1

2 Appendix: Methodology for level 1

A summary of the methodology used for Level 1 estimates is presented in Section 2.2 of the Food
Waste Index (FWI) report. This appendix covers the details of this methodology, in particular:

• How existing food waste studies and estimates were identified and obtained (2.1)
• How the data found from these studies was evaluated to inform its inclusion and our level of
confidence in the estimate (2.2, 2.2.3, 2.2.4)
• Transformations and adjustments applied to data to increase comparability (2.2.2)
• The methods of calculation used to extrapolate data and create relevant estimates (2.3)
• Methods considered for extrapolation but rejected in favour of the one used (2.4)

2.1 Literature Review

This section describes the process of finding relevant studies for this project. The section is split into
two: the first part describes the characteristics of the studies being sought, the second part
describes the methods used for searching, including the limitations.

2.1.1 Characteristics of studies for this project

This section describes the types of food-waste estimates that were sought as part of this study. In
general, studies with comparable boundaries with the definitions of the Food Waste Index (FWI)
were sought (although differences in definition were adjusted for where possible). In addition, the
methodologies of studies included had to be of sufficient accuracy for tracking levels of food waste
over time.
Figure 1: Definitions of food waste used for the Food Waste Index.

Timeframe: When searching, we generally looked for papers which were published after 2005 in
order to find data which had been gathered no longer than 15 years ago. Despite having a cut-off
date of 15 years ago, most of the studies which were used came from the last five years. Many
countries which had studies pre-2010 have refined, repeated or updated those studies and latest
figures were used in the current research.

Table 1: Distribution of studies by publication year

Study publication date Number of studies Number of datapoints in studies

Before 2005 0 0
2005-2009 4 4
2010-2014 21 31
2015-2019 51 98
2020- 8 19
Total 84 152
Note that a single study can contain multiple datapoints (for different sectors, geographies or times). The publication year
does not necessarily reflect the age of a datapoint referred in a study; many are published in recent years but refer to older
datapoints.

Material type: consistent with the FWI methodology, we searched for studies that quantified total
food waste: both edible parts (sometimes referred to as ‘avoidable’ or ‘wasted food’) and inedible
parts (unavoidable food waste). Studies did not need to separate these two parts from one another.
However, where a study only included the edible parts, this was collated and – where possible –
adjusted to account for the unquantified inedible parts (see Section 2.2.2.2).

Destinations: studies were sought that conformed to the destinations defined as food waste by the
FWI: co/anaerobic digestion; compost / anaerobic digestion; land application; controlled
combustion; sewer; litter / discards / refuse; or landfill. Studies were still collated if there were
discrepancies between the destinations covered by the study and those covered by the FWI: where
possible, adjustments were made (e.g., to remove food fed to animals from the estimate). This is
discussed in Section 2.2.2.7.

Sectors: studies were sought covering any of the following: Household, Retail 1, and Food Service. In
the case of Food Service, most studies reviewed did not make estimates for the entire sector. In
general, we pursued papers which from the title and abstract covered a sufficiently large portion of
the sector (i.e., restaurants, or canteens across a range of settings) but ignored those ones which
had a very particular and narrow view (e.g., studies focused on university canteens only).

The definitions of sectors are outlined in Section 3.2.1 of the FWI report according to International
Standard Industrial Classification of all Economic Activities (ISIC). There are occasional differences
between individual studies and these definitions. We applied judgement as to when a study’s
sectoral definition deviated too much from that our aim, with such studies being excluded. In many
cases, insufficient information on sectoral coverage was provided to make this assessment.

A large number of studies present waste data based not on its source but its destination, i.e.,
collection by Municipal Solid Waste (MSW) services and disposed of in landfills, incineration and
other waste destinations. Such studies were only included in situations where the MSW had been
disaggregated by waste source to a sector comparable to the sectors being used here. For a fuller
description of the approach to MSW papers, see Section 2.2.4.1.

Geographic coverage: Studies were considered for inclusion regardless of whether their waste
estimate was formed at a national or subnational level. This meant that subnational studies such as
scoping studies for municipal waste plans, which were not focused on food waste estimation but did
disaggregate waste to that level of detail, were considered. As a result, we may have included a
number of estimates which are otherwise overlooked in the food waste literature. Studies at this level
were particularly relevant for the Household sector, which was often their focus. A distribution of
datapoints by scope of study can be viewed in

1
Note: Wholesale is not covered by the Food Waste Index. Where estimates for retail have also
included wholesale, efforts were made to remove the wholesale estimate, where it was specified in the
study. In some cases, disaggregation was not possible, or it was unclear if wholesale was included in the
study. These were not adjusted further.
Table 2. The majority of studies at a regional level provided by-country estimates, these were
treated separately. Only one paper presented a multi-country sample in aggregated form, this was
not included (Malefors et al., 2019).

Often, the subnational estimate provided a per capita waste generation figure rather than a total
waste generation figure. As many subnational estimates were urban in nature, this per capita waste
may not be considered representative for the entire country. As a result, the confidence in estimates
from these studies is reduced. This is discussed in section 2.2.3.1.
Table 2: Number of datapoints, by sector and geographical scope of study

Household Food Service Retail

Nationwide 38 24 27
Municipality & Sub-national region 53 8 2

Methods and approaches: in this project, we were looking for studies which involve direct
measurement of food waste or use data from other studies that involved direct measurement. This
criterion is important as the purpose of the FWI is to track levels of food waste over time. This
purpose requires estimates to be reasonably accurate, collecting data from the relevant geographic
area and time period and using a methodology without substantial bias (or a methodology where
biases can be adjusted for).

Therefore, studies with the following methodologies were included: waste compositional analysis2,
direct weighing and scanning of wasted items. We also collated studies that involved diaries or
collated information using surveys. For studies focusing on Household food waste that used diaries,
adjustment was made to account for underestimation (see sections 2.2.2.3 and 2.2.2.4 for details),
but our confidence in the resulting estimate was lower than other methods included (see section
2.2.3.2). For surveys, estimates for household food waste obtained directly from surveys (e.g., asking
people to recall the amount of food waste generated) were not used. However, surveys of business
representatives asking them to report their waste generation were included. These surveys, and
data from industry more generally, were included due to the barrier to accessing commercial data.
In very few studies presenting such data was it clear how it was generated, i.e., whether the
businesses directly measured or estimated waste, and how robust measurements taken were. In the
interest of ensuring there was sufficient data, a level of trust that self-reported business estimates
were informed by measurement was therefore applied.

2
Many authors use the term ‘waste audit’, especially when examining the waste of a restaurant or
supermarket; all ‘waste audits’ have been coded as ‘waste composition analysis’ for simplicity.
Table 3 presents datapoints by methodology. For a discussion of method and its relation to
confidence levels, see Section 2.2.3.

Table 3: Number of datapoints, by quantification method and sector

Method Household Food Service Retail

Data from industry 2 6

Diaries 12

Literature 13 10 11

Mixed method 5 7 3

Surveys, questionnaires and interviews 2 1

Waste Composition Analysis 58 8 6

Unclear, Governmental reporting 3 3 2

Shaded blanks refer to methods which are not appropriate for that sector, meaning no studies were found. The solid blank
refers to a rejected methodology.

In the case of studies which combined waste generation factors with some other national statistic,
the determining factor was the origin of these waste generation factors. In some cases, the waste
generation factors were derived from direct observation in the relevant country (see for example,
the USA (U.S. Environmental Protection Agency, 2020)); in others, it is derived from a modelled
estimate, typically using the FAO 2011 estimates (Gustavsson et al., 2011), often from data that is
old and / or from another country. The former would be accepted, and the latter would not for our
purposes.

In a number of countries, there are existing publications which aggregate studies across multiple
sectors for the purposes of estimating and reporting on food waste. In the last few years, a number
of ‘baseline’ studies have been published for this purpose (see, for example, Australia (Arcadis,
2019), Germany (Schmidt et al., 2019) or the United States (U.S. Environmental Protection Agency,
2020)). These typically took the form of meta-analyses for the study scaled by country-specific
factors. Where these studies were identified, they were taken as the authoritative source for the
country and we did not prioritise further searches for those countries, nor the primary data sources
on which the baseline was formed, unless some sectors were not covered by the publication. As a
result, for some countries there are many more studies on food waste than presented in this
database.

2.1.2 Search process

This review took a multi-pronged approach to sourcing data and estimates:

Existing review papers: As a starting point, certain literature reviews were used, notably the Xue et
al. (2017) meta-analysis. This paper collated national estimates of food waste for multiple sectors,
with extensive coverage of studies until approximately 2015, including meta-analyses published up
until that point. The database in the supplementary information of Xue et al. was analysed to
understand the degree to which studies referenced conformed to the required characteristics, as set
out in section 2.1.1. Studies that measured a specific foodstuff or commodity (e.g., wheat) were
excluded from further consideration. In contrast, studies that measured ‘Food & Drink’, ‘Food, Drink
& Tobacco’; ‘Food & Organic Waste’; ‘Total Food’ or ‘Total’ were retained for consideration. The Xue
et al. database was also filtered by measurement method (to remove any calculations using proxy
data) and sector to form a longlist of data that could potentially pass the current study’s criteria.
These papers were then accessed and read to obtain further methodological detail which would help
in determining studies to accept or not.

Several of the studies in the Xue et al. (2017) spreadsheet were listed as ‘Literature’ papers, i.e. the
food waste estimate was derived from a reference in the text. In these cases, the reference was
followed where possible to the original source in order to determine its suitability. Where possible,
the original source paper has been referenced directly in the current study.

In addition to the Xue et al. (2017) data, several other meta-analyses were used in the research
process. This included:

▪ Relevant chapters on regional measurement in the Routledge Handbook of Food Waste

(Reynolds et al., 2020) were read and bibliographies followed if the papers had not
previously been identified.
▪ Several Europe-wide reviews of member-state estimates, notably BIO Intelligence Service
(2010), Stenmarck et al. (2016) for the FUSIONS project and Caldeira et al. (2019). Based on
the information in these documents, some studies referenced were obtained and read. For
studies which were inaccessible to the authors, whether a study was included in the current
research was decided on by the description of the method provided by the meta-analysis.
For example, the Stenmarck et al. (2016) estimate includes judgements of data being a
‘sufficient quality’ with definitions of that quality benchmark. Similarly, Caldeira et al. (2019)
offer in most cases a good level of methodological detail of the studies they reference. In
these cases, the Method was listed as ‘Literature’.
▪ WRAP’s Food Waste Atlas was consulted. This is a repository for food waste data from a
wide range of sources. Some of these sources – such as for specific private companies –
were not relevant for our purposes. However, the data on international estimates and from
academic papers was consulted to check for estimates not otherwise identified.
▪ Abiad and Meho’s (2018) review on food loss and waste in the Arab world was consulted as
a source of information from this region. The individual papers were then followed up and
included based on their suitability.
▪ Two regional analyses by Hamid El Bilali (2018; 2020) for the Gulf Cooperation Council and
North Africa was consulted, with any potentially suitable studies followed to the original
source. However, most of these were speculative figures not derived from direct
measurement. As the author points out for North Africa, there is “in the selected documents
no comprehensive analysis of the extent of food wastage in distribution and consumption”.
Whilst some studies on food losses were identified, they were not relevant for our purposes.
▪ The bibliography of a recently released meta-analysis on directly measured food waste by
Dou and Toth (2020) was kindly provided by the author while still in the review stage. Those
studies in the bibliography which had not previously been identified were accessed and
referenced where applicable.
▪ Van der Werf and Gilliland’s (2017) review article of food waste in developed countries was
consulted, with studies with the appropriate scope being read and referenced directly.
 ▪ The World Bank’s ‘What a Waste’ dataset, which is associated with the report What a Waste
2.0 (Silpa Kaza et al., 2018) was analysed. The information in this dataset contains the
amount and composition of MSW for a wide range of countries and many cities around the
world. Depending on the country, MSW may cover some or all the sectors relevant to the
current study (Household; Food Service; Retail). In some studies referenced in the What a
Waste dataset, information is disaggregated by sector, allowing it to be used in the current
study (see 2.2.4.1). A second complicating factor was that many studies did not report food
in the MSW streams separately from other organic material. To identify studies that were
likely to contain food-specific data, the dataset was filtered to only include cities or countries
where different data for different organic fractions was presented (specifically, garden waste
being separately reported from ‘organics / food’). These papers were searched for: several
were personally communicated to the World Bank for the purposes of their report and were
not public and some were not findable based on the information presented. Any relevant
papers found through this dataset were accessed and added to our dataset directly as
unique paper references.

Searches of academic databases: Additionally, direct searches were conducted online. These were
conducted both using Google Scholar and the Sciences Po Bibliothèque, a dedicated social sciences
library with access to over 20,000 electronic journals. Numerous searches were conducted using
these engines, which combined search terms such as “food waste” and “quantification”,
“measurement”, “national estimate”, “wasted food” “food wastage” and “food loss”. These were
conducted both as time-limited searches after 2014 (to prioritise those papers not considered by
Xue et al. (2017)) and on other occasions, after 2005. These searches were conducted during July-
August 2020. Searches were conducted in English, with supplementary searches conducted in Arabic
and French. Studies identified in other languages were considered where found using online tools
Google Translate and DeepL to check for possible relevance.

The search terms present a huge number of results, many of which were not usable for our
purposes. Papers which were specifically about the potential valorisation of food waste and its
chemical composition were ignored, unless the title alluded to the collection of waste from a specific
geographic area. Similarly, a large number of papers were returned which focus on the demographic
and behavioural determinants of (self-reported) food waste, or food waste-related behaviours,
alongside a large number of papers documenting interventions in specific sectors to reduce food
waste. These papers are important for designing policy to reduce food waste and deliver SDG 12.3.
They were not, however, relevant for our purposes and so were filtered out.

During the review, papers were sought that mentioned a specific geographic area (whether national
or subnational, e.g., a city or a state), direct measurement of food waste, and specific sectors
(Household, Food Service, Retail). If the title, excerpt and/or abstract mentioned some or all of these
elements, it was downloaded and reviewed in more detail. In many cases, the paper reviewed at this
stage did not have an original estimate of food waste but referenced estimates from other papers,
which were then tracked down if perceived relevant. This led to a ‘snowball’ search from the studies
found through the online searches.

It should be noted that due to generic search strings such as “food waste” + “quantification” turning
up as many as 17,000 results on Google Scholar, it was not possible to review every page of search
results for every search conducted. After a few hundred results the search results tended to become
less relevant (e.g., focusing on chemical composition, anaerobic digestion or consumer attitudes).
Once this point was reached, no further citations were reviewed, and a new search was typically
started. Searches conducted using Sciences Po Bibliothèque were more precise and every page of
the search was therefore evaluated. Whilst it is a limitation that searches could not be pursued in
full, it was assumed that by reading the identified papers and following the references to other
estimates, those we did not find directly would be found through the bibliographies of other papers.

In September 2020, after the initial period of review, codifying and building the database, some
country-specific searches were conducted. These were to focus on two regions where no possibly
usable estimates had been identified: Northern Africa and Central Asia. Using Google Scholar and
the main Google search (to possibly include results from NGOs, local associations or government)
searches for [country name] + “food waste” OR “wasted food” OR “food wastage” were carried out
for each country in the regions. None of these searches returned usable estimates.

It should be noted that three other regions have no datapoints in them: Melanesia, Micronesia and
Polynesia. Region-specific, rather than country-specific, searches were carried out and two notable
sources were found: a 2020 paper by Joseph and Prasad (2020) and a 2016 regional strategy
document (SPREP, 2016). The former provides waste compositional analysis information of MSW,
the latter an estimate of household generation but only providing an organic rather than food-
specific share. Whilst some combination of these two sources could provide a rough estimate of
household food waste generation, at least to identify the order of magnitude, MSW studies and
household studies reporting organic waste only were not used in the current study (see Section
2.2.4).

Waste-specific databases: As well as searches of academic databases, some data sources were
provided through personal contacts of the team which helped identify other sources of information.

One contact highlighted the JICA (Japanese International Cooperation Agency) database. Based on
the JICA studies already consulted by that point, “waste management” was searched across the last
ten years of publications on the database. This provided a fruitful avenue and several JICA studies
have been included. Whilst these studies were primarily designed as scoping to understand the scale
of waste problems more generally in a city or country, the inclusion of food-specific measurement
(rather than simply ‘organics’) in detailed, Household-level waste compositional analyses provided
usable data. Unfortunately, often the Retail and Food Service figures were either grouped in ways
we could not disaggregate (i.e., supermarkets and restaurants considered together) or were not
scaled beyond the waste per establishment, limiting their viability in producing an estimate for the
city or country in question without additional, unavailable data. This data gap is discussed more in
Boxes 2 and 4 of the main FWI report.

Based on the experience with the JICA database, it could be possible that other international
development organisations have databases with waste estimating studies. There is some evidence to
suggest that JICA are particularly unique: JICA papers were regularly referenced in the academic
literature, alongside which there was a notable absence of studies from comparable organisations.
Similarly, a UN Habitat study in Nairobi, Kenya (Takeuchi, 2019) based itself off a JICA study in the
same area (JICA, 2010) and replicated its methodology, implying that JICA presents a body of
expertise on this issue. It may still be possible that other development organisations present
‘untapped goldmines’ and this being signposted to the authors would help ensure in future updates
and revisions that these estimates are included.

Reaching out to the research community: Another use of contacts involved posting in a Google
Group for researchers on food waste3 explaining the nature of the review, the boundaries and what

3
The International Food Loss and Food Waste Studies Group: https://foodwastestudies.com/join-the-
discussion/
kind of papers we were looking for. This garnered several responses, especially for very recent
estimates, which were then reviewed for suitability if the paper had not already been identified.

The research group was also the avenue by which contact was made with a researcher, Zhengxia
Dou, whose recently published study (Dou & Toth, 2020) similarly focuses on direct measurements
of food waste. Prof. Dou very kindly shared the reference list of this unpublished study. This
comprehensive list of references served as both a helpful avenue to find some more studies but also
a confirmation of the approach taken up to that point, as most of the studies relevant to food waste
quantification had already been identified.

Key food waste researchers were also contacted individually to see if they were aware of additional
data. This included Felicitas Schneider, Gang Liu, Gustavo Porpino and several regional and sector
specialists at UNEP and FAO.

2.1.3 Limitations with search method

No search will be 100% effective. This wide-ranging searching strategy was designed to obtain the
maximum number of relevant studies within the constraints of the project.

As previously mentioned, resource and time constraints meant it was not feasible to evaluate every
single page of search results from the Google and Google Scholar searches, given both the large
number of results and large number which were not relevant for our purposes here. Whilst it is
reasonable to assume that notable references should be accounted for through the combined
search and ‘snowball’ references method, it is possible that a small number of studies exist which
were not found. For example, many studies were identified through other strands of the search
strategy that were primarily studies evaluating Household solid waste rather than food waste, but,
on closer inspection, disaggregated food waste from other organic wastes, allowing a food-waste
estimate to be obtained. Many of these were at a subnational level. More studies of this nature may
exist that were not identified. Furthermore, there may be other ‘categories’ of study containing
relevant food-waste data that were not identified.

Although studies from a wide range of countries in a range of languages were obtained, there
remains the possibility of geographic bias. We hope that the publication of this study helps
understand whether there are studies, currently missed by this searching strategy, from poorly
represented countries and regions.
2.2 Data Extraction and Adjustment
This section contains details on:

▪ The information recorded from each study (2.2.1)

▪ Adjustments made to data to increase comparability (2.2.2)
▪ Classification of estimates based on our confidence in the estimates (2.2.3)
▪ Decisions relating to whether studies were included in the calculations (2.2.4)

2.2.1 Data extraction

For each relevant study identified, the core information searched for and extracted (beyond
bibliographic information) was as follows:

• Geographic boundaries
• Time of study
• Sectors covered
• Methodological details, including sample size, length of sampling and representativeness
• A share estimate (e.g., x% of household solid waste was food waste, or y% of total national
food waste occurs in a particular sector).
• A total mass estimate for that sector and geography
• A normalised (per capita) mass estimate for that sector and geography
• The share of food waste which was considered edible or avoidable
• The share of organic waste which was food waste
• The waste destinations, particularly if included in the paper estimate was some waste which
goes to an avenue not considered waste in the FWI

Very few studies had all of the above information: in some cases, it was not relevant to the scope of
the study; in others the information was not reported in the publication. As much of the above list as
possible was captured.

All total and normalised mass estimates were input using the measurement scale used in the paper
(e.g., million tonnes / year, g / capita / day) and then adjusted for this study to a single comparable
figure for total mass (tonnes / year) and normalised mass (kg / capita / year).

In some cases, the original mass value was presented as multiple numbers (such as edible and
inedible waste separately) or required some calculation (such as where daily total waste generation
is presented alongside a percentage which was food waste, allowing daily food waste to be derived).
These calculations were carried out to ensure comparable figures.

