commission (EC) in 2011 (EC 2011) put the problem of food waste on the European Union's (EU) agenda. In that document, the EC set the aspirational goal of reducing resource inputs into the food chain by 20% and halving the disposal of edible food waste by 2020 (EC 2011, 18). Consequently, the 7th Environmental Action Plan (EC 2014a, 38) promised a "comprehensive strategy to combat unnecessary food waste." The issue also features prominently in the recently adopted Circular Economy Action Plan (EC 2015). At the international level, food waste is also receiving close attention. For example, the Sustainable Development Goal 12.3 calls for halving "per capita global food waste at the retail and consumer levels" and reducing "food losses along production and supply chains, including post-harvest losses" by 2030 (UN 2015, 22).
The environmental implications of food systems have been assessed from bottom-up (Jungbluth et al. 2000;Foster et al. 2006;Lukas et al. 2015), top-down (Tukker and Jansen 2006;Schmidt and Merciai 2014;Reynolds et al. 2015), or both (Jungbluth et al. 2011;Virtanen et al. 2011) perspectives. Most assessments of potential environmental savings from better practices focus on dietary changes (Baroni et al. 2007;Marlow et al. 2009;Tukker et al. 2011;Wolf et al. 2011;Vanham et al. 2013;Auestad and Fulgoni 2015). In the last years, more attention has been paid to potential savings attributable to food waste prevention practices. Those who have investigated this issue have mainly focused on potential greenhouse gas (GHG) savings (Audsley et al. 2009;Venkat 2011;Matsuda et al. 2012;Bernstad and Andersson 2015), although a few have also addressed other environmental categories (Chapagain and James 2011;Gentil et al. 2011;Vanham et al. 2015;Reynolds et al. 2016a;Reutter et al. 2017). One of the factors limiting the production of similar studies is insufficient data availability. Recently, the amount of studies quantifying the amount of food wasted at different stages of the food value chain has increased significantly (WRAP 2009;Monier et al. 2010;Baptista et al. 2012;Hafner et al. 2012;HISPACOOP 2012;Tatlıdil et al. 2013;WRAP 2013aWRAP , 2013bBrautigam et al. 2014;Silvennoinen et al. 2014;Reynolds et al. 2016b), although considerable methodological challenges remain, thereby limiting the comparability of the available country estimates (Hanssen et al. 2013).
In this article, we use a top-down approach to quantify the environmental pressures of the EU28 (European Union 28 Member States) food system both from the production and consumption perspectives in the period 1995-2011. Likewise, we combine the results with data on food waste to estimate the scale of potential environmental savings in relation to food waste reduction targets in the EU28. We frame the assessment using the environmental themes covered in the dashboard of indicators proposed in the Resource Efficiency Roadmap (EC 2011)-namely carbon, land, materials, and blue water. Thus, we first describe the methodology and main data sources. Next, we present the main results and discuss the findings. The last section concludes.

Some Basic Considerations
This article assesses the environmental pressures associated with food production and consumption in the EU28 during the period 1995-2011. The environmental pressure categories include GHG emissions (excluding those from land use and land-use changes), land use, domestic extraction used (DEU; hereafter referred to as materials), as well as blue water consumption. We offer two complementary views of the system. The first one, the production perspective, accounts for domestic pressures in the production chain of food products. In the second one, the consumption perspective, we allocate the domestic pressures to consumers through the equations presented in the next section. In this article, we use the term foodprint to refer to the cradle-to-gate environmental pressures related to food consumption (i.e., the footprint of food). This should not be mistaken for footprint, which usually covers not only food products, but all the remaining products consumed by final consumers in a country.
The subject of the analysis is food supplied for human consumption. As we use it in this article, the term food covers all products that are commonly part of the human diet. This does not only include solid products, but also beverages. For a better overview, we sort the food products in nine groups (see table 1 below and table S1 in the supporting information available on the Journal's website). Our definition of food consumption excludes edible products that are further processed in the food industry. These are intermediate inputs in the production process of an end product. For instance, this analysis does not consider grapes for wine production to be an end product. By considering the consumption of wine (i.e., the end product), we also cover for the grapes used in its production process.