We searched within papers for estimates of ‘food waste’, sometimes referred to as ‘kitchen waste’.
Definitional consistency was an issue in several papers: many studies used the terms ‘kitchen waste’,
‘organic waste’ and ‘food waste’ interchangeably within the same paper. In some cases, the term
‘organic’ would be used but only foodstuffs listed in the table describing the categories, on other
occasions the term ‘organic’ would be used in a table or graph with ‘food’ being used to label the
same category elsewhere in the paper.

To deal with these problems, we used the definitions applied by the authors. If they labelled a
category as food or kitchen waste (without further elaboration on the definition), this was
understood to mean edible and inedible food waste. In addition, most authors defined garden/yard
waste as a separate category. This presents the most notable bulky organic waste stream outside of
food, so its inclusion as a separate category increased confidence that what was labelled as ‘food’
was, indeed, food waste.
It remains possible that other, non-food wastes have been included in these categories: sanitary
waste such as nappies and animal excrement in particular vary substantially in the regularity with
which they are disaggregated in waste compositional analyses, and it is often unclear what category
they have fallen into (‘organic’, ‘general’, ‘other’ etc.). This is a limitation of the estimate which
unfortunately could not be avoided: we have tried to work as best we can with the data available
from the papers. For a discussion of the organic waste studies which were excluded, see section
2.2.4.2.

2.2.2 Data Adjustments

There are a number of different ways in which food waste can be measured and reported. This
presented a challenge as we were aiming to produce results which are as comparable across
estimates. In order to make the data as comparable as possible, a number of adjustments were
carried out to specific datapoints to account for time difference, measurement boundaries or
measurement bias. These are outlined below. Some of these adjustments add extra uncertainty to
estimates, reducing our confidence (see 2.2.3).

2.2.2.1 Population statistics

In order to create a single comparable food waste baseline, all estimates were normalised to a single
year: 2019. To do this, we assumed that per capita food waste has held constant since the time of
the estimates identified. This enabled us to use 2019 population statistics for the purposes of scaling
per capita waste estimates to country-wide estimates and global food waste extrapolations. All data
on population and other relevant national indicators was downloaded from data.un.org on
03/09/2020.

Some of the identified studies did not present waste as a per capita estimate, but rather total mass
for a specific sector and location. To enable scaling, these estimates were normalised to a per capita
estimate. For this, the same UN data source was used. As this data source does not provide a
continuous time series, a linear interpolation was made between available data points to infer the
population of intermediate years. This allowed an estimate of population in the year of each study to
be used to normalise the total mass estimate. Once expressed as a per capita waste estimate, it was
possible to scale these by 2019 population figures to form a country estimate.

A worked example: Japan’s total waste estimates (Food Industry Policy Office, 2017) come from
MAFF survey data for 2014. Population data was available from the above source for 2010 and 2019
only; 2014 population is inferred by linear interpolation between these two datapoints. The total
mass estimates are used to form per capita estimates from this figure. This normalised, per capita
figure is then multiplied by the 2019 population to form an estimate for 2019.

Some studies did not provide information as to the year the observation took place, or what year the
waste estimate refers to. In these cases, the year of observation was assumed to be two years prior
to the year of publication. In the case of subnational studies presenting total mass rather than per
capita, the population as listed in that study or paper was used to ensure consistency in the
boundaries used to define the area.

2.2.2.2 Edible share adjustment

Food can be divided into the share which is edible by humans (such as the flesh of a fruit or animal
meat) and that which is inedible (such as onion skins, banana peels and animal bones). Due to the
inedible fraction, a world without some degree of food waste is unlikely: eating a banana often leads
to wasting its skin. A reduction in the edible share will have a knock-on effect on the amount of
inedible waste (fewer bananas wasted may mean fewer bananas are grown to meet the same
demand, which may lead to fewer skins wasted in total). As a result, general policies and
interventions which target food waste are usually targeting the edible share. Many studies therefore
focus their analysis on the edible fraction of food waste as this is the portion which is directly
targeted by food waste reduction campaigns.

The definition of food waste used for the purposes of SDG 12.3 encompass both the edible and
inedible fractions of waste. In order to compare studies which only record the edible waste with
those which record edible and inedible parts, the omission of the inedible fraction required
adjusting. Many studies report both the edible and inedible waste (or the similar distinction between
‘avoidable’ and ‘unavoidable’ waste). These were taken to mean the same thing: whilst there are
subtle differences between ‘avoidable’ and ‘edible’, they were considered sufficiently comparable. 4
In the cases where ‘possibly avoidable’ was measured, this was divided into two and allocated
evenly between ‘avoidable’ (edible) and ‘unavoidable’ (inedible)5. The share of waste which was
edible or avoidable was then converted into a percentage share. From here, it was possible to create
sector-specific scaling figures through the following calculation:
1
𝑆𝑒𝑐𝑡𝑜𝑟 𝑒𝑑𝑖𝑏𝑙𝑒 𝑠𝑐𝑎𝑙𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟 =
𝑆𝑒𝑐𝑡𝑜𝑟 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑒𝑑𝑖𝑏𝑙𝑒 𝑠ℎ𝑎𝑟𝑒
Table 4: Edible share of waste only adjustment figure

Household Food Service Retail

Number of studies 17 8 4
Number of countries 16 7 4
Average edible share 49% 66% 85%
Scaling factor 2.04 1.50 1.18

2.2.2.3 Household Diary adjustment

Many countries have used food waste diaries – where a household member weighs and documents
each incidence of food waste for a period of time – as a way of measuring household food waste.
However, as is pointed out in Quested et al. (2020), diaries “substantially underestimate HHFW”.
They identify four main factors for this: behavioural reactivity (people behave differently and waste
less in the study period); misreporting (not all discarded items are recorded); measurement bias (the
respondents do not weigh all their waste) and self-selection bias (those completing a diary are
different from the wider population). Nonetheless, there are many reasons why a diary may be a
suitable study method: it can be cheaper to carry out than a waste compositional analysis; it may
cover all disposal paths not included in a solid waste study, such as down a sink or home
compositing; it can provide the causes behind given instances of waste and how they relate to
specific food products and it can be combined with demographic or attitudinal surveys to find
correlates to increased food waste generation. A recent paper by Withanage et al. (2021) provides a
good overview of when different methodologies for measuring food waste are most applicable.

For our purposes, the incomparability between diaries and waste compositional analysis presented a
problem as it would indicate that some countries or territories have lower waste than others when

4
(For a discussion of these definitions, see Section 2.1.2 of: WRAP, 2018)
5
The assumption behind this decision was based on the analysis in (WRAP, 2018)
this may in fact be a result of methodological differences. In order to correct for this, therefore, the
scaling factor presented in Quested et al. (2020) was used. They identify five studies where diary
estimates of HHFW can be directly compared to waste compositional analysis data and the amount
by which diaries underestimate actual waste (average: 30.2%). This is then converted to a scaling
factor as follows:
1
𝐷𝑖𝑎𝑟𝑦 𝑠𝑐𝑎𝑙𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟 =
(1 − 𝑑𝑒𝑔𝑟𝑒𝑒 𝑜𝑓 𝑢𝑛𝑑𝑒𝑟𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑖𝑜𝑛)
which, when calculated using the average underestimation of 30.2%, comes to 1.43 (rounded). A
preliminary analysis of the (unadjusted) per capita waste data in our database suggested a similar
margin of error, giving us some increased confidence in the use of this figure.

Studies which calculated household food waste through a diary methodology were therefore
multiplied by this scaling factor to get an adjusted figure of waste generation.

2.2.2.4 Household Diary and edible adjustment

In a limited number of cases (see Table 5), Household studies employed a food waste diary
methodology and measured only the avoidable or edible waste. In these cases, both adjustment
factors are applied through multiplication. This decision was felt to be justified in that the reasons
each of these methodological details requiring adjustment were different: reporting edible waste
only is an issue of measurement boundaries, food waste diaries by contrast underestimate due to
bias.

Together, the adjustment factors for Household diary (1.43) and inedible waste (2.04) become a
total adjustment figure of 2.92. This nearly triples the initial estimated waste. Whilst this seems to
be a large increase, the final per capita estimates obtained from this adjustment do not become
unreasonably high, in fact they become more comparable with other countries (see sections 2.5.3
and 2.5.4 of the main FWI report) which would suggest that the level of adjustment is reasonable:

Table 5: Estimates adjusted for diary and inedible parts

Country Reference Adjusted kg / capita / year

China (Li et al., 2021) 21
Finland (Katajajuuri et al., 2014) 67
France (ADEME, 2016) 85
Spain (Caldeira et al., 2019) 78

2.2.2.5 Aggregating socioeconomic groups

In a few papers, particularly JICA papers and ones following a similar methodology to establish
household waste in an urban area, households were grouped into multiple socioeconomic groups
(mostly high, medium and low) and sampled based on that. As a result, the study would generate
multiple shares of food waste in the residual waste and multiple daily waste generation factors. In
some cases, the studies aggregated this information themselves based on the relative population
shares of those socioeconomic groups, but in some cases did not.

Where multiple estimates needed aggregating for the purposes of providing a single per capita
datapoint, the estimates for each socioeconomic group were weighted based on that socioeconomic
group’s share of the total sample. It was therefore assumed that the sampling attempted to mirror
the wider population; in many cases, this was explicitly described as the intention but in other cases
it was not mentioned at all. Therefore, there is a risk that some datapoints have been aggregated in
a way which does not reflect the distribution of socioeconomic groups in the country or territory in
question. This is a limitation but not one easily avoided: typically, studies in sub-national areas used
ad hoc, relative definitions of the socioeconomic groups based on variables such as income,
predominant housing type in an area, classification of a neighbourhood etc. As a result, finding
comparable data which could have been used to weight these estimates more accurately was not
viable, and the share of sample size was taken to be approximate to share of population.

The available disaggregated socioeconomic data and its implications for the report are discussed in
Box 1 of the main FWI report.

2.2.2.6 Aggregating study periods

In some cases, studies were carried out in multiple time periods to estimate seasonal variations,
such as between rainy or dry seasons. In cases where this was averaged by the author to create a
yearly average, this value was taken. In cases where the author did not average the seasonal
variation but instead presented them as multiple tables or datapoints, a simple mean average was
taken of these generation figures. Whilst this would not be quite as accurate as weighted averages
which account for season lengths, it was not considered to make a substantial difference and in
many cases food waste was quite consistent, whereas some other wastes (such as garden) saw
substantial variation.

2.2.2.7 Removing non-waste destinations

For a small number of studies, other adjustments were possible based on information regarding
disposal routes:

Food waste destined for animal feed is not considered waste as part of the FWI. A small number of
studies had estimates of the destinations for waste including that share which was going to animal
feed. In some cases, the authors had already removed this from the estimate which was reported as
waste. In other cases, the share going to animal feed was used to adjust the waste estimate used in
the present study. Similarly, food which is donated to charitable organisations for human
consumption is not considered waste and was removed where the authors had not already done so.

For the household sector, in order to ensure comparability, food waste disposed via the sewer was
removed where possible from studies. This applies to a very small number of studies which detailed
this information. This was primarily to improve comparability, as the majority of household studies
were explicitly conducted on the ‘solid’ waste streams (e.g. residual waste, separate food waste
collections, etc.). Sewer waste is included under the FWI definition and is included under Level 3
reporting (see Section 3.4 of the main FWI report).

2.2.3 Data Classification

All datapoints which fit the above criteria were considered for the purposes of extrapolating
estimates of food waste. However, the studies varied in their methodologies and many were
adjusted to improve comparability. These factors impact our relative confidence in the robustness
and accuracy of each datapoint and, therefore, each estimate for a specific country.

To reflect our confidence in the datapoints, studies were grouped into two ‘tiers’ which correspond
to whether the estimate for a country is High or Medium confidence (for countries without
identified estimates and therefore requiring extrapolation, confidence levels are either Low or Very
Low, see 2.3). These correspond to methodological detail: datapoints in which we have higher
confidence involved more accurate quantification, estimated waste for the entire country and had a
sufficient, representative sample size. The datapoints in which we have Medium confidence
correspond to some studies which required adjustment, namely studies focusing on a specific sub—
national area, only measuring edible waste or using diaries. Similarly, referenced figures which were
unfindable, had unclear methodology or small sample sizes were typically classed as Medium
confidence. Where a High confidence paper was available for a specific country and sector, this was
used and any Medium confidence papers for the same country and sector were excluded from
further analysis. A full description of the boundaries between confidence levels for each sector
follows.

It should be noted that confidence ratings are an assessment – based on our understanding of the
study – of how robust the estimate of food waste is for tracking food waste in the given country,
not a judgement on the quality of the study undertaken. In many cases, food waste measurement
was not an aim of the original study. Hence many good studies will be classified at a Medium
confidence level (or even excluded from consideration altogether) as the aims of the paper did not
include national food waste tracking.

To see the full list of included datapoints and their confidence level, see Section 3.

2.2.3.1 Sub-national studies

A number of studies, particularly in the Household sector, measured food waste in part of a country
(e.g., a state, province or city). These sub-national studies include those with a mixture of urban and
rural (such as Flanders, Belgium (Flemish Food Supply Chain Platform for Food Loss, 2017)) and
studies exclusively in an urban area (such as Beirut, Lebanon (Chalak et al., 2019)). In these cases,
applying the per capita waste figures for each sector to the population of the whole country would
assume comparability between regions and that rural and urban waste generation are comparable,
large assumptions which are likely to be inaccurate. Very few studies focused on rural waste,
meaning it was not possible to form an estimate of the variation between urban and rural waste (for
one example of a study including both rural and urban households, see JICA (2015) in Gujranwala,
Pakistan).

As a result, all studies which were at a sub-national level were classified as Medium confidence with
regard to an estimate of national waste, regardless of whether they met the methodological criteria
for each sector (2.2.3.2, 2.2.3.3, 2.2.3.4) Where a sub-national estimate was identified alongside a
national estimate for the same sector, the national estimate was prioritized unless there was some
methodological reason to exclude it.

All studies are, to some degree, local in their sampling. When a study was in a specific locality but
the authors described this as being representative of the wider country and the authors weighted
their results by national distributions (of income, household size etc.), this was considered a
‘Nationwide’ rather than ‘Sub-national’ study, and therefore could be considered a High confidence
estimate (Hanssen et al. (2016) in Norway is a notable example here).

2.2.3.2 Household
For the Household sector, we have higher confidence in studies which involve the direct weighing
and measurement of food waste by an external researcher. This includes waste compositional
analyses, direct weighing of food-only waste streams and papers which combine waste
compositional analysis with other data for scaling purposes.

Within studies which directly weighed waste, sample size was used as a further determinant of our
confidence in the estimate. The figure of 700 household ‘waste-days’ (households sampled per day *
number of days sampled) was used as a cut-off point. Above this, nationwide studies were
considered High confidence, all papers under this were Medium confidence. In some papers, the
duration of sampling was not specified (see Grover & Singh (2014) in Dehradun, India, for example).
This ambiguity meant it was considered prudent to provisionally classify these as Medium confidence
unless more information were to become available. Choosing a boundary to classify studies is an
imperfect science and there is not a single answer as to whether a larger, more time limited or
smaller, longer sample is preferable for estimating food waste. The 700 waste-day figure was chosen
because it equates to 100 households sampled for a week-long period, which was a common
sampling approach and seems a reasonable benchmark for a sufficient sample within resource
constraints.

Other methodologies required adjustment to be comparable, namely food waste diaries or those
which measured only edible food waste (see 2.2.2.2 and 2.2.2.3). As a result of the uncertainty
stemming from these adjustments, the final waste estimates we use were considered Medium
confidence.

Additionally, a number of papers referred to statistics or figures without presenting methodological

detail or from sources which we were unable to trace further than the secondary reference. In many
cases this was due to unclear referencing or citing papers which could not be found based on online
searches. If unfindable, but referenced in a publication which was peer reviewed, by a reputable
organisation or governmental publication, or a reference of a governmental publication which could
not be found, these datapoints were included. These papers were classified as Medium confidence
based on the uncertainty stemming from being unable to view the primary material. In a project
with additional time or resource these data sources could potentially be found, providing enough
information to reclassify the data.

2.2.3.3 Food Service

The Food Service sector is a notably problematic sector for the generation of High confidence
estimates of the entire sector. Many studies provide a robust measurement of a single
establishment or subsector of establishments (such as hotels, or university canteens) but adequate
collation and scaling of a range of subsectors is needed to form a nationwide estimate. As a result,
the overall level of confidence is lower than in the Household estimates.

We judged ourselves to have High confidence in waste audit studies which met two criteria:

• Sufficient sample size, auditing waste in at least twenty establishments

• Coverage of establishments in both the commercial (such as hotels and restaurants) and
non-commercial sector (such as schools and hospitals)

Many authors identified that chefs and managers were resistant or openly hostile to the prospect of
independent waste audits. In addition, some commercial bodies (particularly larger restaurant
chains or catering providers) may already measure their waste. As a result, surveys of businesses or
chefs are often employed. Surveys of chefs with over 100 respondents or carried out by an
authoritative trade or governmental body, and covering both commercial and non-commercial, were
also considered High confidence (notably, Danish Environmental Protection Agency (2014) from
Denmark). It should be noted that there is insufficient detail in papers to say with confidence that
waste was directly measured by Food Service organisations prior to responding to surveys, or indeed
to submitting their data to governmental auditors. Given the commercial imperative to measure
waste, but also the difficulty in initiating researcher-led audits, this uncertainty was considered
acceptable.
Estimates in which we only have Medium confidence relate to those which had any of the following
limitations:

• Only measured edible waste and therefore required adjustment (as per 2.2.2.2)
• Were referenced in secondary peer reviewed or governmental publications but with an
original source we were unable to trace or access
• Cover Food Service establishments in either commercial or non-commercial sectors only

The inclusion of this third category of paper, which represents an ‘incomplete Food Service’
estimate, means there is a downward bias to the results leading to substantial underestimation and
that actual waste across the Food Service sector is likely to be significantly higher. Some of the
challenges with Food Service measurement are discussed in Box 4 in the main FWI report.

The Food Service estimates have big limitations for three reasons:

Firstly, looking at waste in per capita terms may not be the most suitable metric for this sector.

Secondly, the sheer breadth of the out of home environments in which food waste could be
generated creates problems for quantification. Section 3.2.1 in the main FWI report demonstrates
the wide range of establishments which could be considered under the Food Service sector.
Measuring waste in all these locations is practically very difficult, and the relative scale of each
subsector will vary significantly based on the national context. This leads to an inconsistency in
scopes: to our knowledge, only the US (U.S. Environmental Protection Agency, 2020) and UK (WRAP,
2020) estimates include waste in sports stadia. Balancing an accurate estimate of out of home waste
with the limitations of practicality and resources remains a challenge.

Thirdly, food waste going down the sewer is inconsistently measured. In some settings, this could be
considerable.

2.2.3.4 Retail
Retail, like Food Service, has the problem of being considered commercially sensitive data, making it
more difficult for researchers to carry out audits or access existing records which may be carried out
internally. Whilst some supermarkets are publishing their data, a sufficient number of supermarkets
in any given country needs to do so for this to give insight into national waste.

The inconsistency with which sample information was provided meant that it could not be used to
form an assessment of confidence in the estimate. Instead, differences in methodology were
grouped. The High confidence estimates refer to those in which a waste audit was carried out by or
with the assistance of external researchers, whether weighing or using supermarket scanning
systems, and those estimates which involved the disclosure of internally collated supermarket data
to a relevant body, whether governmental surveys (for Japan, see Food Industry Policy Office
(2017)), industry agreements (for the UK, see WRAP (2020)) or other forms of public disclosure.

The estimates which were judged Medium confidence included studies with less transparency or
potentially less robust data, including any of the following limitations:

• Interviews with supermarket representatives where it is unclear whether the estimates

provided came from direct measurement within the retail establishment
• Estimates referenced in secondary peer reviewed or governmental publications but with an
original source we were unable to trace or access
• Estimates which only measured edible waste and therefore required adjustment, as per
2.2.2.2.
2.2.4 Rejected estimates
Below is a brief description of the two primary categories of papers which were narrowly rejected
but could be applicable in other scenarios for forming very rough, ‘order of magnitude’ estimates of
food waste.

2.2.4.1 MSW Papers

There are many papers which document waste compositional analyses which we were unfortunately
unable to consider here due to the sectoral uncertainty around them. Papers which analysed the
MSW of a geographical area often disaggregate food from other organic and biological waste,
however, the uncertainty of the origin source of the waste means it could not be said with any
certainty what was being measured.

MSW will typically be dominated by household waste, but other wastes from litter bins on the
streets, commercial waste from small businesses including restaurants, retailers and street vendors
may make their way into the MSW. Furthermore, not all households or businesses will necessarily
have access to MSW collection rounds: their waste may be processed through informal or illegal
routes. As a result, papers which analyse MSW without disaggregating the source are difficult to use
in the current study. This includes information about waste samples from landfills or waste transfer
stations.

There were a few MSW-based papers that provided usable estimates. These were typically when a
residential solid waste specific estimate was provided (i.e., disaggregation of the total MSW
estimates). To provide an example: Denafas et al. (2014)’s waste compositional analyses in four East
European cities had three MSW estimates which were unusable and one residential solid waste
which was usable. For Kaunas, St Petersburg and Boryspil, the methodology describes taking a waste
sample from a transfer station or landfill. This sectoral uncertainty means it could not be used. By
contrast, the sample taken for Kutaisi in the same study was specifically a sample from residential
areas. By virtue of being residential only, it can be used as an estimate for household. This example
demonstrates how the specificities of method and where the sample was taken could be the
difference between inclusion and exclusion.

Some MSW papers claimed to be looking at the household share of MSW, without clarifying how
exactly that was determined (see Zhang et al. (2018), for example). In these cases, the claims of the
researchers have been trusted and they have been codified as Household estimates, although the
uncertainty means they are classified as Medium confidence (see 2.2.3).

By not including MSW papers, we are not able to use the insights provided by another big source of
waste data: the World Bank’s ‘What a Waste’ dataset (Silpa Kaza et al., 2018). This was searched
within for possibly relevant papers (see 2.1.2), but for the reasons described in this section, the data
could not be used directly.

2.2.4.2 Organic estimates

There were several papers, particularly in the household sector, which evaluated only the total
organic rather than food waste. The organic fraction could contain a wide range of materials,
including food, garden (green) waste, wood and leather. The relative fractions of these materials
within the total organics will depend on a range of factors, most notably around garden waste,
including: presence of gardens, feeding of domestic animals. climate (affecting the amount of
growth) and whether the geographical area in question provides collection of garden waste (e.g., for
industrial composting or anaerobic digestion).
For studies where there was no disaggregation of total, we considered the possibility of calculating
the approximate amount of food waste from the total organics. This could be achieved by taking the
average percentage of food waste within total organics from studies that did provide this
information and applying this average to those studies that only provided total organics.

Two problems arose in applying this method:

▪ The average percentage of food waste in total organics varied widely between studies: the
lowest value was 24%; the highest 98%. (Mean = 81%; standard deviation 17%). This wide
variation likely reflects the factors affecting garden waste mentioned above and makes
applying an average value to obtain even an approximate estimate of food waste in a county
with an organic-only estimate problematic.
▪ If food waste estimates were to be estimated using this method, they are much more
scattered than those directly measured (even with other adjustments): of the 14 studies
identified that this method could be applied to, three of the resulting estimates of
households food waste would be conspicuously high (greater than 150 kg / person / year)
and another would be less than 1 kg / person / year). This is further illustrated by the
standard deviation: for the organics-based estimates (62 kg / person / year), it would be
around twice that of other estimates used.

Furthermore – and as mentioned in 2.2.1 – terminology used around these concepts is not
standardised. Some studies use the terms ‘kitchen waste’, ‘organic waste’ and ‘food waste’
interchangeably within the same paper.

For these reasons, we deemed that estimates obtained in this way (i.e., applying the average
percentage of organics waste which is food to studies with a total ‘organics’ category) insufficiently
accurate. We believed that it would be slightly more accurate to extrapolate from similar countries
than to use these calculations.

2.2.4.3 Surveys
Household surveys in which a representative of a household is tasked with recalling the waste they
or their household has generated over a period of time were considered too inaccurate and
incomparable with the other measurement methods considered here. As a result, no studies which
only distributed a survey to households were included. (Many studies distribute a survey alongside a
diary or waste compositional analysis.)6

By contrast, surveys of organisations such as manufacturers, managers of retail organisations or

restaurant chefs were considered sufficient for inclusion. Whilst they still have problems typically
tied to underestimation (ADEME (2016), in France, demonstrates this point of substantial chef
underestimation when compared to observation), the commercially sensitive nature of the
information makes direct external observation more difficult. The commercial incentive to reduce
waste means many companies may have internal procedures which would inform survey responses,
making them more accurate than household estimation via surveys. However, the quality of opaque
internal measurement is hard to verify: this problem is true both for researcher surveys and
governmental reporting requirements (see Sections 2.2.3.3-2.2.3.4).

6
(C Cicatiello, 2018; Delley & Brunner, 2018; Giordano et al., 2018; For discussions on survey methodology,
see: van Herpen et al., 2019)
2.2.4.4 Superseded studies
In order to make our estimate as relevant for 2019 as possible, we have used the latest available
estimate of food waste available in each country. As previously mentioned, a few countries have
repeated estimates of food waste to provide a time series (such as the UK (WRAP, 2020) or the
Netherlands(The Netherlands Nutrition Centre Foundation, 2019)). When an estimate for which we
have High confidence was available, this was taken and prior studies were not considered.

In the cases of Medium confidence estimates, if a study was a nationwide, direct repetition of a prior
study for comparison purposes (such as Hungary (Kasza et al., 2020), which uses a diary
methodology) only the most recent estimate was taken. In the cases where multiple Medium
confidence estimates exist but they are not directly comparable to one another (such as Iraq (Al-
Maliky & ElKhayat, 2012; Al-Mas’udi & Al-Haydari, 2015; Al-Rawi & Al-Tayyar, 2013; Sulaymon et al.,
2010; Yasir & Abudi, 2009) which has distinct studies across a range of cities) the average of all of the
relevant datapoints was taken to form the estimate.

As a result of this ‘superseding’ process, many studies are excluded from the calculations due to
their being older than comparable studies. Therefore, the total number of food waste quantification
studies originally considered is much higher than those in the final calculations. Similarly, due to
prioritising High confidence estimates (2.2.3), all Medium confidence estimates in countries and
sectors with High estimates were superseded, meaning a wealth of additional sub-national
estimates and Household diary studies exist beyond the final list of estimates.

2.2.4.5 Other unsuitable studies

Alongside the above categories and papers which were incompatible with the methodological
requirements (2.1.1), a number of studies required closer consideration before being excluded. In
particular, this was the case if studies were more qualitative or centred around a case study with a
very small sample, such as a single restaurant or a handful of households, usually measured for
testing methodologies, or lacking the information necessary to scale the information to a national
estimate. Below is a small sample of excluded studies with the reasons for exclusion, to offer an
insight into studies which fell narrowly under the line of usability:

• A study on household waste from rural Czechia, whilst very detailed and including an
ethnographic study, had a sample of seven households over the period of two weeks each
year across two years. As each year amounted to fewer than 100 ‘waste days’ (number of
households sampled * days of waste generation sampled), this was considered too small a
sample (Sosna et al., 2019).
• Numerous studies, including for example by JICA in Pakistan (JICA, 2015), collect a
reasonable waste sample (ten restaurants and institutions over a week) but the waste is
presented only in terms of the waste per establishment or member of staff. Such
information can be very useful and allow other forms of comparison between countries.
However, for our purposes extensive data on meals consumed out of home, the number of
restaurants in a country etc. would be required to scale the information to national
estimates. As a result, this data was not usable in our estimates.
• A study of a supermarket in Viterbo, Italy, focused on food that is recovered for charitable
purposes. ‘Rescued’ food for human consumption is not considered waste, and food taken
by a charitable organisation is not a proxy for waste generation. Furthermore, this study
looks only at a single establishment and extrapolates based on floor size. This was
considered too small a sample to have confidence in the conclusions at a national level
(Cicatiello et al., 2016).
2.3 Calculations: Quantifying food waste in each country
This section details the calculation methods used to obtain an estimate of food waste for each
country in the world, for each of the three sectors under consideration.

Multiple methods were trialled to explore their appropriateness to meet the objectives of this study.
The method in this section was assessed to be the most accurate and most appropriate given the
nature of the data collected. Other methods are outlined in (2.4), with discussion of why they were
deemed less appropriate to this study.

The method used for Household is different from the other three sectors, so is presented separately.
This reflects the low data coverage for non-household sectors outside the HIC bracket. This data
scarcity means that the estimates for the non-household sectors have low accuracy and therefore
have a confidence level of Very Low. This reflects the substantial assumptions required to obtain
these estimates. They are intended to give an approximate indication of the scale of the problem
where these assumptions hold true. Without more data, we cannot say with confidence whether
these estimates under- or over-state the true scale of the food waste problem.

2.3.1 Household
There are two broad approaches to obtaining a household-food waste estimate for a given country.
This depends on whether a country has an estimate of food waste (classified as either High or
Medium confidence, see 2.2.3.2) or no usable data for quantification purposes.

▪ Countries with data: For countries with a single usable estimate of household food waste,
this is taken as the estimate for that country. When a country has multiple estimates (e.g.
multiple household studies have been undertaken and we have a similar level of confidence
in each), the average (mean) of those estimates is taken. If High confidence estimates
existed for a country and sector, any Medium confidence estimates were removed, so
averaging only happens at the same confidence level. See 2.2.4.4 for detail on when studies
were superseded and when they were grouped. Only nationwide studies are considered
High confidence to reduce possible bias from sub-national studies overrepresenting specific
population groups, although sampling methodologies and in-country variation may still lead
to uncertainty in the results.

▪ Countries without data: For countries without a usable study, we calculate an extrapolation
using data from similar countries. For this calculation, two figures are calculated, and the
average taken:

The average waste (kg / capita / year) for data points from all countries with estimates in the
same income group as the country in question (using World Bank classification)7 and

7
As previously mentioned, income groups refer to World Bank classification, for the 2021 fiscal year.
There are four categories: Low-income countries (LIC), defined as those with Gross National Income
(GNI) per capita of $1,035 or less; lower middle-income economies (LMC), with GNI per capita
between $1,036 and $4,045; upper middle-income economies (UMC) with a GNI per capita between
$4,046 and $12,535; high-income economies (HIC), those with GNI per capita of $12,536 or more.
 The average waste (kg / capita / year) for data points from all countries with estimates in the
same region of the country in question (using UNSD sub-region). For a list of all countries
and regions, see Appendix 4.

These two figures are averaged (i.e., combined with equal weight) to generate an estimate
for the country:

𝐸𝑥𝑡𝑟𝑎𝑝𝑜𝑙𝑎𝑡𝑒𝑑 𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒 = (𝐴𝑣𝑔. 𝑜𝑓 𝑖𝑛𝑐𝑜𝑚𝑒 𝑔𝑟𝑜𝑢𝑝 ∗ 50%) + (𝐴𝑣𝑔. 𝑜𝑓 𝑟𝑒𝑔𝑖𝑜𝑛 ∗ 50%)

All averages are means. If there is no regional average, the income group average alone is used to
inform the extrapolated estimate.

Due to the small number of estimates in Low-Income Countries (LICs), the income group average for
LICs is calculated by averaging the data points for Low-Medium Income Countries (LMCs) and LICs
into a single figure. Table 6 displays the average per capita waste by income group.

Table 6: Average Household food waste, by World-Bank income group

WB Income Group kg / capita / year Number of countries with estimates

HIC 79 28
UMC 78 12
LMC 91 10
LIC 97* 2(+10)
* Due to the small number of LIC studies, the LIC average waste is calculated as the combined average of LMC
and LIC, hence a total of 12 countries informing the average. There is insufficient data to make conclusions
about food waste in low income countries.
Table 7 displays the average per capita waste by region. For a discussion of some of the specific
regions and the papers used to inform the estimates, see Section 2.4 of the main FWI report. Whilst
the calculations are based on the averages presented in
Table 7, the small number of datapoints for many regions and differences in methodology mean that
drawing meaningful comparisons from differences in this table is not possible and should be
avoided.
Table 7: Average Household food waste, by region

kg / capita / Number of countries with

Region year estimates
Southern Asia 66 4
Southern Europe 90 5
Northern Africa n/a 0
Polynesia n/a 0
Sub-Saharan Africa 108 8
Latin America and the
69 4
Caribbean
Western Asia 110 6
Australia and New Zealand 81 2
Western Europe 65 6
Eastern Europe 61 3
Northern America 69 2
South-eastern Asia 82 3
Eastern Asia 64 2
Northern Europe 74 7
Melanesia n/a 0
Micronesia n/a 0
Central Asia n/a 0

Because the income groups typically include more countries than the regions, in nearly all cases the
country has more estimates from similar income level countries than it does for countries within its
region. As a result, by applying the average of the region and income group evenly, each regional
estimate is given more weight than economic group estimates. The extent of the bias towards
regional estimates depends on the number of papers in each category. This is considered justifiable
as there are more likely to be regional similarities in diet and food culture (and thus food waste)
than there are between geographically dispersed countries of similar income.

Countries which are in both the same income group and region are counted twice, once in the
regional and once in the income average. As a result, this ‘double-weights’ the data from these
countries. This is again considered justifiable as the data is coming from countries most alike to that
in question.

We assess our confidence in the extrapolated estimates based on the number of countries which
inform the extrapolation. None of the extrapolations are considered High or Medium confidence
estimates, as these classifications are reserved for countries in which a study was identified.
Extrapolations with low confidence are those which are informed by at least ten countries in total of
which at least five must be countries from the same region. These are based on countries rather
than datapoints, so even if a country in the same region had five studies informing its estimate, this
would only count as one for the purposes of extrapolation. All extrapolations in LICs are Very Low
due to having to use averages largely derived from LMCs. For a summary of confidence ratings, see
Section 2.3.3.

2.3.2 Retail and Food Service

For non-household sectors (i.e., Retail and Food Service), the data coverage geographically and
across income levels is insufficient to fully replicate the approach taken for households. There are
three ways in which country estimates have been made:
 ▪ Countries with data: Similar to the method used for households, if a country has usable
estimate(s) of food waste in that sector, this is taken as the estimate for that country. As
with Household, when a country has multiple estimates, the average (mean) of those
estimates is taken. Where a country has a High confidence estimate for a sector, this is
prioritised and Medium confidence estimates are not included. These are classified as either
High or Medium confidence depending on the method and scope of the study (Sections
2.2.3.3 and 2.2.3.4).

▪ High-Income Countries without data: For HICs, there is sufficient data to extrapolate to
other HICs without data. This extrapolation uses the average (mean) per capita waste of HICs
with data. There is insufficient information in most regions to support the use of regional
estimates in this extrapolation. These estimates are classed as ‘low’ confidence.

▪ Other countries without data: For UMCs, LMCs and LICs the average per capita waste for all
countries with estimates is taken. This amounts to a very rough global average being used
for the extrapolation. These estimates have a Very Low confidence classification.

As previously mentioned, the method for non-HICs is will result in estimates with very low levels of
accuracy. For these countries, the global average used for extrapolation is mainly based on data
from HICs, which may not be suitable proxies, hence the Very Low confidence classification.

Table 8: Average per capita waste by sector, HIC and all countries combined

Food Service Retail

Average (kg / No. of countries Average (kg / No. of countries
capita/ year) with estimates capita / year) with estimates
HIC 26 18 13 20
Global 28 23 16 23

2.3.3 Calculating Confidence per sector

Four confidence brackets are applied. High and Medium correspond to when a country has an
existing estimate, with classification of that estimate corresponding to the boundaries set out in
section 2.2.3. Low and Very Low confidence are calculated differently for household and non-
household sectors, as outlined in Sections 2.3.1 and 2.3.2 respectively. This categorisation is
summarised in the
Table 9.
Table 9: Description of confidence classification in this study

When the classification has been used for… Approximate

… Food Service and confidence Suitable for tracking?
… Household interval
Retail
Often in range
±10-20%. See
High If there is a country-specific high confidence
specific study for Highly likely
Confidence estimate
value or data to
calculate CI
Possibly for larger
Often in range
changes in FW, although
If there is a country-specific medium ±20-50%. See
Medium study may have potential
confidence estimate and no high confidence specific study for
Confidence for higher accuracy and /
estimate value or data to
or comparability with
calculate CI
other countries
Extrapolation from Extrapolation with
No – but may provide
estimates from at least sufficient estimates
Low approximate estimate to
10 similar countries, of in income Around ±50%
Confidence inform FW-prevention
which at least five are in classification (i.e.,
strategy
the same region for HIC countries)
All others: Extrapolation
No – but may provide
from few than 10 All others:
Very Low very approximate
estimates or fewer than extrapolation for At least ±50%
Confidence estimate to inform FW-
five from the same non-HIC countries
prevention strategy
region

The confidence rating in this report is not a judgement on the quality of the study undertaken. It is
an assessment – based on our understanding of the study – of how robust the estimate of food
waste is for tracking food waste in the given country. In many cases, this was not an aim of the
original study. Hence many good studies will be classified at a Medium confidence level (or even
excluded from consideration altogether) as the aims of the paper did not include national food-
waste tracking.
2.4 Additional extrapolation methods considered for Household sector
Several different approaches were trialled for obtaining global estimates of food waste in each
sector. The most appropriate approach given the data is described in 2.3, with results in Section
2.4.3 of the main FWI report. However, those approaches trialled but rejected are listed below.

Firstly, it was important to note that had significant correlations been observed between levels of
food waste and potential explanatory variables for which there is good national coverage (such as
GDP per capita), that this relationship could be used to build a regression model, and extrapolation
to countries without data be informed by this model. However, none of the sectors had a discernible
relationship with a nation’s GDP / capita. In fact, Household waste, which might be expected to be
the one with the clearest association, had a very consistent clustering of results around the 60-100
kg/capita/year mark, regardless of GDP. The most notable change across the range of GDP per capita
was the spread of results: whilst the averages stayed largely consistent, the upper extremes varied
significantly. This can be viewed in Figure 5 in the main FWI report.

As a result, a model built from the correlation between these two figures was not considered
appropriate and more general approach were considered. Example values are given for the
Household sector.

Method 1: simple global mean (kg / capita / year) used for all countries, irrespective of whether they
had data. This simplistic approach assumes that average global data will be more appropriate for a
country, even if that country has a relevant study. As this assumption was known to be large, this
approach was more to provide a check to other methods. (Household estimate using method 1 =
635 million tonnes.)

Method 2: This approach repeats Method 1 but derives a median average rather than a mean in
order to quickly see the effect of outliers and their effect on the estimate. Like Method 1, this
applies the average to all countries, including those which already have datapoints, which are
replaced by the average. Again, this method was never intended as a serious ‘contender’ for use.
(597 million tonnes).

Method 3: This approach uses a country’s data where it is present. Where a country does not have
data, it uses a global average (mean), as in Method 1. All countries are scaled by their 2019
population figures. For households, this provided a similar estimate to the method eventually used,
illustrating the similarity in per capita food waste figures in countries of different income levels and
regions. (560 million tonnes.)

Method 4: As for method 3, but using a median average for countries without their own data. For
the household sector, as there were a few high values (potential outliers), this estimate is slightly
lower than method 3. (551 million tonnes).

Method 5: As for method 3, but the mean average applied to countries without data is calculated
from the dataset with high values (potential outliers) removed. For households, the threshold
applied was 120 kg / capita / year, which removed data from six of the 52 countries with data. The
resultant estimate was similar to method 4, and slightly lower than method 3 (544 million tonnes.)

Method 6: A forerunner to the method eventually adopted for household. For countries without
data, a combination of global average, regional average and average from countries in the UN
classification of Developed or Developing were used8. These three averages were then combined in
proportions depending on the number of data points that each average was calculated from.

The proportions were:

▪ The global average had a fixed weight of 0.33

▪ The regional average from the table below
▪ The Developed/Developing average makes up the rest of the weighting

Share of countries with estimate Regional Average Confidence Regional Confidence Weight
0 None 0
>0% Low 0.33
33% Medium 0.50
50% High 0.67

For each country, the weighted averages are summed to generate an estimate. This is summarised
below:

𝐹𝑜𝑜𝑑 𝑤𝑎𝑠𝑡𝑒 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒

= (𝑅𝑒𝑔. 𝐴𝑣𝑔. ∗ 𝑅𝑒𝑔. 𝑊𝑒𝑖𝑔ℎ𝑡) + (𝐷𝑣𝑙𝑝𝑑𝐷𝑣𝑙𝑝𝑛𝑔𝐴𝑣𝑔 ∗ 𝐷𝑣𝑙𝑝𝑑𝐷𝑣𝑙𝑝𝑛𝑔𝑊𝑒𝑖𝑔ℎ𝑡)
+ (𝐺𝑙𝑜𝑏𝑎𝑙𝐴𝑣𝑔 ∗ 𝐺𝑙𝑜𝑏𝑎𝑙𝑊𝑒𝑖𝑔ℎ𝑡 (0.33))
This method gives a value of 567 million tonnes. This indicates the similarity of measured values of
food waste between income groups and regions.

The value of food waste used for this report (569 million tonnes, method 7, Section 2.3.1) is very
similar to method 6 (567 million tonnes). Given the similarity between the results of the seven
different methods used for estimating a global average, the exact nature of the calculations has only
a small impact on the results. Therefore, we have focused on a method that is likely to yield the
most accurate results for each country (as presented in Section 2.3).