When setting the boundaries of the system, one should bear in mind that food is not only consumed by final consumers such as households. Although final consumers also comprise governments, nongovernmental organizations, and similar organizations, their relevance in relation to generation of food waste is quite small. Hence, we refer to all final consumer groups as households for better readability. A non-negligible amount of food is consumed in several service industries, such as hospitals, universities, schools, prisons, stadiums, cinemas, restaurants, hotels, and so on. In input-output (I-O) tables, these food purchases take place within those service industries and, ultimately, form part of their value chain. We refer to these sectors as food-related services (see table S2 in the supporting information on the Web).
As for food losses and waste, we follow Parfitt and colleagues (2010), who describe food losses as food wasted in preconsumer stages, while food waste covers food wasted by end consumers. In this assessment, we only address the latter. In line with the previous paragraph, we split up the consumer group into households and bulk consumers (e.g., food-related services). Further, we only assess avoidable and possibly avoidable food waste-in other words, food that was edible at some pointthereby ignoring unavoidable waste such as bones and peelings.

Calculating the Environmental Foodprint in EU28
The EXIOBASE v3.3 multiregional input-output (MRIO) model (Stadler et al. 2018) is the main data source in this calculation. The monetary structure of EXIOBASE represents 200 product groups for 44 countries that account for more than 90% of the world's gross domestic product (GDP). The remaining countries are grouped in five rest of world regions. Likewise, it contains substantial industry-specific environmental information, including extraction of raw materials, air emissions, water use, and land use. This assessment aggregates the 49 countries and regions to 12 (see table S3 in the supporting information on the Web for the aggregation key used).
In environmentally extended I-O analysis, the formula to calculate the direct and indirect environmental pressures associated with final consumption is well known (Miller and Blair 2009) where m denotes the environmental footprint, B the direct environmental pressure intensity, (I-A) −1 represents the Leontief inverse, and y the final demand. Yet, in line with the explanations in the previous section, this formula needs to be adapted to consider the intermediate consumption of food products in food-related services. We do this in several steps. First, we filter the final consumption of countries to remove the nonfood products, which yields the final consumption of food (y F ). With the standard formula of I-O analysis, we can calculate the monetary output required (x ) to satisfy the final demand of nonfood products (y )-which include food-related services-as shown below (equations 2 and 3).
The intermediate consumption (z ) matrix can be reproduced by multiplying the monetary output (x ) by the row-and column-specific input coefficients (a ij ). The resulting matrix does not contain any of the inputs required in the value chain of the final demand of food (equation 4).
The second step consists of identifying the food consumed in food-related services. That is defined as follows (equation 5): In this case, row i refers to a food product, column j to foodrelated services, and l jj is the own use in food-related services as given in the Leontief inverse (I-A) −1 , that is, the fraction of food-related services required to produce one unit of such services. The sum of each z ij element yields the intermediate food demand vector (z F ). Removing the own use within the food-related services avoids double accounting.
The foodprint (m F ) is estimated as in equation (1), but using the sum of the final (y F ) and intermediate (z F ) demand food vectors as reference product (equation 6): The foodprint is here given according to the final and intermediate demand of the consumer, but it can also be represented according to the place where the pressures occur. We use that as the equivalent to the production perspective.

Estimating the Potential Environmental Savings from Food Waste Prevention
In order to estimate the potential environmental benefits from reducing avoidable and possibly avoidable food waste from consumers, we have taken product-specific food waste fractions from national studies as references. In spite of the growing number of such studies, there are several limitations that need to be overcome to increase the reliability of, and comparability between, studies. Examples of such limitations include lack of a common methodology, the classification used for waste statistics, which is not clear enough to isolate food waste in certain aggregated categories, and the difficulty of identifying the fraction of food waste included in household waste (Hanssen  (2009, 2013b) due to the comprehensiveness of the studies. These data have been adapted to match the product classification used in this article as explained in the Supporting Information on the Web. Given that the values refer to the UK, the results should be interpreted carefully and taken only indicative of the extent of potential savings. This is also the main source for consumer food waste used by the Food and Agriculture Organization of the United Nations (FAO) for European countries (Gustavsson et al. 2013).
The potential savings (m sav ) is the result of multiplying the total footprint by total avoidable and possibly avoidable food waste. The latter is the sum of the waste fraction of bulk consumers (w z ) multiplied by the share of bulk food consumption over the total food consumption plus the waste fraction of final consumers (w y ) multiplied by the share of final food consumption over the total food consumption. All these parameters take the form of vectors in which the rows match the product classification used in the MRIO model (equation 7).