8
Later analysis within this project suggested that the four groups of income level, as per the World
Bank classification, were more appropriate. They provided more insight into the coverage of food
waste data and amounts of food waste generated.
 3 Appendix: Available estimates / datapoints used for level 1
modelling
This section contains tables with the datapoints used with the Level 1 modelling. This includes:

▪ The country and any other geographical boundaries of the study,

▪ A description of the method used in the study,
▪ The amount of food waste (normalised and, where necessary, adjusted for differences in
scope or methodology)
▪ The study author and year to correspond with the bibliography
▪ The confidence in the estimate for the purposes of tracking food waste over time (see
Section 2.3.3 for more details)

It is important to note that the confidence rating in this report is not a judgement on the quality of
the study undertaken, it is an assessment – based on our understanding of the study – of how robust
the estimate of food waste is for tracking food waste in the given country. In many cases, this was
not an aim of the original study. Hence many good studies will be classified at a ‘Medium’
confidence level (or even excluded from consideration altogether) as the aims of the paper did not
include national food waste tracking.

3.1 Available estimates / datapoints: Household

Table 10: Available datapoints for Household Sector, by country

Country Subnational Adjusted

study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
This paper represents the national baseline estimate for food waste commissioned by
the Australian government in order to be the basis for measurement and reporting of
progress on Australia's National Food Waste Strategy 2017. To generate the estimate
Australia 101.70 (Arcadis, 2019)
for household, they combined state government data and modelling forward
previously aggregated state government estimates which informed the National Waste High
Report (2016). Confidence
This presentation (in English) by the Environment Agency Austria references a range of
other studies in German, summarising the current state of knowledge. These came
from Waste Compositional Analyses, although precise details are not in the
(Environment
presentation (apart from for one region, Lower Austria). For the one paper where
Austria 39.00 Agency
methodological aspects are given, it refers to sampling from three areas: rural (without
Austria, 2017)
centre); rural (with centre); urban, with 796 samples. Whilst more methodology on the
other primary studies would have made this clearer, assuming similar methodologies High
are used in the other cases suggests a sufficiently large and diverse sample. Confidence
The source link is to a newspaper article which refers to a report by the Center for
Waste Management. The original Center for Waste Management Report could not be
found. However, the infographic (clearly copied from the original report) and the
(Alayam,
Bahrain article make clear that a waste compositional analysis was undertaken, referring to 131.71
2018)
'huge quantities of household waste collected from the various region of Bahrain' being
sorted. The inability to find the source paper means we cannot have high confidence in Medium
the results. Confidence
55 households in five different socioeconomic groups across three different areas had
their waste sampled daily, using plastic bags provided to them. It was unclear for how (Salam et al.,
73.63
long the sampling ran for each household. This small sample size and unknown 2012) Medium
Bangladesh Chittagong duration means we cannot have high confidence. Confidence
75 households across five socioeconomic groups in the Rahman Nagar Residential Area
(Sujauddin et
had their waste sampled. The length of sampling is unknown. The small sample with 56.61 Medium
al., 2008)
Chittagong unknown duration means we cannot have high confidence in the results. Confidence
Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
This paper combines a number of different data sources. An existing OVAM study was
used for waste composition analysis in residual waste. This was combined with (Flemish Food
information from other surveys self-reporting different disposal methods to capture Supply Chain
Belgium the share going to home compost, pets etc. and data from the Netherlands was used to 49.92 Platform for
inform the disposal to sink and toilet. The share which goes to feeding animals (28%) Food Loss,
was removed as it is not considered waste, and the amount going to sinks/toilets (3%) 2017) Medium
Flanders was removed to improve comparability. Confidence
San Ignacio / 174 households across three socioeconomic groups had their waste sampled, with at 95.05 Medium
Santa Elena least 100kg collected each sampling day. Measurement was for 8 days. Confidence
132 households across three socioeconomic groups had their waste sampled, with at Medium
45.23 (Inter-
Caye Caulker least 100kg collected each sampling day. Measurement was for 8 days. American Confidence
Belize
169 households across three socioeconomic groups had their waste sampled, with at Development Medium
least 100kg collected each sampling day. Measurement was for 8 days. 36.20 Bank, 2011) Confidence
San Pedro
183 households across three socioeconomic groups had their waste sampled, with at Medium
least 100kg collected each sampling day. Measurement was for 8 days. 34.21
Belize City Confidence
686 households participated in "a food diary with photo analysis of wasted food", with
(Araujo et al.,
Brazil more households also participating in a questionnaire about consumption habits. As a 59.60 Medium
2018)
diary result, it has been scaled to adjust for the measurement bias. Confidence
56 different waste compositional analyses studies are analysed and averaged to form a (Environment
national average. The studies analysed involved a mixture of kerbside analysis and at and Climate
Canada 78.54
sorting facilities. The share which is food waste has been multiplied by the total Change High
residential waste to form a food waste estimate. Canada, 2019) Confidence
The paper references Zhang et al. (2010) but does not provide sufficient (Gao et al., Medium
methodological detail to offer high confidence. 25.57
Beijing 2013) Confidence
140 households participated in a compositional analysis. This involved their waste
being collected each day for a week and was repeated in each season. They also (Gu et al.,
67.31
completed a survey. The household sizes are considered representative of the wider 2015) Medium
Suzhou city. Confidence
207 houses across 21 villages in three prefectures (Jinan, Weifang, Dezhou) studied,
tracked for 9 meals across 3 days. The weight of food recorded before cooking and
after a meal's disposal to form 'meal waste' figure, but also measure 'non meal' waste
21.20 (Li et al., 2021)
(such as removal of mouldy stored items) once a day. It was a form of diary
documenting this only looking at edible waste, so it is adjusted for both diary bias and Medium
Shandong the inedible share of waste. Confidence
The paper cites the Hong Kong Environment Bureau's official statistics. It is assumed to
(Lo & Woon,
be from Waste Compositional Analysis but is not made explicit, nor other details of the 101.46 Medium
China 2016)
Hong Kong method (such as sample). Confidence
113 households across six districts in Beijing city had their waste collected and analysed (Qu et al., Medium
daily for a period of 10 days. 58.98
Beijing 2009) Confidence
Paper quantifying the carbon, water and ecological footprints of food consumed and
food wasted in Chinese households. Uses the China Health and Nutrition Survey (CHNS) (Song et al.,
22.92
2011 database, with data from 12,850 households across 2004, 2006 and 2009 using a 2015) Medium
diary methodology over three days. Confidence
The household estimate uses a huge range of local MSW figures and studies estimating
the share of household food waste in the entire MSW. 196 samples obtained from the
literature across 2001-2016. (Supplementary Info, Table S21-2). All literature values
(Zhang et al.,
cited reported the value Household Food Waste in MSW, though it is unclear how 150.00
2020)
exactly it was disaggregated if samples were taken at landfilling or transport sites. The
Urban China per capita figure only applies to the urban population as this was where the study was Medium
Total concentrated. Confidence
The paper cites 3259 samples, although it's unclear if this is referring to households or
individuals, taken across a single twenty-four-hour period, across 19 localities and six
Colombia 70.43 (JICA, 2013)
socioeconomic categories. Whilst the length of sample is small, the size was considered Medium
Bogota to compensate for this. Confidence
1474 households were randomly selected across five Danish municipalities in which
(Edjabou et
Denmark there is no source segregation of household food waste. Depending on their collection 83.20 High
al., 2016)
schedule, their waste from one or two weeks was collected and sorted. Confidence
Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
1600 houses sampled across four municipalities. It specifically looks at and (Danish
disaggregates between single family homes and blocks of flats in each of those Environmental
79.47
municipalities. Protection High
Agency, 2018) Confidence
100 households of various income levels and living arrangements took part in a diary (Moora,
Estonia study alongside a questionnaire survey. It has been adjusted to account for diary bias. 77.51 Evelin, et al., Medium
2015) Confidence
Bags distributed to 92 'residential households' for waste collection and sorting every
day. From this waste compositional data, food waste can be derived. It is unclear for
how long this compositional analysis took place. Survey in Laga Tafo Laga Dadi
(sometimes written Legetafo Legedadi) town, Oromia, a small area on the outskirts of
Ethiopia Addis Ababa. 92.14 (Assefa, 2017)
Laga Tafo Note: different 'residential' groups are included in the paper, including 'real estate
Laga Dadi residential' and 'ropack village residential'. Due to some confusion over the
town, terminology and these types having very high bone waste, only 'residential households' Medium
Oromia were considered here. Confidence
References three different papers: from Katajajuuri et al. (2014), Silvenoinen et al.
(2014) and HSY (2012). The Katajajuuri paper is also a separate data point in this
database. The first two (diary studies) only estimate avoidable waste (120,000 tonnes),
the latter (appears to be a waste compositional analysis) was used to estimate
avoidable waste (230,000 tonnes). This was done to make them more comparable to (Stenmarck et
63.60
other studies of all food waste. Since this adjusted total food waste estimate comes al., 2016)
from Finland-specific data, it was judged that leaving it as a separate data point to
Finland compare and average with our generic edible and diary share adjusted figure would be
prudent. The adjusted per capita figures are highly comparable to one another, giving Medium
some confidence in our estimate. Confidence
380 households participating in diary study for two weeks. Only looking at edible food.
It is adjusted for diary bias and inedible share. The end result is highly comparable with (Katajajuuri et
67.25
the combined estimate cited in Stenmarck et al. (2016), which used Helsinki al., 2014) Medium
metropolitan area waste compositional data to adjust the original figure. Confidence
Paper in French, with summary available in English and details in Caldeira et al. (2019).
50 households in a representative sample of the French population were measured
(ADEME,
France across 7 days using an online diary survey in which they documented waste each day 84.79
2016)
and uploaded pictures. Only measuring edible food. This has been adjusted to account Medium
for both diary bias and the inedible waste share. Confidence
Between 400-600kg of residual waste from residential areas taken and sorted. Done
each month for a period of a year. Compositional information combined with MSW
(Denafas et al.,
Georgia data to understand total waste. The paper does specify these samples came from 100.97
2014)
residential areas, but they were collected from waste trucks rather than homes directly Medium
Kutaisi leading to some increased uncertainty. Confidence
This is an English summary paper of a study in German, which means some of the
details of the methodology are unclear. They claim to use "the best available data at
the time of the study", which will include papers in German. Figure S.1 shows the
admissible measurement methods and those which were applied for each sector,
which is in line with the methods we are targeting here. However, the English summary
(Schmidt et al.,
Germany does not include detail on the specific sources used so it is unclear how many of each 75.00
2019)
method there are. For household they use Direct Measurement; Waste Composition
Analysis; Records.
This paper forms the 2015 baseline for food waste against which the national strategy
for reducing food waste will be measured, it is therefore the preferred figure for High
Germany. Confidence
1014 households representing 6083 people were randomly selected in 10 different
districts across the country across three socioeconomic groups (low, medium, high).
The households were provided with two bags, one for biodegradable waste and one for
other waste, and were taught how to separate accordingly. Employed sorters then (Miezah et al.,
Ghana 84.01
collected and did further sorting and disaggregation between every two days and twice 2015)
a week for a period of five weeks, including sorting the biodegradable waste into a food
subcategory. The per capita figure taken is the average across the socioeconomic High
groups provided in the paper. Confidence
Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
252 households in urban and semi-urban areas of Athens, Heraklion and Chania were
involved in a diary study, asked to weigh and record their waste (both avoidable and (Abeliotis et
Greece 141.69
unavoidable) for a period of two weeks. This has been adjusted to account for diary al., 2015) Medium
bias. Confidence
165 households measured their waste in a diary for a week. This is a repeat of a 2016
(Kasza et al.,
Hungary study to identify changes in household waste. The definitions used are designed to be 93.83 Medium
2020)
compliant with the EU FUSIONS project. It has been adjusted to account for diary bias. Confidence
144 households across three different socioeconomic groups in Dehradun city were
given a large bag in which to dispose their waste, which was then sorted and classified. (Grover &
72.99
It is unclear for how long the survey took place, so is assumed to have not met the '700 Singh, 2014) Medium
Dehradun waste day' baseline and we therefore cannot have high confidence in the estimate. Confidence

India Rajam, 25 households from 5 different segments of Rajam town were given two bags; one for
wet and one for dry waste, collected each day. Segregated their waste for seven (Ramakrishna,
Andhra 57.84 Medium
consecutive days, which was then taken for sorting. 2016)
Pradesh Confidence
144 households from 11 major blocks of Dehradun city provided with waste bags in (Suthar & Medium
which to put their waste from a 24-hour period, which was then sorted and classified. 20.13
Dehradun Singh, 2015) Confidence
100 households in Surabaya were provided with bags in which to put all of their daily
(Dhokhikah et
Indonesia waste for a period of 8 consecutive days. This was then collected and sorted, including 77.37 Medium
al., 2015)
Surabaya into a separate food waste category. Confidence
20 families in Baghdad, a mixture of family sizes and income levels, weighed their
weekly purchases (this seems to have excluded meat) and kitchen waste on kitchen
scales and documented them for a period of 8 months. This has been adjusted to
(Al-Maliky &
account for diary bias. There is some uncertainty about the scope as the term kitchen
75.07 ElKhayat,
waste and leftovers are used interchangeably: this could possibly be looking at edible
2012)
waste only. Adjusting it as such would lead to an extreme food waste estimate, so this
has been assumed not to be the case, but this uncertainty prevents us from having high Medium
Baghdad confidence in the results. Confidence
60 households, 10 from each sector of Mosul were given plastic bags and told to
deposit their waste from a 24-hour period into it. It is unclear if this was repeated for
(Al-Rawi & Al-
individual houses or for how many days, though the paper does say that the study 84.66
Tayyar, 2013)
period was between February and July which would suggest it was repeated for Medium
Iraq Mosul households for some duration. A total of 1680 solid waste samples were collected. Confidence
70 households in Karbala distributed plastic bags in which to put their waste from a 24- (Al-Mas’udi &
hour period. Done once a month for three months in winter and three months in 141.61 Al-Haydari, Medium
Karbala summer. 2015) Confidence
80 households across three income groups in Al-Kut had their waste collected daily for
a period of one week, which was repeated one week per month for seven months.
Whilst this is a large sample, there remains some uncertainty around definitions as to (Sulaymon et
137.53
whether food or organic waste was measured, which could explain the quite al., 2010)
substantial waste generation. As a result, we cannot have high confidence in the Medium
Al-Kut City estimate. Confidence
65 households representing 417 people across three income groups in Nassiriya were
(Yasir & Abudi,
randomly selected. Distributed plastic bags in which to put waste which were collected 163.33 Medium
2009)
Nassiriya daily and replaced over a period of seven months. Confidence
Cited as being from the Irish Environmental Protection Agency (2015). The original
source and weight estimates were not found based on the bibliography information or
(Stenmarck et
Ireland direct searches. However, it was judged by Stenmarck et al. (2016) to be 'data of 54.70
al., 2016)
sufficient quality'. The inability to find the source paper means we cannot have high Medium
confidence. Confidence
192 households across three neighbourhoods in Eastern Haifa, primarily middle-class
households, provided with waste bags which were collected daily for the period of one (Elimelech et
Israel 94.18
week. Due to being a study within a specific unrepresentative area, we only have al., 2018) Medium
Haifa medium confidence. Confidence
 Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
The paper describes the figure as coming from "A comprehensive value chain model for
various food production and consumption stages was designed to assess food waste
and the potential for food rescue in Israel. The model is based on a bottom-up
approach, and includes analysis of data relevant to agricultural production, import,
export, industry, distribution and a sample of consumption patterns of 50 different
types of food." [...] "For each type of food, the volume of input and output was
measured in terms of gross agricultural product and loss rate for every stage of the
(Leket Israel,
value chain in the food production, distribution and consumption process." [...] "This 104.99
2019)
data is indicative and intended to serve as the basis for public debate, and for further
research and study".
As a result, it is not that clear exactly the calculation that has taken place nor the
original data sources, other than that they are 'bottom up'. Leket is a primary food
rescue organisation in Israel and this paper was referred to us by contacts in the region.
The lack of clarity on the methodological details means we cannot have high Medium
confidence in the estimate. Confidence
388 families completed a food diary in which they recorded food disposed over the
course of one week. The households were chosen by random stratified sampling and (Giordano et
Italy 67.05
were considered representative in terms of region and household size. As a diary al., 2019) Medium
estimate, it has been adjusted to account for methodological bias. Confidence
This presentation (in English) by the Food Industrial Policy Office presents statistics
from the MAFF statistical survey, estimating waste from each of the sectors: Household
(Food Industry
(solid waste); Restaurant (assumed to mean all Food Service) and Retail. Wholesale
Japan 64.32 Policy Office,
estimated separately, not included. Estimates are for the 2014 financial year. The
2017)
survey methodology is not presented. As a result of this methodological uncertainty, it Medium
is considered 'medium confidence' until more information is received. Confidence
150 households were sampled across five income groups (High, Middle, Low-Middle,
Low, Slum), which are grouped in Table 2.2.7 into three residential groups (High,
Middle, Low) with a subset of those sampled for composition. Total of 8 days collected 99.92 (JICA, 2010)
but the first one was discounted as not representing daily generation, so 7 days of Medium
Kenya Nairobi sample. A subset of this waste was then sorted and classified. Confidence
90 households across three income areas (high, middle, low) received plastic bags for
disposing daily waste. Total of 8 days collected but the first one was discounted as not (Takeuchi,
98.56
representing daily generation, so 7 days of sample. This waste was then sorted and 2019) Medium
Nairobi classified. Confidence
250 respondents completed a seven-day food waste diary in Beirut, alongside an
attitudinal survey which a total of 500 people completed. Neighbourhoods were
(Chalak et al.,
Lebanon selected to be representative of the wider population, but the sampling of households 104.66
2019)
within neighbourhoods was systematic random sampling. As a diary study, it has been Medium
Beirut adjusted for methodological bias. Confidence
This figure comes from a review paper which references a primary source (in German)
which is mentioned as being a waste composition study, although further (Caldeira et
90.57
methodological details are not provided. This estimate has been complemented by a al., 2019) Medium
Luxembour more recent study in the same country. Confidence
g Combination of waste statistics and a waste composition analysis undertaken in 2018- (Luxembourg
19, alongside other secondary and tertiary data. Further details are not known, but it is Environment
88.50
published by the Luxembourg Environment Administration. Ministry, Medium
2020) Confidence
Table 1 cites 'MHLG' (Ministry of Housing and Local Government) 2011, estimating food
(Jereme et al.,
waste generation by source. This was not findable by the bibliography nor through a 111.54 Medium
2013)
direct internet search. As a result, we cannot have high confidence in the estimate. Confidence
282 households were sampled across four neighbourhoods, which represent a mixture
of different housing types (terraced housing, bungalows, flats). All in an area of
Malaysia Selangor described as a typical suburban area in the Kuala Lumpur area. Waste from a
single day sampled in each area, sampling from the normal disposal routine rather than
(Watanabe,
asking households to dispose of their waste differently. Panel 3 has breakdown of food 71.35
2012)
into 'Unused food' (7.71% total household waste), 'General kitchen waste' (24.83%
total household waste), 'big fruit peels' (10.32% total waste).
Bandar Baru Although this has a large sample, it is geographically restricted to one area so can only Medium
Bangi have medium confidence when used for the whole of Malaysia. Confidence
 Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
The figure comes from a review paper which details the methodology of the original
(Caldeira et
Malta study. 700 households were randomly sampled for one week in each of July, October 129.00 High
al., 2019)
and April as part of a waste compositional analysis. Confidence
This figure combines a number of sources, detailed in Appendix 5 of the report. Studies
were identified in 3 states and 5 municipalities which directly measured the share of
waste which was food waste at the household level. This is then scaled up using figures
from the urban solid waste, which is primarily but not exclusively household waste: (Kemper et al.,
Mexico 93.90
some small businesses and some larger ones (operating illegally) dispose of waste in 2019)
the household municipal waste. The scale of non-household contamination is not
known. As a result, it is no more than a medium confidence estimate for household Medium
food waste that likely slightly exaggerates its extent (in urban solid waste). Confidence
The study combines multiple methods: 130 households across 13 municipalities had
(The
waste compositional analyses conducted, in order to determine solid waste; 1000
Netherlands
respondents to a consumer survey in order to determine waste destinations (i.e., how
Netherland Nutrition
much was composted, fed to animals etc.); 1013 respondents to an app used to 50.00
s Centre
document liquid waste. This is the same methodology as was applied in a 2016 study.
Foundation,
Avoidable and unavoidable waste were presented separately and have been summed. High
2019)
Liquid waste has not been included in our figure. Confidence
597 households across six different local authorities had their waste audited. This only (Sunshine
New considers the kerbside domestic waste. Yates
61.00
Zealand Consulting, High
2018) Confidence
100 households covering a total of 334 people were selected by stratified random
(Orhorhoro et
Nigeria sampling, all in the Sapele area. Waste was collected from households after seven days 188.80 Medium
al., 2017)
Sapele and sorted. Confidence
210 households in the Fredrikstad and Hallingdal areas were randomly selected for
(Hanssen et
Norway waste sampling. These areas were considered to be sufficiently representative of the 78.80 High
al., 2016)
wider population for scaling purposes. Confidence
60 urban households across three income groups (high middle and low). Provided with
plastic bags which were collected daily for 8 days, though the first bag was disregarded
for containing more than one day's waste. The sample was repeated across three
seasons to account for variation. Rural households were considered within the study, 87.55
these have been treated as a separate data point. As it is a study specific to a smaller
geographic area, it is considered medium confidence for analysing the whole of Medium
Pakistan Gujranwala Pakistan. (JICA, 2015) Confidence
10 households in rural areas provided with plastic bags in which to deposit waste.
Collected for 8 days but the first day was discounted due to covering more than a day's
waste. The survey was repeated across three different seasons to account for variation. 59.73
The small sample means we cannot have high confidence. Urban households were also Medium
Gujranwala studied but treated as a separate data point. Confidence
21 households, representing 83 people, were audited. None of them were involved in
agricultural production. They were provided with three bags for sorting (bio-waste,
(Steinhoff-
hygienic waste, all other waste) and had waste collected in each of the four seasons. It
Poland 55.94 Wrześniewska,
is unclear for how long during each season the measurement took place. As a result of
2015)
Surrounding small sample size and unknown length, we cannot have high confidence in the Medium
Wroclaw estimate. Confidence
The paper cites what is assumed to be a waste composition analysis by the Higher
School of Economics (which was not found when searched for) and data from Rosstat.
In addition, the shares of waste at each stage are calculations based on data from
Russian Russian Agriculture Ministry (2017). The estimate provides a total food waste estimate (Tiarcenter,
33.38
Federation as well as the amount of waste at each stage of the chain, these have been combined 2019)
to form sector-specific estimates. The inability to trace the original source data and the
lack of transparency on the calculations means we cannot have high confidence in this Medium
estimate. Confidence
90 households surveyed in 3 districts. For each district, 10 households from each
socioeconomic group (low, medium, high). Bags and scales were distributed to the
Rwanda households and they were told to separate food waste and other waste. The 164.36 (Mucyo, 2013)
households weighed this each day for a period of two weeks, but regularly received Medium
Kigali visits from the researchers. Confidence
Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
This study forms the Saudi waste Baseline, conducted by Saudi Grains Organisation
(SAGO). 20,090 samples of domestic consumption were taken across 19 food products
across 13 regions in Saudi Arabia. It is unclear, however, from how many households
these samples arise. These were separated and weighed. This compositional analysis
Saudi
was supplemented by a behavioural study. The Household estimate is the share of 104.88 (SAGO, 2019)
Arabia
waste attributed to 'Consumption'. Additional information and images to supplement
the main study can be found here: https://www.macs-
g20.org/fileadmin/macs/Activities/2020_FLW_WS/4_Session_3_FW_at_HH_level_smal High
l.pdf Confidence
Data from the Slovenian Statistical Office, but the exact methodology is unclear. The
methodological explanation from the same site mentions three different annual (Republic of
surveys on waste collection, generation and recovery/disposal, alongside an ad-hoc Slovenia
32.83
questionnaire of public waste collection services. It is unclear to whom the surveys are Statistical
sent (i.e., to waste generators or to waste collection services) and if it requires Office, 2019) Medium
submission of observed data or some other form of self-reporting. Confidence
Slovenia
Data from the Slovenian Statistical Office, but the exact methodology is unclear. The
methodological explanation from the same site mentions three different annual (Republic of
surveys on waste collection, generation and recovery/disposal, alongside an ad-hoc Slovenia
35.58
questionnaire of public waste collection services. It is unclear to whom the surveys are Statistical
sent (i.e., to waste generators or to waste collection services) and if it requires Office, 2020) Medium
submission of observed data or some other form of self-reporting. Confidence
Richards 554 participants taking place in a face-to-face survey. They were asked to measure and
(Chakona &
Bay, Dundee record their waste for the 48hr before the survey. Survey across three towns of
17.69 Shackleton,
and Richards Bay, Dundee and Harrismith. Mixture of urban, peri-urban and rural Medium
2017)
Harrismith households. Confidence
This paper combines a literature review of waste compositional analyses disaggregated
by income group across three cities (Cape Town, Johannesburg and Rustenburg). These
(Nahman et
are then scaled by the waste generation of those specific income groups nationally. 27.33
al., 2012)
Due to the comparison with other datapoints from South Africa and their large Medium
variation, this was not considered an estimate in which we could have high confidence. Confidence
44,927 households across 74 collection routes in Johannesburg were sampled during a
6-week period, with random grab sub-samples from municipal waste collection trucks
in residential areas, which were then analysed for composition.
The result is particularly low which is notable when compared to other studies in 12.00
nearby countries. This could suggest that some other waste (such as from small
Johannesbur businesses, or illegal dumping) is being collected as part of the household waste Medium
g stream. (Oelofse et al., Confidence
20,439 households across 41 collection routes in Ekurhuleni were sampled during a 6- 2018)
week period, with random grab sub-samples from municipal waste collection trucks in
South residential areas, which were then analysed for composition.
Africa The result is particularly low which is notable when compared to other studies in 8.00
nearby countries. This could suggest that some other waste (such as from small
businesses, or illegal dumping) is being collected as part of the household waste Medium
Ekurhuleni stream. Confidence
123 households across 5 areas of Tshwane Metropolitan Municipality had their food
waste weighed. The food waste was collected separately and weighed on a weekly
basis for a period of 3 weeks. The sample of 123 are out of 133 respondents on a
survey who indicated that they wasted food. Another 77 respondents indicated that
they did not waste food, and were seemingly not asked to weigh their waste. This may
bias the results by only auditing those who self-describe themselves as someone who
wastes food, and not including measurements from much smaller waste generators.
The paper does not present a single waste figure. Instead, it has been derived from (Ramukhwath
133.85
Table 4.9 using the waste generation rate per household, number of people in o, 2016)
household and share of that household size in the sample to get a weighted per capita
estimate (the Sum of [household waste / number of people in household] * [share of
total sample which is this household size] for each household size).
Tshwane The paper does include some disposal method information but not enough to adjust
Metropolita the figures. For example, 14% of respondents claimed they fed food waste to pets, but
n this does not clearly translate to 14% of food waste being fed to animals. As a result, no Medium
Municipality adjustment was carried out. Confidence
Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
The figure comes from a review paper which cites a study produced by the Spanish
Environment Ministry. It involved 4000 households doing an online waste diary survey,
2000 of which documented their shopping and waste with the other half used to 77.65
develop user profile and preferences. This only looks at edible waste so has been Medium
adjusted. (Caldeira et Confidence
Spain
The figure comes from a review paper which cites a study produced by the Spanish al., 2019)
Environment Ministry. It is similar to the other Spanish datapoint but involves a higher
number of households, with 12,000 considered for building the profile of shopping 77.00
habits and 4,000 doing the waste diary. This only looks at edible waste so has been Medium
adjusted. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
118.28
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Jaffna confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
95.37
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Nuwara Eliya confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
94.79
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Kataragama confidence in the estimates, so they are rated 'medium confidence' only. Confidence
Sri Lanka (JICA, 2016)
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
79.31
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
Thamankadu level, such as at the landfill. This and the methodological uncertainty reduce our Medium
wa confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
78.47
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Katunayake confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
74.78
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Moratuwa confidence in the estimates, so they are rated 'medium confidence' only. Confidence
 Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
74.78
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Kesbewa confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
74.78
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
Dehiwala Mt level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Lavinia confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
47.48
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Kurunegala confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The JICA study refers to a range of locally conducted surveys on waste generation units
and waste composition, combined with waste generation rates obtained by SATREPS
(Science and Technology Research Partnership for Sustainable Development) in 2014, a
previous JICA project. The methodological details of the locally outsourced surveys are
20.60
not clear. Although the waste generation rates are captured at a household level, it
appears as though the compositional analysis may have been done at an aggregated
level, such as at the landfill. This and the methodological uncertainty reduce our Medium
Trincomalee confidence in the estimates, so they are rated 'medium confidence' only. Confidence
The estimate comes from Avfall Sverige (Swedish Waste management and Recycling
(Swedish
Association) data on the amounts of separate food waste collection and composition
Environmental
Sweden estimates for mixed municipal waste. Home composting is included in the separate 81.00
Protection
food waste estimate. The waste estimates are adjusted by a figure of 0.80 to account High
Agency, 2014)
for home-like business waste collected by municipalities. Confidence
Household data comes from a combination of data on the composition and weight of
residual and organic recycling schemes from Local Authorities. The share of waste
United which is disposed of via the sewer has been removed from the estimate, though the
Kingdom of waste composted at home has not.
Great
77.00 (WRAP, 2020)
Britain and
Northern
Ireland
High
Confidence
75 households in middle and low-income settlements, primarily in high population
density informal settlements. Provided with waste bags for three different days which
United
Kinondoni were collected and sorted. (Oberlin,
Republic of 119.09
municipality, 2013)
Tanzania
Dar es Medium
Salaam Confidence
US EPA synthesis of food waste consistent with the FLW protocol. Combines waste
(U.S.
United generation factors from other studies with relevant scaling statistics. 12 studies which
Environmental
States of directly measure waste were used to inform residential generation rate, nearly all of 58.83
Protection
America which were waste compositional analyses. 15% going to sewer / wastewater treatment High
Agency, 2020)
has been removed. Confidence
100 households across ten different sampling points were selected. The sample is
considered to be representative of Can Tho City in terms of household size. They had (Thanh et al.,
Viet Nam 85.38
Mekong their waste analysed at two points in time: once in the dry season for a month, once in 2010) Medium
Delta the rainy season for a two-week period. Confidence
Country Subnational Adjusted
study area kg/capita/year Confidence
(if relevant) Methodological Description estimate Source Level
120 households were sampled for the period of one week. They were provided with
(Vetter-
plastic bags in which to put household waste which were collected daily. Satellite
66.94 Gindele et al.,
imagery on the distribution of housing types in Da Nang were used to scale the data Medium
2019)
Da Nang according to those housing types and form an estimate for the city. Confidence
60 households across three areas (distinguished by housing density) sorted their waste
weekly for a period of one month. The households were given plastic containers for (Edema et al.,
Zambia 77.92
different wastes: food, plastics, paper, textile, grass and other wastes. They therefore 2012) Medium
Ndola separated it themselves but did not weigh or estimate it themselves. Confidence
3.2 Available estimates / datapoints: Food Service
Table 11: Available datapoints for Food Service, by country

Identified food service food waste data

Adjusted
Subnational study kg/capita/year Source (study
Country area (if relevant) Methodological Description estimate number) Confidence Level
This paper represents the national baseline
estimate for food waste commissioned by the
Australian government in order to be the basis for
measurement and reporting of progress on
Australia's National Food Waste Strategy 2017. The
figure was calculated through a number of surveys
and published figures for hospitality and food
service (commercial), and a range of small samples
Australia at institutions such as hospitals and universities. 21.68 (Arcadis, 2019)
These were scaled based on relevant figures such
as the number of students and inmates for schools
and prisons respectively. The authors highlight low
confidence in the institutional estimates due to
small samples. However, it represents wide
coverage and the low confidence is "not
considered material given the estimated scale of
food waste produced". High Confidence
50 establishments had waste compositional
analyses carried out (23 canteen kitchens, including
hospitals and nursing homes; 13 accommodation
establishments; 13 restaurants, 1 catering). This
(Environment
represents wide coverage of both commercial and
25.68 Agency Austria,
non-commercial out of home consumption. From
2017)
the compositional analyses, extrapolation was
based on food waste quantities per meals, number
Austria of meals per establishment type and number of
establishment types. High Confidence
The original paper referenced is in German. The
reference explains that 29 hospitality companies
including restaurants, hotels and canteen kitchens (Caldeira et al.,
31.09
recorded their food waste. This therefore has 2019)
coverage across both commercial and non-
commercial dining. High Confidence
The JICA study team conducted waste generation
source surveys in both the dry and wet season. The
exact number of restaurants is not known, only
that 50 non-households were sampled across 8
days in each season (this figure will include other
buildings, such as shops or offices). It was scaled to
the whole of Dhaka based on the Revenue
Department's licensing list. Being incomplete in its
Bangladesh 3.34 (JICA, 2005)
sectoral coverage and being quite a dated estimate
limit its robustness. Note: the JICA paper does also
estimate the waste for hotels, which would be
considered here, however the table defining the
share of food waste from hotels groups it with
shops, which would otherwise be considered retail.
To avoid this overlap they have not been Medium
Dhaka documented. Confidence
 The data comes from a range of sources, many of
which are commercial and therefore not
transparently disclosed. The authors collaborated
with food chain partners (horeca Vlaanderen and
Unie Belgische Catering) and used existing surveys
including Annual Foodservice Monitor and the
(Flemish Food
Integrated Environmental Report. These sources
Supply Chain
Belgium allowed a per person daily waste to be derived, 19.70
Platform for Food
which was then scaled. It covers both hospitality
Loss, 2017)
(commercial) and catering (non-commercial).
Hospitality and catering are presented separately
in the paper, they are summed together here. Their
figures have been scaled for the whole of Belgium,
assuming that Flanders can be considered Medium
Flanders representative of the whole nation. Confidence
This paper aggregates a huge number of what they
term 'catering waste' papers. 47 of them in total
from various areas of China. It's a mixture of
surveys, official statistics, author's calculations etc.
to create data on catering food waste across
different regions. This is then associated to a 51.87
number of correlates which are used to predict
growing food waste in future. The authors worked
with datapoints from a range of years and other
data to form a 2019 estimate which is what we use
East China here. High Confidence
This paper aggregates a huge number of what they
term 'catering waste' papers. 47 of them in total
from various areas of China. It's a mixture of
surveys, official statistics, author's calculations etc.
to create data on catering food waste across
different regions. This is then associated to a 49.31
number of correlates which are used to predict
growing food waste in future. The authors worked
with datapoints from a range of years and other
data to form a 2019 estimate which is what we use
Middle China here. High Confidence
This paper aggregates a huge number of what they (Zhang et al.,
China term 'catering waste' papers. 47 of them in total 2020)
from various areas of China. It's a mixture of
surveys, official statistics, author's calculations etc.
to create data on catering food waste across
different regions. This is then associated to a 47.48
number of correlates which are used to predict
growing food waste in future. The authors worked
with datapoints from a range of years and other
data to form a 2019 estimate which is what we use
West China here. High Confidence
This paper aggregates a huge number of what they
term 'catering waste' papers. 47 of them in total
from various areas of China. It's a mixture of
surveys, official statistics, author's calculations etc.
to create data on catering food waste across
different regions. This is then associated to a
number of correlates which are used to predict
45.00
growing food waste in future. The authors worked
with datapoints from a range of years and other
data to form a 2019 estimate which is what we use
here. The total waste figure itself was not listed in
the text but was confirmed with the authors as
being 38 million tonnes. This refers only to urban
Urban China Total catering waste. High Confidence
 This paper aggregates a huge number of what they
term 'catering waste' papers. 47 of them in total
from various areas of China. It is a mixture of
surveys, official statistics, author's calculations etc.
to create data on catering food waste across
different regions. This is then associated to a 34.33
number of correlates which are used to predict
growing food waste in future. The authors worked
with datapoints from a range of years and other
data to form a 2019 estimate which is what we use
Northeast China here. High Confidence
Electronic questionnaire studies were issues to
chefs among hotels, restaurants and canteens. 474
chefs reported information about their food waste.
The paper also refers to 53 businesses being
studied; it is unclear if this number includes out of (Danish
home consumption businesses or just retail. The Environmental
Denmark 20.64
authors claim it was designed to be as Protection
representative as possible. The figure taken is the Agency, 2014)
sum of Hotels (4), Restaurants (5), Institutions (6)
and Canteens (7). Confidence intervals of ±25 are
provided for hotels and restaurants, ±50 for
institutions and ±10 for canteens. High Confidence
20 catering institutions had waste audits run: 3
restaurants, 3 bars/pubs, 3 cafes, 4
canteens/buffets, 3 schools, 3 kindergartens, 1
hospital. This therefore offers a good variation of
(Moora, Evelin, et
Estonia commercial and non-commercial out of home 16.61
al., 2015)
consumption. Across 5 days, waste was measured
at preparation, serving, consumption and storing
food to provide waste generation figures which
was scaled up. High Confidence
References three different papers: from Katajajuuri
et al. (2014), Silvenoinen et al. (2014) and HSY
(2012). The Katajajuuri paper is also a separate
data point in this database. The first two only
estimate avoidable waste (80,000 tonnes), the
latter was used to estimate avoidable waste
(50,000 tonnes). This was done to make them more
(Stenmarck et al.,
comparable to other studies of all food waste. 24.04
2016)
Since this adjusted total food waste estimates
comes from Finland-specific data, it was judged
that leaving it as a separate data point to compare
and average with our generic edible share adjusted
figure would be prudent. The adjusted per capita
figures are highly comparable to one another, Medium
giving some confidence in our estimate. Confidence
Finland
This paper includes the results of two separate
studies, one on the commercial sector and one on
the non-commercial sector. The commercial sector
study involved 72 restaurants across 17 businesses
- a range of diners, cafes, restaurants, hotels etc. -
having their waste measured for a single day. The
non-commercial study involved 55 outlets
(Katajajuuri et al.,
representing three companies providing food for 22.57
2014)
day care, hospitals and workplace canteens,
measured over one week. Table 5 in the paper
refers to "avoidable food waste in the Finnish food
supply chain" suggesting that this did not measure
total food waste, only the avoidable share, and the
figure is therefore adjusted to estimate this Medium
inedible or unavoidable share. Confidence
 Combination of studies across canteens and
restaurants. 22 canteens and 37 restaurants
interviewed. Most of these were self-estimation
without measurement, though some
measurements were taken independent of the
study. The study also conducted a waste audit in
one canteen and one restaurant. These were
compared to a similar 2011 Agriculture Ministry
study of out of home waste. They used this
31.60 (ADEME, 2016)
information to confirm the underestimation of out
of home actors to their own waste and prioritise
France the 'realised' measure over the actor estimate. The
waste per meal (including storage waste from
kitchen, not just plate waste) is scaled for the
respective commercial and non-commercial sector.
However, it only looks at edible waste (destined for
human consumption) and no preparation waste. Medium
This is therefore adjusted. Confidence
Estimate judged to be of 'sufficient quality' by
Stenmarck et al. (2016), it combines literature (BIO Intelligence
17.37
references from a Danish Environmental Ministry Service, 2010) Medium
Report (2010) and ADEME (2004). Confidence
This is an English summary paper of a study in
German, which means some of the details of the
methodology are unclear. They claim to use "the
best available data at the time of the study", which
will include papers in German. Figure S.1 shows the
admissible measurement methods and those which
were applied for each sector, which is in line with
the methods we are targeting here. However, the
(Schmidt et al.,
Germany English summary does not include detail on the 20.58
2019)
specific sources used so it is unclear how many of
each method there are. For out of home, they use
Direct Measurements; Waste Composition
Analysis; Records.
This paper forms the 2015 baseline for food waste
against which the national strategy for reducing
food waste will be measured, it is therefore the
preferred figure for Germany. High Confidence
Cited as being from the Irish Environmental
Protection Agency (2015). The original source and
weight estimates were not found based on the
bibliography information or direct searches. (Stenmarck et al.,
Ireland 56.15
However, it was judged by Stenmarck et al. (2016) 2016)
to be 'data of sufficient quality'. The inability to find
the source paper means we cannot have high Medium
confidence. Confidence
 The paper describes the figure as coming from "A
comprehensive value chain model for various food
production and consumption stages was designed
to assess food waste and the potential for food
rescue in Israel. The model is based on a bottom-up
approach, and includes analysis of data relevant to
agricultural production, import, export, industry,
distribution and a sample of consumption patterns
of 50 different types of food." [...] "For each type of
food, the volume of input and output was
measured in terms of gross agricultural product
and loss rate for every stage of the value chain in (Leket Israel,
Israel 27.44
the food production, distribution and consumption 2019)
process." [...] "This data is indicative and intended
to serve as the basis for public debate, and for
further research and study".
As a result, it is not that clear exactly the
calculation that has taken place nor the original
data sources, other than that they are 'bottom up'.
Leket is a primary food rescue organisation in Israel
and this paper was referred to us by contacts in the
region. The lack of clarity on the methodological
details means we cannot have high confidence in Medium
the estimate. Confidence
This presentation (in English) by the Food Industrial
Policy Office presents statistics from the MAFF
statistical survey, estimating waste from each of
the sectors: Household (solid waste); Restaurant
(assumed to mean all Food Service) and Retail.
Wholesale estimated separately, not included. (Food Industry
Japan Estimates are for the 2014 financial year. The 14.75 Policy Office,
survey methodology is not presented. As a result of 2017)
this methodological uncertainty, it is considered
'medium confidence' until more information is
received. For Retail and Food Service, the figures
have been adjusted to account for the share going Medium
to animal feed, also included in the presentation. Confidence
Across retail and out of home consumption, 90
locations had their waste analysed for a period of
seven days, which was preceded by a one-day test
measurement, which was excluded from analysis.
Kenya The figure presented is the sum of Restaurants, 31.14 (JICA, 2010)
Hotels and Public Facilities, each of which had a
distinct waste generation rate and food waste
generation share. The original study scales this by Medium
Nairobi the number of institutions in Nairobi. Confidence
Combination of waste statistics and interviews and
(Luxembourg
surveys with enterprises. Further details are not
Luxembourg 20.90 Environment
known, but it is published by the Luxembourg Medium
Ministry, 2020)
Environment Administration. Confidence
Table 1 cites 'MHLG' (Ministry of Housing and Local
Government) 2011, estimating food waste
generation by source. This was not findable by the (Jereme et al.,
Malaysia 89.56
bibliography nor through a direct internet search. 2013)
As a result, we cannot have high confidence in the Medium
estimate. Confidence
 References the research project KuttMatsvinn2020,
which focused on hotels, convenience stores and
employee cafeterias, sampling around 2000
catering outlets. Work on other subsectors such as
schools and restaurants is underway; as a result it
appears as though the estimate has not accounted
for these other sectors, making it an incomplete
Food Service estimate. Uses the ForMat definition (Stensgård et al.,
Norway 4.96
of food which only measures edible food waste, 2019)
not inedible parts. It has therefore been adjusted
to increase comparability. In this definition, food
sent to animal feed is considered waste. It was not
possible to adjust and account for this as the share
of food going to animal feed was only included for
the production stage; it is therefore assumed no Medium
food waste is going to animals at other stages. Confidence
Interviews conducted with approximately 100
hotels, restaurants and caterers to determine the
share of food waste at the stages of kitchen
preparation and plate waste. Unclear to what (Bogdanović, et
Serbia 6.00
extent survey respondents were estimating or al., 2019)
based on internal measurement. The waste
generation factors from this were applied to CEVES Medium
estimates on food purchases in Serbian HoReCa. Confidence
Data from the Slovenian Statistical Office, but the
exact methodology is unclear. The methodological
explanation from the same site mentions three
different annual surveys on waste collection, (Republic of
generation and recovery/disposal, alongside an ad- Slovenia
19.54
hoc questionnaire of public waste collection Statistical Office,
services. It is unclear to whom the surveys are sent 2019)
(i.e., to waste generators or to waste collection
services) and if it requires submission of observed Medium
data or some other form of self-reporting. Confidence
Slovenia
Data from the Slovenian Statistical Office, but the
exact methodology is unclear. The methodological
explanation from the same site mentions three
different annual surveys on waste collection, (Republic of
generation and recovery/disposal, alongside an ad- Slovenia
20.25
hoc questionnaire of public waste collection Statistical Office,
services. It is unclear to whom the surveys are sent 2020)
(i.e., to waste generators or to waste collection
services) and if it requires submission of observed Medium
data or some other form of self-reporting. Confidence
This estimate takes waste factors derived from two
different papers, both in Swedish. One is a
restaurant-specific factor taken from Jensen et al.
(2011) which was then scaled by waste per
employee. The other is a catering factor from Stare 21.00
et al. (2013), which has a factor of waste per
portion in schools. This has been applied to schools
and other catering facilities, including prisons and (Swedish
healthcare. Environmental High Confidence
Sweden
Protection
This estimate takes waste factors derived from two Agency, 2014)
different papers, in Swedish. One is a restaurant-
specific factor taken from Jensen et al. (2011)
which was then scaled by waste per employee. The
20.00
other is a catering factor from Stare et al. (2013),
which has a factor of waste per portion in schools.
This has been applied to schools and other catering
facilities, including prisons and healthcare. High Confidence
This paper cites Baier and Reinhard (2007) for food
service installations, and Andrini and Bauen (2005)
for restaurants. Both are in German and involved
(Beretta et al.,
Switzerland waste audits. 40 food service installations were 40.00
2013)
audited and 20 restaurants. These estimates have
been combined to create an estimate of total Out Medium
of Home waste. Confidence
 United Food waste data remodelled based on WRAP's
Kingdom of 2013 analysis of food waste in the hospitality and
Great Britain food service sector, a study which employed waste 16.50 (WRAP, 2020)
and compositional analyses and analysis of DEFRA
Northern survey information. This data was re-weighted to
Ireland account for the change in number and size of
premises, number of pupils served by school
catering etc. High Confidence
US EPA synthesis of food waste consistent with the
FLW protocol. Combines waste generation factors
from other studies with relevant scaling statistics,
such as employees or revenue for a sector. Food
Service as an estimate which combines hospitality
(hotels, restaurants/food service; sports venues)
and institutional (healthcare, office buildings, (U.S.
United
military and prisons, schools and universities). 51 Environmental
States of 63.62
studies in total used to inform Food Service Protection
America
generation rate across the various subsectors. Agency, 2020)
Waste management pathways were provided for
the commercial and institutional subsectors
separately; only in the commercial hospitality
subsector was some surplus managed to a non-
waste destination (14% donated), this has been
accounted for. High Confidence