Results
Between 1995 and 2011, the amount of food provided for domestic consumption in the EU28 has grown by 7% (figure 1). Supply is dominated by vegetables and fruits, milk and dairy products, followed by other processed products such as bread and pasta, beverages, and meat products. Although there have been no major changes in the broad demand structure, there have been shifts within the aggregated product groups. For instance, over the last two decades, pork products still represent around half of meat consumption, but there seems to be a shift from cattle toward poultry products. Similar shifting consumer preferences relate to fruits where oranges and other citrus fruits are gaining market shares from of apples and bananas. There are also shifts among vegetable oils where olive and palm oils have increased their relevance at the expense of sunflower seed oil.
The environmental intensity of food consumption is very diverse. Meat products, other animal products, and other processed products have a much higher contribution to the EU28's foodprint in relation to their mass. At the opposite end of the spectrum, vegetables, dairy products, and beverages rank much lower in environmental intensity. In absolute terms, the environmental foodprints of the EU28 related to GHG emissions and land use have dropped by 11% and 13%, respectively, between 1995 and 2011. In contrast, the material (DEU) and blue water foodprint have risen by 8% and 17%.
From the production perspective, domestic GHG emissions, land use, and material extraction related to the food system have decreased 18%, 11%, and 2%, respectively, between 1995 and 2011. In contrast, domestic blue water consumption has experienced a small increase (9%) in the same period (figure S1 in the supporting information on the Web).
The absolute domestic pressures and the environmental foodprints of all environmental pressure categories shows that the EU28 is a net exporter of pressures (figure 2). The ratio of consumption-based against production-based pressures is higher in the case of blue water consumption (2.13) and lower for GHG emissions (1.25). Depending on the pressure category there are different regions affected. When looking at net pressures shifted elsewhere by the EU28, the Asia and Pacific region-excluding the Middle East-ranks first in GHG emission and blue water consumption. America appears first in land and materials. Africa also ranks high in terms of land and materials, while imports from the Middle East embody high quantities of water. Environmental pressures shifted to other European countries are relatively small compared to the other world regions.
Given the magnitude of the environmental pressures of the European food system, measures to improve the efficiency of the value chain, including the reduction of waste, offers a huge potential to reduce environmental pressures. Theoretically, around 20% of the European environmental foodprint could be reduced by completely eliminating avoidable and possibly avoidable consumer food waste (figure S2 in the supporting information on the Web). Relative to the total footprint, the potential savings range between 3% for GHG emissions and 15% for blue water consumption. According to our estimates, more than 90% of the potential environmental benefits are attributable to eliminating the fraction of household food waste that has been edible at some point as shown in figure 3.
Regarding the contribution of individual product groups, figure 4 shows the hotspots of avoidable and possibly avoidable consumer food waste. For different product groups, the Y axis indicates the potential savings that could be obtained by reducing household food waste to zero. The environmental pressure differs depending on the environmental pressure category as shown in figure 4. For readability, the points in figure 4 represent a mean value of all four environmental pressure categories to give an overall impression of the environmental relevance. The X axis shows the fraction of food wasted. Other processed products offer considerable opportunities to reduce the pressures associated with consumer food waste. This category comprises, among others, wheat-based products, such as bread, pastries, or cakes, that are thrown away when not consumed shortly after having been baked. Thus, this type of product is not only thrown away in high proportions by consumers and vendors alike, but, at the same time, also represents the products whose contribution to total food waste is highest. Vegetables and fruits rank second in relevance. In this case, more than one quarter of the acquired product is discarded. This waste fraction represents more than one third of the potential savings in freshwater consumption and more than 10% in material and land use. Waste of meat products is also relevant mainly due to its high environmental pressure-to-mass ratio. GHG emissions, the use of (pasture) land, and materials are the main environmental categories that would benefits from decreasing the fraction of this product (see figure S3-6 in the supporting information on the Web for more details).