3.3 Available estimates / datapoints: Retail

Table 12: Available datapoints for Manufacturing, by country

Country Subnatio Adjusted

nal study kg/capita/y Source
area (if ear (study Confiden
relevant) Methodological Description estimate number) ce Level
This paper represents the national baseline estimate for food waste
commissioned by the Australian government in order to be the basis for
measurement and reporting of progress on Australia's National Food
(Arcadis,
Australia Waste Strategy 2017. The Retail figure was calculated through publicly 9.45
2019)
available waste estimates for major supermarkets and data from large High
shopping centre organisations. The source paper also has an estimate for Confiden
Wholesale, which has been excluded here. ce
Five different retail companies, with market share of about 83%, (Environme
contributed data on their food waste. This paper was also referenced by nt Agency High
Austria 8.63
EU FUSIONS and judged to be 'data of sufficient quality'. Austria, Confiden
2017) ce
A combination of data from the Integrated Environmental Report and (Flemish
sector surveys by COMEOS and Buurtsuper.be. Where these were national Food
surveys, the authors adjusted to Flanders based on sales figures. Their Supply
Belgium figures have been scaled for the whole of Belgium, assuming that Flanders 9.71 Chain
can be considered representative of the whole nation. 3% of this sector is Platform Medium
estimated to go to animal feed, which is not considered waste. The for Food Confiden
Flanders numbers have been adjusted to reflect this. Loss, 2017) ce
 Paper is in Danish but has a summary in English. The waste from 53
businesses was studied, 69 were visited to discuss their food waste (Danish
management. It's unclear what share of this sample was retail, but the Environme
paper claims the sample was aimed to be representative. It analyses not ntal
Denmark 29.80
just non-specialised stores (supermarkets etc.) but also specialised stores Protection
such as butchers, fishmongers and greengrocers. The value is the sum of Agency, High
non-specialised (1) and specialised retail (2) but does not include 2014) Confiden
wholesale (3). Confidence intervals of ±25 are given for each retail sector. ce
This study cites a paper in Estonian from which the figures are taken. 11
retailers had detailed analyses carried out including interviews, on-site
observations and weighing. An additional questionnaire was sent to 600
(Caldeira et
retailers and 183 wholesalers, scaled by grocery store distribution across 4.70
al., 2019)
Estonia. This study was also cited in Stenmarck et al. (2016) with a slightly Medium
different waste figure, it is unclear if this is rounding error or some other Confiden
Estonia calculation. ce
Study in Estonian, referenced in Stenmarck et al. (2016), which details
that 11 retailers had detailed analyses carried out including interviews,
(Moora,
on-site observations and weighing. An additional questionnaire was sent
4.72 Piirsalu, et
to 600 retailers and 183 wholesalers, scaled by grocery store distribution Medium
al., 2015)
across Estonia. This paper is also referenced in Caldeira et al. (2019) with a Confiden
slightly different waste figure, it is unclear why these are different. ce
Paper in French, with summary available in English and details in Caldeira
et al. (2019). Across the supply chain, 582 interviews took place: 512 (ADEME, Medium
France 25.60
'quali-quantitative', 80 qualitative. Looks only at edible waste (food for 2016) Confiden
human consumption), so has been adjusted. ce
This is an English summary paper of a study in German, which means
some of the details of the methodology are unclear. They claim to use
"the best available data at the time of the study", which will include
papers in German. Figure S.1 shows the admissible measurement
methods and those which were applied for each sector, which is in line
with the methods we are targeting here. However, the English summary
(Schmidt et
Germany does not include detail on the specific sources used so it's unclear how 5.97
al., 2019)
many of each method there are. For 'trade', which we are understanding
as retail, they use Direct Measurements and Counts and Scans. There is a
risk that wholesale has been included in this figure.
This paper forms the 2015 baseline for food waste against which the High
national strategy for reducing food waste will be measured, it is therefore Confiden
the preferred figure for Germany. ce
Paper cites 'Greek Waste Statistics' from 2011 which could not be found
based on the bibliography reference or internet searches. However, it was (Stenmarck Medium
Greece 7.38
judged to be 'data of sufficient quality' by Stenmarck et al. (2016). The et al., 2016) Confiden
inability to find the source paper means we cannot have high confidence. ce
The paper describes the figure as coming from "A comprehensive value
chain model for various food production and consumption stages was
designed to assess food waste and the potential for food rescue in Israel.
The model is based on a bottom-up approach, and includes analysis of
data relevant to agricultural production, import, export, industry,
distribution and a sample of consumption patterns of 50 different types of
food." [...] "For each type of food, the volume of input and output was
(Leket
measured in terms of gross agricultural product and loss rate for every
Israel 51.41 Israel,
stage of the value chain in the food production, distribution and
2019)
consumption process." [...] "This data is indicative and intended to serve
as the basis for public debate, and for further research and study".
As a result, it is not that clear exactly the calculation that has taken place
nor the original data sources, other than that they are 'bottom up'. Leket
is a primary food rescue organisation in Israel and this paper was referred Medium
to us by contacts in the region. The lack of clarity on the methodological Confiden
details means we cannot have high confidence in the estimate. ce
17 retail stores, including both supermarkets and hypermarkets had their
(Clara
waste weighed. The linked infographic suggests a combination of direct High
Italy 3.63 Cicatiello et
weighing and item scanning, which was used to form a waste generation Confiden
al., 2019)
factor by retail space, which was then scaled up nationwide. ce
 This presentation (in English) by the Food Industrial Policy Office presents
statistics from the MAFF statistical survey, estimating waste from each of
the sectors: Household (solid waste); Restaurant (assumed to mean all (Food
Food Service) and Retail. Wholesale estimated separately, not included. Industry
Japan Estimates are for the 2014 financial year. The survey methodology is not 8.63 Policy
presented. As a result of this methodological uncertainty, it is considered Office,
'medium confidence' until more information is received. For Retail and 2017) Medium
Food Service, the figures have been adjusted to account for the share Confiden
going to animal feed, also included in the presentation. ce
Across retail and out of home consumption, 90 locations had their waste
analysed for a period of seven days, which was preceded by a one-day
Kenya test measurement, which was excluded from analysis. Figure is a sum of 10.96 (JICA, 2010) Medium
Shop and Market, which are measured separately. The original study Confiden
Nairobi scales this by the number of institutions in Nairobi. ce
This figure comes from a review paper which references a primary source
(in German) which is mentioned as being a survey to retail and
(Caldeira et
distribution companies, although further methodological details are not 9.10 Medium
al., 2019)
provided. This estimate has been complemented by a more recent study Confiden
in the same country. ce
The paper cites 'Luxembourg Waste Statistics 2015'. This was not found
either through the bibliography or through a direct internet search.
Luxembo (Stenmarck
However, it was considered 'data of sufficient quality' for the purposes of 3.90 Medium
urg et al., 2016)
EU FUSIONS. The inability to find the source paper means we cannot have Confiden
high confidence. ce
Combination of waste statistics and interviews and surveys with (Luxembou
enterprises. Further details are not known, but it is published by the rg
Luxembourg Environment Administration. 8.70 Environme Medium
nt Ministry, Confiden
2020) ce
Table 1 cites 'MHLG' (Ministry of Housing and Local Government) 2011,
estimating food waste generation by source. This was not findable by the (Jereme et Medium
Malaysia 78.82
bibliography nor through a direct internet search. As a result, we cannot al., 2013) Confiden
have high confidence in the estimate. ce
The document cites a paper (De Waart, 2011) which could not be found
through a direct internet search for the paper. As a result, it cannot be
Netherlan (Stenmarck
verified any further. It was judged to be data of 'sufficient quality' for the 11.00 Medium
ds et al., 2016)
EU FUSIONS project, however. The inability to find the source paper Confiden
means we cannot have high confidence. ce
This summary document refers to a University of Otago Master's student
having conducted waste audits at three supermarket chains. It also (Love Food
New
presents the final destinations of retail waste, which has been used to 3.12 Hate Waste High
Zealand
adjust the waste figure. The share going to Animal Feed, Donation and NZ, 2020) Confiden
Protein Reprocessing has been removed from the waste figure. ce
89 stores from 3 different retail chains provided data, alongside
wholesalers covering a large share of the market, both upscaled to the
whole of Norway. It was not possible to disaggregate wholesale. Only 12.93 Medium
edible food reported, and the figures have therefore been adjusted to Confiden
estimate the entire food waste. (Caldeira et ce
89 stores from 3 different retail chains provided data, alongside al., 2019)
wholesalers covering a large share of the market, both upscaled to the
whole of Norway. It was not possible to disaggregate wholesale. Only 14.11 Medium
edible food reported, and the figures have therefore been adjusted to Confiden
estimate the entire food waste. ce
89 stores from 3 different retail chains provided data, alongside
Norway wholesalers covering a large share of the market, both upscaled to the (Stensgård
whole of Norway. It was not possible to disaggregate wholesale. Only 14.39 & Hanssen, Medium
edible food reported, and the figures have therefore been adjusted to 2016) Confiden
estimate the entire food waste. ce
Retailers report data on waste and sales by product group and cause.
Wholesalers also provide estimates this way, these have not been
included here. Uses the ForMat definition of food which only measures
edible food waste, not inedible parts. It has therefore been adjusted to
(Stensgård
increase comparability. In this definition, food sent to animal feed is 13.67
et al., 2019)
considered waste. It was not possible to adjust and account for this as the
share of food going to animal feed was only included for the production Medium
stage; it is therefore assumed no food waste is going to animals at other Confiden
stages. ce
 The paper cites what is assumed to be a waste composition analysis by
the Higher School of Economics (which was not found when searched for)
and data from Rosstat. In addition, the shares of waste at each stage are
Russian calculations based on data from Russian Agriculture Ministry (2017). The
(Tiarcenter,
Federatio estimate provides a total food waste estimate as well as the amount of 13.72
2019)
n waste at each stage of the chain, these have been combined to form
sector-specific estimates. The inability to trace the original source data Medium
and the lack of transparency on the calculations means we cannot have Confiden
high confidence in this estimate. ce
This study forms the Saudi waste Baseline, conducted by Saudi Grains
Organisation (SAGO). For Retail, over 7,000 samples across 19 product
groups were taken. It is unclear, however, from how many retailers
samples were taken. Wholesale is not disaggregated from Retail so is
Saudi included. Samples taken across 13 regions in Saudi Arabia. The value (SAGO,
19.65
Arabia taken is the share of total waste attributed to 'Distribution'. Additional 2019)
information and images to supplement the main study can be found here:
https://www.macs- High
g20.org/fileadmin/macs/Activities/2020_FLW_WS/4_Session_3_FW_at_H Confiden
H_level_small.pdf ce
Data from the Slovenian Statistical Office, but the exact methodology is
unclear. The methodological explanation from the same site mentions (Republic
three different annual surveys on waste collection, generation and of Slovenia
Slovenia recovery/disposal, alongside an ad-hoc questionnaire of public waste 6.62 Statistical
collection services. It is unclear to whom the surveys are sent (i.e., to Office, Medium
waste generators or to waste collection services) and if it requires 2020) Confiden
submission of observed data or some other form of self-reporting. ce
The paper is produced by Naturvårdsverket, the Swedish Environmental (Swedish
Protection Agency, and is based on industry reporting to the EPA. The Environme
summary (in English) does not provide detailed methodology on number ntal
Sweden 10.00
of companies reporting. Protection High
Agency, Confiden
2020) ce
Data provided by Retail signatories to Courtauld 2025, which cover more
than 95% of the food retail sector (by sales), were used. They were scaled
United up based on market coverage.
Kingdom
of Great
(WRAP,
Britain 4.20
2020)
and
Northern
Ireland High
Confiden
ce
US EPA synthesis of food waste consistent with the FLW protocol.
Combines waste generation factors from other studies with relevant
(U.S.
scaling statistics, such as employees or revenue for a sector. Nine studies
Environme
United were used, a mixture of waste audits and surveys, to form the waste
ntal
States of generation factors. Wholesale also estimated, not included in the paper. 9 15.65
Protection
America studies used to inform supermarket and supercenter waste generation
Agency,
rate. Waste management pathways for wholesale/retail were provided, High
2020)
these were used to remove the share donated (21%) and going to animal Confiden
feed (14%) ce
3.4 Bibliography
This bibliography contains all of the studies referenced in the Appendices. For a list only of the
references used to inform the Level 1 analysis, see the spreadsheet published alongside the report.

Abeliotis, K., Lasaridi, K., Costarelli, V., & Chroni, C. (2015). The implications of food waste generation

on climate change: The case of Greece. Sustainable Production and Consumption, 3, 8–14.

https://doi.org/10.1016/j.spc.2015.06.006

Abiad, M. G., & Meho, L. I. (2018). Food loss and food waste research in the Arab world: A systematic

review. Food Security, 10(2), 311–322. https://doi.org/10.1007/s12571-018-0782-7

ADEME. (2016). Pertes et gaspillages alimentaires: L’état des lieux et leur gestion par étapes de la

chaîne alimentaire (p. 165). ADEME. https://www.ademe.fr/etat-lieux-masses-gaspillages-

alimentaires-gestion-differentes-etapes-chaine-alimentaire

Alayam. (2018). Minister of works: 195 thousand tons of food waste annually. Alayam.

https://www.alayam.com/online/local/737712/News.html

Al-Maliky, S. J. B., & ElKhayat, Z. Q. (2012). Kitchen Food Waste Inventory for Residential Areas in

Baghdad City. Modern Applied Science, 6(8), p45. https://doi.org/10.5539/mas.v6n8p45

Al-Mas’udi, R. M., & Al-Haydari, M. A. S. (2015). Spatial Analysis of Residential Waste Solid in the City

of Karbala. Journal of Kerbala University, 13(2), 132–154.

Al-Rawi, S. M., & Al-Tayyar, T. A. (2013). A Study on Solid Waste Composition And Characteristics of

Mosul City/Iraq. Journal of University of Zakho, 1(2), 496–507.

Araujo, G. P. de, Lourenço, C. E., Araújo, C. M. L. de, & Bastos, A. (2018). Intercâmbio Brasil-União

Europeia sobre desperdício de alimentos: Relatório final (p. 40).

https://www.embrapa.br/busca-de-publicacoes/-/publicacao/1105525/intercambio-brasil-

uniao-europeia-sobre-desperdicio-de-alimentos-relatorio-final

Arcadis. (2019). National Food Waste Baseline: Final assessment report. Arcadis.

https://www.environment.gov.au/system/files/pages/25e36a8c-3a9c-487c-a9cb-

66ec15ba61d0/files/national-food-waste-baseline-final-assessment.pdf
Assefa, M. (2017). Solid Waste Generation Rate and Characterization Study for Laga Tafo Laga Dadi

Town, Oromia, Ethiopia. International Journal of Environmental Protection and Policy, 5(6),

84. https://doi.org/10.11648/j.ijepp.20170506.11

Beretta, C., Stoessel, F., Baier, U., & Hellweg, S. (2013). Quantifying food losses and the potential for

reduction in Switzerland. Waste Management, 33(3), 764–773.

https://doi.org/10.1016/j.wasman.2012.11.007

BIO Intelligence Service. (2010). Preparatory study on food waste across EU 27. European

Commission. http://ec.europa.eu/environment/eussd/pdf/bio_foodwaste_report.pdf

Bogdanović, M., Bobić, D., Danon, M., & Suzić, M. (2019). Circular Economy Impact Assessment: Food

waste in HORECA sector. Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ).

https://www.giz.de/en/downloads/CE%20impact%20assessment_HORECA.pdf

Caldeira, C., Barco Cobalea, H., Serenella, S., De Laurentiis, V., European Commission, & Joint

Research Centre. (2019). Review of studies on food waste accounting at Member State level.

https://op.europa.eu/publication/manifestation_identifier/PUB_KJNA29828ENN

Chakona, G., & Shackleton, C. M. (2017). Local setting influences the quantity of household food

waste in mid-sized South African towns. PLOS ONE, 12(12), e0189407.

https://doi.org/10.1371/journal.pone.0189407

Chalak, A., Abiad, M. G., Diab, M., & Nasreddine, L. (2019). The Determinants of Household Food

Waste Generation and its Associated Caloric and Nutrient Losses: The Case of Lebanon. PLOS

ONE, 14(12), e0225789. https://doi.org/10.1371/journal.pone.0225789

Cicatiello, C. (2018). Measuring household food waste at national level: A systematic review on

methods and results. CAB Reviews: Perspectives in Agriculture, Veterinary Science, Nutrition

and Natural Resources, 13(056). https://doi.org/10.1079/PAVSNNR201813056

Cicatiello, Clara, Franco, S., & Falasconi, L. (2019). Gli sprechi alimentari nella grande distribuzione

organizzata in Italia. Quantificazione e analisi dei prodotti alimentari smaltiti nei

 supermercati e ipermercati. REDUCE. https://www.sprecozero.it/wp-

content/uploads/2020/07/Report-AR4-GDO.pdf

Cicatiello, Clara, Franco, S., Pancino, B., & Blasi, E. (2016). The value of food waste: An exploratory

study on retailing. Journal of Retailing and Consumer Services, 30, 96–104.

https://doi.org/10.1016/j.jretconser.2016.01.004

Danish Environmental Protection Agency. (2014). Kortlægning af madaffald i servicesektoren: Detail

handel, restauranter og storkøkkener.

https://www2.mst.dk/Udgiv/publikationer/2014/07/978-87-93178-75-5.pdf

Danish Environmental Protection Agency. (2018). Kortlægning af sammenstaetningen af

dagrenovation og kildesorteret oranisk affald fra husholdninger.

https://www2.mst.dk/Udgiv/publikationer/2018/03/978-87-93614-78-9.pdf

Delley, M., & Brunner, T. A. (2018). Household food waste quantification: Comparison of two

methods. British Food Journal, 120(7), 1504–1515. https://doi.org/10.1108/BFJ-09-2017-

0486

Denafas, G., Ruzgas, T., Martuzevičius, D., Shmarin, S., Hoffmann, M., Mykhaylenko, V., Ogorodnik,

S., Romanov, M., Neguliaeva, E., Chusov, A., Turkadze, T., Bochoidze, I., & Ludwig, C. (2014).

Seasonal variation of municipal solid waste generation and composition in four East

European cities. Resources, Conservation and Recycling, 89, 22–30.

https://doi.org/10.1016/j.resconrec.2014.06.001

Dhokhikah, Y., Trihadiningrum, Y., & Sunaryo, S. (2015). Community participation in household solid

waste reduction in Surabaya, Indonesia. Resources, Conservation and Recycling, 102, 153–

162. https://doi.org/10.1016/j.resconrec.2015.06.013

Dou, Z., & Toth, J. D. (2020). Global primary data on consumer food waste: Rate and characteristics –

A review. Resources, Conservation and Recycling, 105332.

https://doi.org/10.1016/j.resconrec.2020.105332
Edema, M. O., Sichamba, V., & Ntengwe, F. W. (2012). Solid waste management—Case study of

Ndola, Zambia. International Journal of Plant, Animal and Environmental Sciences, 2(3).

https://www.academia.edu/30874341/SOLID_WASTE_MANAGEMENT_CASE_STUDY_OF_N

DOLA_ZAMBIA

Edjabou, M. E., Petersen, C., Scheutz, C., & Astrup, T. F. (2016). Food waste from Danish households:

Generation and composition. Waste Management, 52, 256–268.

https://doi.org/10.1016/j.wasman.2016.03.032

El Bilali, H. (2018). Research on food losses and waste in North Africa. The North African Journal of

Food and Nutrition Research, 02(03), 8.

El Bilali, H., & Ben Hassen, T. (2020). Food Waste in the Countries of the Gulf Cooperation Council: A

Systematic Review. Foods, 9(4), 463. https://doi.org/10.3390/foods9040463

Elimelech, E., Ayalon, O., & Ert, E. (2018). What gets measured gets managed: A new method of

measuring household food waste. Waste Management, 76, 68–81.

https://doi.org/10.1016/j.wasman.2018.03.031

Environment Agency Austria. (2017). Food Waste Statistics Austria.

https://ec.europa.eu/food/sites/food/files/safety/docs/fw_eu-platform_20170925_sub-

fwm_pres-02b.pdf

Environment and Climate Change Canada. (2019). National Waste Characterization Report: The

Composition of Canadian Residual Municipal Solid Waste.

http://publications.gc.ca/collections/collection_2020/eccc/en14/En14-405-2020-eng.pdf

Flemish Food Supply Chain Platform for Food Loss. (2017). Food Waste and Food Losses: Prevention

and Valorisation.

https://www.voedselverlies.be/sites/default/files/atoms/files/Monitor_EN_final.pdf

Food Industry Policy Office. (2017, August). Reducing Food Loss and Waste & Promoting Recycling.

https://www.maff.go.jp/e/policies/env/attach/pdf/frecycle-3.pdf
Gao, L., Cheng, S., Cao, X., Zhang, D., Xiaojie Liu, Qin, Q., & Liu, Y. (2013). An Overview of the

Resources and Environmental Issues from Wasted Food in Urban Catering Across China.

Journal of Resources and Ecology, 4(4), 337–343. https://doi.org/10.5814/j.issn.1674-

764x.2013.04.006

Giordano, C., Alboni, F., & Falasconi, L. (2019). Quantities, Determinants, and Awareness of

Households’ Food Waste in Italy: A Comparison between Diary and Questionnaires

Quantities’. Sustainability, 11(12), 3381. https://doi.org/10.3390/su11123381

Giordano, C., Piras, S., Boschini, M., & Falasconi, L. (2018). Are questionnaires a reliable method to

measure food waste? A pilot study on Italian households. British Food Journal, 120(12),

2885–2897. https://doi.org/10.1108/BFJ-02-2018-0081

Grover, P., & Singh, P. (2014). An Analytical Study of Effect of Family Income and Size on Per Capita

Household Solid Waste Generation in Developing Countries. Review of Arts and Humanities,

3(1), 127–143.

Gu, B., Wang, H., Chen, Z., Jiang, S., Zhu, W., Liu, M., Chen, Y., Wu, Y., He, S., Cheng, R., Yang, J., & Bi,

J. (2015). Characterization, quantification and management of household solid waste: A case

study in China. Resources, Conservation and Recycling, 98, 67–75.

https://doi.org/10.1016/j.resconrec.2015.03.001

Gustavsson, J., Cederberg, C., & Sonesson, U. (2011). Global food losses and food waste: Extent,

causes and prevention (p. 38). FAO. http://www.fao.org/3/a-i2697e.pdf

Hanssen, O. J., Syversen, F., & Stø, E. (2016). Edible food waste from Norwegian households—

Detailed food waste composition analysis among households in two different regions in

Norway. Resources, Conservation and Recycling, 109, 146–154.

https://doi.org/10.1016/j.resconrec.2016.03.010

Inter-American Development Bank. (2011). Waste Generation and Composition Study for the

Western Corridor, Belize C.A. (2056/)C-BL). http://belizeswama.com/wp-

 content/uploads/2018/12/Waste-Generation-Composition-Study-for-Western-Corridor-

Belize-C.A.-2056-OC-BL1.pdf

Jereme, I., Chamhuri, S., Talib, B. A., & Begum, R. A. (2013). Household food composition and

disposal behaviour in Malaysia. The Social Sciences, 8(6), 553–539.

https://doi.org/10.3923/sscience.2013.533.539

JICA. (2005). The Study on the Solid Waste Management in Dhaka City (Volume 2). Japan

International Cooperation Agency. https://openjicareport.jica.go.jp/pdf/11785243.pdf

JICA. (2010). Preparatory Survey for Integrated Solid Waste Management in Nairobi City in the

Republic of Kenya (Volume 1). Japan International Cooperation Agency.

https://openjicareport.jica.go.jp/pdf/12005443.pdf

JICA. (2013). Project on Master Plan Study for Integrated Solid Waste Management in Bogota, D.C.

(Volume 2). Japan International Cooperation Agency.

https://openjicareport.jica.go.jp/pdf/12126843.pdf

JICA. (2015). Project for Integrated Solid Waste Management Master Plan in Gujranwala (Volume 3).

Japan International Cooperation Agency.

https://openjicareport.jica.go.jp/pdf/12246336_01.pdf

JICA. (2016). Data Collection Survey on Solid Waste Management in Democratic Socialist Republic of

Sri Lanka. Japan International Cooperation Agency.

https://openjicareport.jica.go.jp/pdf/12250213.pdf

Joseph, L. P., & Prasad, R. (2020). Assessing the sustainable municipal solid waste (MSW) to

electricity generation potentials in selected Pacific Small Island Developing States (PSIDS).

Journal of Cleaner Production, 248, 119222. https://doi.org/10.1016/j.jclepro.2019.119222

Kasza, G., Dorkó, A., Kunszabó, A., & Szakos, D. (2020). Quantification of Household Food Waste in

Hungary: A Replication Study Using the FUSIONS Methodology. Sustainability, 12(8), 3069.

https://doi.org/10.3390/su12083069
Katajajuuri, J.-M., Silvennoinen, K., Hartikainen, H., Heikkilä, L., & Reinikainen, A. (2014). Food waste

in the Finnish food chain. Journal of Cleaner Production, 73, 322–329.

https://doi.org/10.1016/j.jclepro.2013.12.057

Kemper, K., Voegele, J., Hickey, V., Ahuja, P. S., Poveda, R., Edmeades, S., Kneller, C., Swannell, R.,

Gillick, S., Corallo, A., Aguilar, G., Alencastro, S., Felix, E., & Sebastian, A. (2019). Mexico

Conceptual Framework for a National Strategy on Food Loss and Waste (p. 68).

https://beta.wrap.org.uk/resources/report/conceptual-framework-national-strategy-food-

loss-and-waste-mexico

Leket Israel. (2019). Food Waste and Rescue in Israel: The Economic, Social and Environmental

Impact. https://www.leket.org/en/food-waste-and-rescue-report/

Li, Y., Wang, L., Liu, G., & Cheng, S. (2021). Rural household food waste characteristics and driving

factors in China. Resources, Conservation and Recycling, 164, 105209.

https://doi.org/10.1016/j.resconrec.2020.105209

Lo, I. M. C., & Woon, K. S. (2016). Food waste collection and recycling for value-added products:

Potential applications and challenges in Hong Kong. Environmental Science and Pollution

Research, 23(8), 7081–7091. https://doi.org/10.1007/s11356-015-4235-y

Love Food Hate Waste NZ. (2020). What is known about food waste in New Zealand.

https://lovefoodhatewaste.co.nz/wp-content/uploads/2020/09/What-is-known-about-food-

waste-in-New-Zealand.pdf

Luxembourg Environment Ministry. (2020). Génération, traitement et prévention des déchets

alimentaires. http://environnement.public.lu/fr/offall-ressourcen/types-de-

dechets/Biodechets/Gaspillage_alimentaire/Etudes_et_resultats.html

Malefors, C., Callewaert, P., Hansson, P.-A., Hartikainen, H., Pietiläinen, O., Strid, I., Strotmann, C., &

Eriksson, M. (2019). Towards a Baseline for Food-Waste Quantification in the Hospitality

Sector—Quantities and Data Processing Criteria. Sustainability, 11(13), 3541.

https://doi.org/10.3390/su11133541
Miezah, K., Obiri-Danso, K., Kádár, Z., Fei-Baffoe, B., & Mensah, M. Y. (2015). Municipal solid waste

characterization and quantification as a measure towards effective waste management in

Ghana. Waste Management, 46, 15–27. https://doi.org/10.1016/j.wasman.2015.09.009

Moora, H., Evelin, U.-P., & Õunapuu, K. (2015). Toidujäätmete ja toidukao teke Eesti

kodumajapidamistes ja toitlustusasutustes. Stockholm Environment Institute.

https://www.sei.org/wp-content/uploads/2017/12/sei-2015-report-food-waste-and-food-

loss-in-estonian-households-and-catering-institutions-sei-tallinn1.pdf

Moora, H., Piirsalu, E., & Viilvere, T. (2015). Analysis of food waste in the Estonian food trade sector

and the food industry Summary (p. 2). Stockholm Environment Institute.

https://www.sei.org/wp-content/uploads/2017/12/summary-food-waste-in-estonian-food-

trade-sector-and-industry-sei-tallinn-2016.pdf

Mucyo, S. (2013). Analysis of Key Requirements for Effective Implementation of Biogas Technology

for Municipal Solid Waste Management in Sub-Saharan Africa. A Case Study of Kigali City,

Rwanda [PhD, Abertay University]. https://rke.abertay.ac.uk/en/studentTheses/analysis-of-

key-requirements-for-effective-implementation-of-biog

Nahman, A., de Lange, W., Oelofse, S., & Godfrey, L. (2012). The costs of household food waste in

South Africa. Waste Management, 32(11), 2147–2153.

https://doi.org/10.1016/j.wasman.2012.04.012

Oberlin, A. S. (2013). Characterization of Household Waste in Kinondoni Municipality, Dar Es Salaam.

Academic Journal of Interdisciplinary Studies. https://doi.org/10.5901/ajis.2013.v2n13p35

Oelofse, S., Muswema, A., & Ramukhwatho, F. (2018). Household food waste disposal in South

Africa: A case study of Johannesburg and Ekurhuleni. South African Journal of Science,

114(5/6). https://doi.org/10.17159/sajs.2018/20170284

Orhorhoro, E. K., Ebunilo, P. O., & Sadjere, G. E. (2017). Determination and Quantification of

Household Solid Waste Generation for Planning Suitable Sustainable Waste Management in

Nigeria. International Journal of Emerging Engineering Research and Technology, 5(8), 10.
Qu, X., Li, Z., Xie, X., Sui, Y., Yang, L., & Chen, Y. (2009). Survey of composition and generation rate of

household wastes in Beijing, China. Waste Management, 29(10), 2618–2624.

https://doi.org/10.1016/j.wasman.2009.05.014

Quested, T. E., Palmer, G., Moreno, L. C., McDermott, C., & Schumacher, K. (2020). Comparing

diaries and waste compositional analysis for measuring food waste in the home. Journal of

Cleaner Production, 262, 121263. https://doi.org/10.1016/j.jclepro.2020.121263

Ramakrishna, V. (2016). Municipal Solid Waste Quantification, Characterization and Management in

Rajam. The International Journal of Engineering and Science, 5(2), 8.

Ramukhwatho, F. R. (2016). An Assessment of the Household Food Wastage in a Developing Country:

A Case Study of Five Areas in the City of Tshwane Metropolitan Municipality, Guateng

Province, South Africa [University of South Africa].

https://www.semanticscholar.org/paper/An-assessment-of-the-household-food-wastage-in-

a-a-Ramukhwatho/6a1e01c9218dc0dd7b7cd6566fbcec3eda81bda0

Republic of Slovenia Statistical Office. (2019, October 22). Food Waste, Slovenia, 2018.

https://www.stat.si/StatWeb/en/news/Index/8433

Republic of Slovenia Statistical Office. (2020, November 26). Food waste and waste indicators,

Slovenia, 2019. https://www.stat.si/StatWeb/en/News/Index/9230

Reynolds, C., Soma, T., Spring, C., & Lazell, J. (Eds.). (2020). Routledge handbook of food waste.

Routledge/Taylor & Francis Group.

SAGO. (2019). Saudi FLW Baseline: Food Loss & Waste Index in Kingdom of Saudi Arabia. Saudi

Grains Organization. https://www.sago.gov.sa/Content/Files/Baseline_230719.pdf

Salam, M. A., Hossain, L., Das, S. R., Wahab, R., & Hossain, M. K. (2012). Generation and Assessing

the Composition of Household Solid Waste in Commercial Capital City of Bangladesh. 1, 12.

Schmidt, T., Schneider, F., Leverenz, D., & Hafner, G. (2019). Food waste in Germany – Baseline 2015

– Summary (No. 71). Thünen Institute.

 https://www.researchgate.net/publication/339473130_Food_waste_in_Germany_-

Baseline_2015_-_Summary

Silpa Kaza, Lisa C. Yao, Perinaz Bhada-Tata, & Frank Van Woerden. (2018). What a Waste 2.0: A

Global Snapshot of Solid Waste Management to 2050. World Bank.

https://openknowledge.worldbank.org/handle/10986/30317

Song, G., Li, M., Semakula, H. M., & Zhang, S. (2015). Food consumption and waste and the

embedded carbon, water and ecological footprints of households in China. Science of The

Total Environment, 529, 191–197. https://doi.org/10.1016/j.scitotenv.2015.05.068

Sosna, D., Brunclíková, L., & Galeta, P. (2019). Rescuing things: Food waste in the rural environment

in the Czech Republic. Journal of Cleaner Production, 214, 319–330.

https://doi.org/10.1016/j.jclepro.2018.12.214

SPREP (Ed.). (2016). Cleaner Pacific 2025: Pacific regional waste and pollution management strategy,

2016-2025. SPREP.

Steinhoff-Wrześniewska, A. (2015). The Pilot Study of Characteristics of Household Waste Generated

in Suburban Parts of Rural Areas. Journal of Ecological Engineering, 16, 92–100.

https://doi.org/10.12911/22998993/1862

Stenmarck, Å., Jensen, C., Quested, T., Moates, G., Buksti, M., Cseh, B., Juul, S., Parry, A., Politano, A.,

Redlingshofer, B., Scherhaufer, S., Silvennoinen, K., Soethoudt, H., Zübert, C., & Östergren, K.

(2016). Estimates of European food waste levels. http://edepot.wur.nl/378674

Stensgård, A. E., & Hanssen, O. J. (2016). Food Waste in Norway 2010-2015 (OR.17.16). Matvett

AS/ForMat Project. https://norsus.no/wp-content/uploads/or1716-format-sluttrapport-

english.pdf

Stensgård, A. E., Prestrud, K., Hanssen, O. J., & Callewaert, P. (2019). Food Waste in Norway: Report

on Key Figures, 2015-2018 (OR.32.19). Matvett AS.

https://www.matvett.no/uploads/documents/OR.32.19-Edible-food-waste-in-Norway-

Report-on-key-figures-2015-2018.pdf
Sujauddin, M., Huda, S. M. S., & Hoque, A. T. M. R. (2008). Household solid waste characteristics and

management in Chittagong, Bangladesh. Waste Management, 28(9), 1688–1695.

https://doi.org/10.1016/j.wasman.2007.06.013

Sulaymon, D. A. H., Ibraheem, D. J. A., & Graimed, B. H. (2010). Household Behavior on Solid Waste

Management a Case of Al-Kut City. Engineering and Technology Journal, 28(24), 11.

Sunshine Yates Consulting. (2018). New Zealand Food Waste Audits. Prepared for WasteMINZ.

https://lovefoodhatewaste.co.nz/wp-content/uploads/2019/02/Final-New-Zealand-Food-

Waste-Audits-2018.pdf

Suthar, S., & Singh, P. (2015). Household solid waste generation and composition in different family

size and socio-economic groups: A case study. Sustainable Cities and Society, 14, 56–63.

https://doi.org/10.1016/j.scs.2014.07.004

Swedish Environmental Protection Agency. (2014). Food waste volumes in Sweden 2012.

http://www.naturvardsverket.se/Documents/publikationer6400/978-91-620-8695-4.pdf

Swedish Environmental Protection Agency. (2020). Matavfall I Sverige: Uppkomst och benhandling

2018. https://www.naturvardsverket.se/Documents/publ-filer/8800/978-91-620-8861-

3.pdf?pid=26710

Takeuchi, N. (2019, September 18). Linkages with SDG 11.6.1 on MSW and composition analysis

[Unpublished UN Habitat presentation].

Thanh, N. P., Matsui, Y., & Fujiwara, T. (2010). Household solid waste generation and characteristic

in a Mekong Delta city, Vietnam. Journal of Environmental Management, 91(11), 2307–2321.

https://doi.org/10.1016/j.jenvman.2010.06.016

The Netherlands Nutrition Centre Foundation. (2019). Synthesis report on food waste in Dutch

households in 2019.

https://www.voedingscentrum.nl/Assets/Uploads/voedingscentrum/Documents/Profession

als/Pers/Persmappen/Verspilling%202019/VC_Synthesis%20report%20on%20food%20wast

e%20in%20Dutch%20households%202019.pdf
Tiarcenter. (2019). Foodsharing in Russia. https://tiarcenter.com/wp-

content/uploads/2019/11/ENG_Foodsharing-in-Russia_2019.pdf

U.S. Environmental Protection Agency. (2020). 2018 Wasted Food Report (p. 42).

https://www.epa.gov/sites/production/files/2020-

11/documents/2018_wasted_food_report-11-9-20_final_.pdf

van der Werf, P., & Gilliland, J. A. (2017). A systematic review of food losses and food waste

generation in developed countries. Proceedings of the Institution of Civil Engineers - Waste

and Resource Management, 170(2), 66–77. https://doi.org/10.1680/jwarm.16.00026

van Herpen, E., van der Lans, I. A., Holthuysen, N., Nijenhuis-de Vries, M., & Quested, T. E. (2019).