At least in the short-to mid-term perspective, a complete reduction of avoidable and possibly avoidable consumer food waste seems unattainable. For this reason, we focus on the potential gains from achieving the food waste reduction target formulated in the Resource Efficiency Roadmap, which calls for halving the disposal of edible food waste by 2020. As explained earlier, we only consider consumer food waste. In the absence of reasonable projections for 2020, we use 2011 data to give an indication of the magnitude of the potential yearly benefits. The blue bars in figure 5 indicate the scale of the savings (as percentage of the EU28 foodprint) if the EU would succeed in cutting the waste of each of the foodstuffs by 50%. The orange bars represent the same savings, but as percentage of the EU28's total environmental footprint instead of only the foodprint. The error bars in the figure indicate the variability of the environmental savings. The upper end of the bar shows the potential benefits of achieving the target of halving food waste by addressing the most environmentally intensive products first (i.e., those that have high pressure-to-mass ratios), while the  lower part of the bar indicates the benefits of halving avoidable and possibly avoidable food waste by focusing on the least intensive products first.

Discussion
This article has associated the food consumed by Europeans with considerable environmental pressures across different pressure categories both in the domestic territory as well as abroad. Our results show that European food consumption increases GHG emissions and material use primarily in Asia and the Pacific region. It contributes to the scarcity of blue water mainly in the Middle East. Primarily in America and Africa, European food consumption has an impact on land use. Although not addressed here, the direct and indirect environmental consequences associated with the food system go beyond these areas and include biodiversity loss mainly as a result of land use and land-use changes (Newbold et al. 2015), soil erosion (Verheijen et al. 2009), and many other (UNEP 2016). The trade-induced shift of environmental pressures also has major social and political impacts, for example, the export of blue water from the Middle East into the EU contributes to social tension and political instability of the region (Kliot 1994). Likewise, using land for African exports to Europe can get in conflict with the local food production.
At the same time, the food system is highly inefficient, for a lot of edible food is lost along the value chain. The problem is likely to increase in the coming years unless political measures are adopted (EC 2014b).
According to our estimates, consumers throw away almost one fifth of the edible food they acquire (i.e., excluding bones, egg shells, and the like), which represents between 3% and 15% of the EU28's footprint depending on the environmental pressure category. In our analysis, we find that reducing the waste of meat products could have relevant environmental benefits due to its environmental intensity per mass unit, while other processed products, such as bread, pasta, etc., and fruits and vegetables, would bring considerable benefits due to the quantities wasted. When looking at the actors involved, households account for more than 90% of the potential savings.
Although the magnitude of the potential environmental benefits and the potential positive social and geopolitical impacts should sufficiently justify ambitious food waste prevention policies, so far, only a handful of EU Member States have implemented targeted food waste prevention initiatives. Most of these strategies either lack quantifiable results or they are on such a small scale that no significant impacts can be expected at the macro level (EC 2014b). So far, the EC is acting as an enabling agent that supports the Member States (EC 2014b), for example, by contributing to the establishment of a common methodology to quantify food waste (EC 2015). Nonetheless, the EC has not yet adopted the food waste prevention strategy promised in the latest Environmental Action Plan (EC 2014a). From an economic point of view, food waste reductions could yield net monetary gains, since the cost of prevention would be outweighed by the savings in waste management and environmental externalities (EC 2014b). How these gains are reinvested in the economy could also have relevant environmental implications if they lead to a rebound effect (Salemdeeb et al. 2017).
Only a few Member States have adopted quantitative food waste reduction targets, all of which are nonbinding (EC 2014b). At the EU level, the commitment to meet the reduction target from the Sustainable Development Goals by 2030, and the aspirational goals of the Resource Efficiency Roadmap (reducing the resource inputs in the food chain by 20% and halving the disposal of edible food waste by 2020) still stand. As shown in this paper, halving avoidable and possibly avoidable consumer food waste would result in a 2% to 7% yearly reduction of the environmental footprint of the EU28, depending on the pressure category. At the same time, meeting the 50% reduction target for avoidable and possibly avoidable consumer food waste would lead to an 11% decrease in land, material, and blue water inputs along the food and drinks value chain, thereby contributing strongly toward the second target of the Resource Efficiency Roadmap (reduction of resource inputs by 20%). The reduction potential depends on the food product-specific environmental pressure, which can vary considerably. Thus, the composition of the food waste avoided will determine the extent to which environmental pressures are reduced.
The results we present here provide an overview on the system that delivers the food consumed by Europeans and the level of inefficiency of that system in terms of food waste by consumers. In this vein, the MRIO model used seems to be accurate enough to show major trends, although the results at the individual product level are more uncertain (Stadler et al. this issue). The results are also influenced by the inherent limitations of monetary MRIO models, for example, relatively low-resolution compared to life cycle assessment databases. Likewise, the homogeneous price assumption can lead to misallocation of environmental pressures between the final consumption of countries with different prices for the same food commodities. When it comes to food acquisition data, differences between EXIOBASE and Food Balance Sheets are to be expected, not least because in the latter the amount of food available for human consumption is commonly calculated as a residual of the other components of the domestic supply. Although the effects at the macro level do not seem to be very relevant, important yearly changes are sometimes visible at the product level. In this context, it should be highlighted the disagreement between the Food Balance Sheets and other data sources on food consumption. This issue has been documented by several authors (Rodrigues et al. 2007;Vandevijvere et al. 2013). The hybridization of life cycle and I-O databases could be used to provide more accurate results at the product level. Such a combination of methods could also be used to include the waste management stages of food, which are allocated to the waste sector in I-O models and omitted from this assessment. The uncertainty at the product level increases due to the different classifications in the databases used, which limits the accuracy of bridge tables used to reconcile data from different sources. For instance, sugars and sweeteners, vegetable oils, and wheat products are an important energy source in the European diet. While these are represented in individual groups in the Food Balance Sheets and EXIOBASE, this is not the case for studies addressing consumer food waste. In the latter case, an important share of sugars and sweeteners is embodied in confectionary, juices and refreshments, bakery products, dairy products, etc. Likewise, part of vegetable oils remains in fried food products. Further, wheat products, such as bread, pastries, pasta, or wheat-based cereals, are included in the same aggregated category in the Food Balance Sheets, which makes it difficult to assign waste factors to each of these items. The same applies to plate scraps reported in food waste studies, which do not have a one-to-one equivalent in the other classifications.
It should also be noted that the food waste factors used refer to the UK and will hence most likely differ in other EU countries. This also adds uncertainty to the results. Therefore, they only indicate the scale of the environmental problems and potential benefits of appropriate policies in the EU28. In absence of better food waste data, the FAO has also used UK data as a proxy for Europe (Gustavsson et al. 2013), which increases the acceptability of the assumption.

Conclusions
In this paper, we have quantified the environmental pressures in four categories (GHG emissions, land use, material use, and blue water consumption) associated with food consumption in the EU28 both from the production as well as from the consumption perspectives for the period 1995-2011. Likewise, we provide insights on the environmental savings that could be achieved by complying with the aspirational targets calling for halving (avoidable and possibly avoidable consumer) food waste by the end of this decade.
Our results support the need to speed up the adoption of measures to combat food waste in the EU28, especially focusing on household waste. The level of policy intervention since the EC flagged the problem in 2010 (Monier et al. 2010) is not sufficient to reverse trends both at the EU and Member State level. Considering the limited progress to this day and the expected evolution of food waste in the absence of additional measures (EC 2014b), there is no indication that the Resource Efficiency Roadmap's target of halving food waste by 2020 will become more than a political aspiration. Unless new targets are adopted as contemplated in the Commission's impact assessment (EC 2014b), the next milestone is defined by the Sustainable Development Goals (SDG) to reduce per-capita food waste generation by 50% by 2030. It remains to be seen whether the EU will rise to the challenge.
Against this background, we illustrated the use of the EXIOBASE database for characterizing the food system at the macro level from an environmental standpoint, and for quantifying a wide range of domestic and global environmental pressures and potential benefits arising from food waste reduction policies. The usefulness of this type of database for evidencebased policy making in the short-to mid-term perspective will depend on advances happening in both official institutions and the research community. Regarding the former, an institutionalization process is already taking place to some extent. In this line, Eurostat is generating MRIO tables for EU28 countries and the Organization for Economic Cooperation and Development (OECD) produced a global MRIO model. Yet, the low resolution of official databases is still a limiting factor. Likewise, more resources are needed for the compilation of comparable food waste statistics. The manual recently published by the FUSIONS project (Tostivint et al. 2016) and the ongoing efforts at Eurostat are encouraging steps in the right direction. In the research community, collaborative virtual laboratories (e.g., Lenzen et al. 2014) promise a highly automated and flexible process to build and enhance the geographical and product resolution of MRIO tables, potentially extending the amount of policy-relevant research questions that can be addressed with this type of databases.