Comparing wasted apples and oranges: An assessment of methods to measure household

food waste. Waste Management, 88, 71–84. https://doi.org/10.1016/j.wasman.2019.03.013

Vetter-Gindele, J., Braun, A., Warth, G., Bui, T. T. Q., Bachofer, F., & Eltrop, L. (2019). Assessment of

Household Solid Waste Generation and Composition by Building Type in Da Nang, Vietnam.

Resources, 8(4), 171. https://doi.org/10.3390/resources8040171

Watanabe, K. (2012). The 3R Potential of Household Waste in Bangi, Malaysia. In Understanding

Confluences and Contestations, Continuities and Changes: Towards Transforming Society and

Empowering People (pp. 116–126).

https://www.researchgate.net/publication/280642994_The_3R_Potential_of_Household_W

aste_in_Bangi_Malaysia?enrichId=rgreq-7baab3bfdb1ebd5045b09cfee5cc62f4-

XXX&enrichSource=Y292ZXJQYWdlOzI4MDY0Mjk5NDtBUzoyNTg2ODczMzU5MjM3MTJAMT

QzODY4NzI3NTUyOQ%3D%3D&el=1_x_2&_esc=publicationCoverPdf

Withanage, S. V., Dias, G. M., & Habib. (2021). Review of household food waste quantification

methods: Focus on composition analysis. Journal of Cleaner Production, 15.

WRAP. (2018). Household food waste: Restated data for 2007-2015.

https://www.wrap.org.uk/sites/files/wrap/Household%20food%20waste%20restated%20da

ta%202007-2015%20FINAL.pdf
WRAP. (2020). UK Progress against Courtauld 2025 targets and UN Sustainable Development Goal

12.3 (p. 54).

https://wrap.org.uk/sites/files/wrap/Progress_against_Courtauld_2025_targets_and_UN_S

DG_123.pdf

Xue, L., Liu, G., Parfitt, J., Liu, X., Van Herpen, E., Stenmarck, Å., O’Connor, C., Östergren, K., & Cheng,

S. (2017). Missing Food, Missing Data? A Critical Review of Global Food Losses and Food

Waste Data. Environmental Science & Technology, 51(12), 6618–6633.

https://doi.org/10.1021/acs.est.7b00401

Yasir, R. A., & Abudi, Z. N. (2009). Characteristics and Compositions of Solid Waste in Nassiriya City.

Al-Qadisiya Journal for Engineering Sciences, 2, 13.

Zhang, H., Duan, H., Andric, J. M., Song, M., & Yang, B. (2018). Characterization of household food

waste and strategies for its reduction: A Shenzhen City case study. Waste Management, 78,

426–433. https://doi.org/10.1016/j.wasman.2018.06.010

Zhang, H., Liu, G., Xue, L., Zuo, J., Chen, T., Vuppaladadiyam, A., & Duan, H. (2020). Anaerobic

digestion based waste-to-energy technologies can halve the climate impact of China’s fast-

growing food waste by 2040. Journal of Cleaner Production, 277, 123490.

https://doi.org/10.1016/j.jclepro.2020.123490
4 Appendix: Level 1 data by country for all sectors
This section contains three tables – one for each sector – including estimates of food waste for all
countries. These estimates include those taken from one or more measured datapoints (classified as
High or Medium confidence) and those extrapolated from other countries’ data (classified as Low or
Very Low confidence).

It is important to note that the confidence rating in this report is not a judgement on the quality of
the study undertaken, It is an assessment – based on our understanding of the study – of how robust
the estimate of food waste is for tracking food waste in the given country. In many cases, this was
not an aim of the original study. Hence many good studies will be classified at a ‘Medium’
confidence level (or even excluded from consideration altogether) as the aims of the paper did not
include national food waste tracking.

4.1 Household
See Erreur ! Source du renvoi introuvable. in Section Erreur ! Source du renvoi introuvable..

4.2 Food Service

To briefly reiterate the methodology: the best available food waste data was collected, adjusted to
account for biases and improve comparability, and grouped into confidence ratings. Where
available, the average of these datapoints was applied for a country. Where not available, an
extrapolation was made based on the observed estimates. In Food Service, for HICs, the average
waste in HICs was used. Due to insufficient regional or income group data in UMC, LMC and LICs, the
global average is used in all other cases. The High and Medium confidence refer only to estimates
from the data collected. HIC extrapolations are Low confidence and all others are Very Low
confidence. This methodology is detailed in Appendix Section 1.

Table 13: Level 1 food-service food-waste estimates for each country

Food service Food service
M49 Confidence in
Region Country estimate estimate
code estimate
(kg/capita/year) (tonnes/year)

Australia and New Zealand 36 Australia 22 546,340 High Confidence

Australia and New Zealand 554 New Zealand 26 122,256 Low Confidence

Central Asia 398 Kazakhstan 28 512,760 Very Low Confidence

Central Asia 417 Kyrgyzstan 28 177,335 Very Low Confidence

Central Asia 762 Tajikistan 28 257,632 Very Low Confidence

Central Asia 795 Turkmenistan 28 164,239 Very Low Confidence

Central Asia 860 Uzbekistan 28 911,613 Very Low Confidence

Eastern Asia 156 China 46 65,377,741 High Confidence

Eastern Asia 344 China, Hong Kong SAR 26 190,070 Low Confidence

Eastern Asia 446 China, Macao SAR 26 16,369 Low Confidence

Dem. People's Rep.
Eastern Asia 408 Korea 28 709,413 Very Low Confidence

Eastern Asia 392 Japan 15 1,870,735 Medium Confidence

Eastern Asia 496 Mongolia 28 89,144 Very Low Confidence

Eastern Asia 410 Republic of Korea 26 1,309,322 Low Confidence

Eastern Europe 112 Belarus 28 261,264 Very Low Confidence

Eastern Europe 100 Bulgaria 28 193,483 Very Low Confidence

Eastern Europe 203 Czechia 26 273,217 Low Confidence

Eastern Europe 348 Hungary 26 247,542 Low Confidence

Eastern Europe 616 Poland 26 968,415 Low Confidence

Eastern Europe 498 Republic of Moldova 28 111,757 Very Low Confidence

Eastern Europe 642 Romania 26 494,961 Low Confidence

Eastern Europe 643 Russian Federation 28 4,031,907 Very Low Confidence

Eastern Europe 703 Slovakia 26 139,481 Low Confidence

Eastern Europe 804 Ukraine 28 1,215,982 Very Low Confidence

Latin America and the
* *
Caribbean 660 Anguilla No Estimate
Latin America and the
Caribbean 28 Antigua and Barbuda 26 2,482 Low Confidence
Latin America and the
Caribbean 32 Argentina 28 1,237,737 Very Low Confidence
Latin America and the
Caribbean 533 Aruba 26 2,717 Low Confidence
Latin America and the
Caribbean 44 Bahamas 26 9,956 Low Confidence
Latin America and the
Caribbean 52 Barbados 26 7,336 Low Confidence
Latin America and the
Caribbean 84 Belize 28 10,791 Very Low Confidence
Latin America and the
Caribbean 68 Bolivia (Plurin. State of) 28 318,222 Very Low Confidence
Latin America and the Bonaire, St. Eustatius &
* *
Caribbean 535 Saba No Estimate
Latin America and the
Caribbean 76 Brazil 28 5,833,403 Very Low Confidence
Latin America and the
Caribbean 92 British Virgin Islands 26 767 Low Confidence
Latin America and the
Caribbean 136 Cayman Islands 26 1,659 Low Confidence
Latin America and the
Caribbean 152 Chile 26 484,414 Low Confidence
Latin America and the
Caribbean 170 Colombia 28 1,391,380 Very Low Confidence
Latin America and the
Caribbean 188 Costa Rica 28 139,516 Very Low Confidence
Latin America and the
Caribbean 192 Cuba 28 313,258 Very Low Confidence
Latin America and the
Caribbean 531 Curaçao 26 4,177 Low Confidence
Latin America and the
Caribbean 212 Dominica 28 1,985 Very Low Confidence
Latin America and the
Caribbean 214 Dominican Republic 28 296,826 Very Low Confidence
Latin America and the
Caribbean 218 Ecuador 28 480,209 Very Low Confidence
Latin America and the
Caribbean 222 El Salvador 28 178,377 Very Low Confidence
Latin America and the Falkland Islands
* *
Caribbean 238 (Malvinas) No Estimate
Latin America and the
* *
Caribbean 254 French Guiana No Estimate
Latin America and the
Caribbean 308 Grenada 28 3,096 Very Low Confidence
Latin America and the
* *
Caribbean 312 Guadeloupe No Estimate
Latin America and the
Caribbean 320 Guatemala 28 485,952 Very Low Confidence
Latin America and the
Caribbean 328 Guyana 28 21,637 Very Low Confidence
Latin America and the
Caribbean 332 Haiti 28 311,312 Very Low Confidence
Latin America and the
Caribbean 340 Honduras 28 269,382 Very Low Confidence
Latin America and the
Caribbean 388 Jamaica 28 81,491 Very Low Confidence
Latin America and the
* *
Caribbean 474 Martinique No Estimate
Latin America and the
Caribbean 484 Mexico 28 3,526,184 Very Low Confidence
Latin America and the
* *
Caribbean 500 Montserrat No Estimate
Latin America and the
Caribbean 558 Nicaragua 28 180,917 Very Low Confidence
Latin America and the
Caribbean 591 Panama 26 108,538 Low Confidence
Latin America and the
Caribbean 600 Paraguay 28 194,713 Very Low Confidence
Latin America and the
Caribbean 604 Peru 28 898,589 Very Low Confidence
Latin America and the
Caribbean 630 Puerto Rico 26 74,978 Low Confidence
Latin America and the
* *
Caribbean 652 Saint Barthélemy No Estimate
Latin America and the
Caribbean 659 Saint Kitts and Nevis 26 1,350 Low Confidence
Latin America and the
Caribbean 662 Saint Lucia 28 5,053 Very Low Confidence
Latin America and the Saint Martin (French
Caribbean 663 part) 26 971 Low Confidence
Latin America and the Saint Vincent &
Caribbean 670 Grenadines 28 3,057 Very Low Confidence
Latin America and the Sint Maarten (Dutch
Caribbean 534 part) 26 1,084 Low Confidence
Latin America and the
Caribbean 740 Suriname 28 16,070 Very Low Confidence
Latin America and the
Caribbean 780 Trinidad and Tobago 26 35,656 Low Confidence
Latin America and the
Caribbean 796 Turks and Caicos Islands 26 976 Low Confidence
Latin America and the United States Virgin
Caribbean 850 Islands 26 2,674 Low Confidence
Latin America and the
Caribbean 858 Uruguay 26 88,481 Low Confidence
Latin America and the Venezuela (Boliv. Rep.
Caribbean 862 of) 28 788,176 Very Low Confidence

Melanesia 242 Fiji 28 24,600 Very Low Confidence

Melanesia 540 New Caledonia 26 7,228 Low Confidence

Melanesia 598 Papua New Guinea 28 242,571 Very Low Confidence

Melanesia 90 Solomon Islands 28 18,513 Very Low Confidence

Melanesia 548 Vanuatu 28 8,289 Very Low Confidence

Micronesia 316 Guam 26 4,276 Low Confidence

Micronesia 296 Kiribati 28 3,250 Very Low Confidence

Micronesia 584 Marshall Islands 28 1,625 Very Low Confidence

 Micronesia (Fed. States
Micronesia 583 of) 28 3,145 Very Low Confidence

Micronesia 520 Nauru 26 276 Low Confidence

Northern Mariana
Micronesia 580 Islands 26 1,462 Low Confidence

Micronesia 585 Palau 26 460 Low Confidence

Northern Africa 12 Algeria 28 1,189,987 Very Low Confidence

Northern Africa 818 Egypt 28 2,774,725 Very Low Confidence

Northern Africa 434 Libya 28 187,330 Very Low Confidence

Northern Africa 504 Morocco 28 1,008,080 Very Low Confidence

Northern Africa 729 Sudan 28 1,183,356 Very Low Confidence

Northern Africa 788 Tunisia 28 323,241 Very Low Confidence

Northern Africa 732 Western Sahara * * No Estimate

Northern America 60 Bermuda 26 1,598 Low Confidence

Northern America 124 Canada 26 956,228 Low Confidence

Northern America 304 Greenland 26 1,449 Low Confidence

Saint Pierre and
* *
Northern America 666 Miquelon No Estimate

Northern America 840 United States of America 64 20,934,827 High Confidence

Northern Europe 208 Denmark 21 119,134 High Confidence

Northern Europe 233 Estonia 17 22,013 High Confidence

Northern Europe 234 Faroe Islands 26 1,245 Low Confidence

Northern Europe 246 Finland 23 128,927 Medium Confidence

Northern Europe 352 Iceland 26 8,665 Low Confidence

Northern Europe 372 Ireland 56 274,135 Medium Confidence

Northern Europe 833 Isle of Man 26 2,162 Low Confidence

Northern Europe 428 Latvia 26 48,735 Low Confidence

Northern Europe 440 Lithuania 26 70,536 Low Confidence

Northern Europe 578 Norway 5 26,685 Medium Confidence

Northern Europe 752 Sweden 21 205,746 High Confidence

Northern Europe 826 United Kingdom 17 1,114,248 High Confidence

Polynesia 16 American Samoa 28 1,528 Very Low Confidence

Polynesia 184 Cook Islands * * No Estimate

Polynesia 258 French Polynesia 26 7,139 Low Confidence

Polynesia 570 Niue * * No Estimate

Polynesia 882 Samoa 28 5,448 Very Low Confidence

Polynesia 772 Tokelau * * No Estimate

Polynesia 776 Tonga 28 2,888 Very Low Confidence

Polynesia 798 Tuvalu 28 321 Very Low Confidence
Wallis and Futuna
* *
Polynesia 876 Islands No Estimate

South-eastern Asia 96 Brunei Darussalam 26 11,075 Low Confidence

South-eastern Asia 116 Cambodia 28 455,686 Very Low Confidence

South-eastern Asia 360 Indonesia 28 7,480,085 Very Low Confidence

South-eastern Asia 418 Lao People's Dem. Rep. 28 198,165 Very Low Confidence

South-eastern Asia 458 Malaysia 90 2,861,537 Medium Confidence

South-eastern Asia 104 Myanmar 28 1,493,814 Very Low Confidence

South-eastern Asia 608 Philippines 28 2,988,340 Very Low Confidence

South-eastern Asia 702 Singapore 26 148,358 Low Confidence

South-eastern Asia 764 Thailand 28 1,924,450 Very Low Confidence

South-eastern Asia 626 Timor-Leste 28 35,741 Very Low Confidence

South-eastern Asia 704 Viet Nam 28 2,666,210 Very Low Confidence

Southern Asia 4 Afghanistan 28 1,051,474 Very Low Confidence

Southern Asia 50 Bangladesh 3 544,436 Medium Confidence

Southern Asia 64 Bhutan 28 21,092 Very Low Confidence

Southern Asia 356 India 28 37,767,754 Very Low Confidence

Southern Asia 364 Iran (Islamic Republic of) 28 2,291,738 Very Low Confidence

Southern Asia 462 Maldives 28 14,677 Very Low Confidence

Southern Asia 524 Nepal 28 790,744 Very Low Confidence

Southern Asia 586 Pakistan 28 5,985,859 Very Low Confidence

Southern Asia 144 Sri Lanka 28 589,387 Very Low Confidence

Southern Europe 8 Albania 28 79,628 Very Low Confidence

Southern Europe 20 Andorra 26 1,971 Low Confidence

Southern Europe 70 Bosnia and Herzegovina 28 91,240 Very Low Confidence

Southern Europe 191 Croatia 26 105,571 Low Confidence

Southern Europe 292 Gibraltar 26 861 Low Confidence

Southern Europe 300 Greece 26 267,703 Low Confidence

Southern Europe 336 Holy See * * No Estimate

Southern Europe 380 Italy 26 1,547,665 Low Confidence

Southern Europe 470 Malta 26 11,257 Low Confidence

Southern Europe 499 Montenegro 28 17,358 Very Low Confidence

Southern Europe 807 North Macedonia 28 57,588 Very Low Confidence

Southern Europe 620 Portugal 26 261,382 Low Confidence

Southern Europe 674 San Marino 26 866 Low Confidence

Southern Europe 688 Serbia 6 52,633 Medium Confidence

Southern Europe 705 Slovenia 20 41,354 Medium Confidence

Southern Europe 724 Spain 26 1,194,596 Low Confidence

Sub-Saharan Africa 24 Angola 28 879,650 Very Low Confidence

Sub-Saharan Africa 204 Benin 28 326,185 Very Low Confidence

Sub-Saharan Africa 72 Botswana 28 63,674 Very Low Confidence

Sub-Saharan Africa 854 Burkina Faso 28 561,683 Very Low Confidence

Sub-Saharan Africa 108 Burundi 28 318,705 Very Low Confidence

Sub-Saharan Africa 132 Cabo Verde 28 15,199 Very Low Confidence

Sub-Saharan Africa 120 Cameroon 28 715,223 Very Low Confidence

Sub-Saharan Africa 140 Central African Republic 28 131,157 Very Low Confidence

Sub-Saharan Africa 148 Chad 28 440,772 Very Low Confidence

Sub-Saharan Africa 174 Comoros 28 23,519 Very Low Confidence

Sub-Saharan Africa 178 Congo 28 148,717 Very Low Confidence

Sub-Saharan Africa 384 Côte d’Ivoire 28 710,803 Very Low Confidence

Sub-Saharan Africa 180 Dem. Rep. of the Congo 28 2,398,890 Very Low Confidence

Sub-Saharan Africa 262 Djibouti 28 26,910 Very Low Confidence

Sub-Saharan Africa 226 Equatorial Guinea 28 37,480 Very Low Confidence

Sub-Saharan Africa 232 Eritrea 28 96,660 Very Low Confidence

Sub-Saharan Africa 748 Eswatini 28 31,733 Very Low Confidence

Sub-Saharan Africa 231 Ethiopia 28 3,097,852 Very Low Confidence

Sub-Saharan Africa 266 Gabon 28 60,051 Very Low Confidence

Sub-Saharan Africa 270 Gambia 28 64,890 Very Low Confidence

Sub-Saharan Africa 288 Ghana 28 840,750 Very Low Confidence

Sub-Saharan Africa 324 Guinea 28 352,996 Very Low Confidence

Sub-Saharan Africa 624 Guinea-Bissau 28 53,094 Very Low Confidence

Sub-Saharan Africa 404 Kenya 31 1,637,020 Medium Confidence

Sub-Saharan Africa 426 Lesotho 28 58,743 Very Low Confidence

Sub-Saharan Africa 430 Liberia 28 136,470 Very Low Confidence

Sub-Saharan Africa 450 Madagascar 28 745,431 Very Low Confidence

Sub-Saharan Africa 454 Malawi 28 514,897 Very Low Confidence

Sub-Saharan Africa 466 Mali 28 543,347 Very Low Confidence

Sub-Saharan Africa 478 Mauritania 28 125,090 Very Low Confidence

Sub-Saharan Africa 480 Mauritius 26 32,454 Low Confidence

Sub-Saharan Africa 175 Mayotte * * No Estimate
Sub-Saharan Africa 508 Mozambique 28 839,315 Very Low Confidence

Sub-Saharan Africa 516 Namibia 28 68,948 Very Low Confidence

Sub-Saharan Africa 562 Niger 28 644,307 Very Low Confidence

Sub-Saharan Africa 566 Nigeria 28 5,554,629 Very Low Confidence

Sub-Saharan Africa 638 Réunion * * No Estimate

Sub-Saharan Africa 646 Rwanda 28 349,010 Very Low Confidence

Sub-Saharan Africa 654 Saint Helena * * No Estimate

Sub-Saharan Africa 678 Sao Tome and Principe 28 5,945 Very Low Confidence

Sub-Saharan Africa 686 Senegal 28 450,432 Very Low Confidence

Sub-Saharan Africa 690 Seychelles 26 2,497 Low Confidence

Sub-Saharan Africa 694 Sierra Leone 28 215,957 Very Low Confidence

Sub-Saharan Africa 706 Somalia 28 426,841 Very Low Confidence

Sub-Saharan Africa 710 South Africa 28 1,618,550 Very Low Confidence

Sub-Saharan Africa 728 South Sudan 28 305,756 Very Low Confidence

Sub-Saharan Africa 768 Togo 28 223,397 Very Low Confidence

Sub-Saharan Africa 800 Uganda 28 1,223,611 Very Low Confidence

Sub-Saharan Africa 834 United Rep. of Tanzania 28 1,603,271 Very Low Confidence

Sub-Saharan Africa 894 Zambia 28 493,678 Very Low Confidence

Sub-Saharan Africa 716 Zimbabwe 28 404,801 Very Low Confidence

Western Asia 51 Armenia 28 81,751 Very Low Confidence

Western Asia 31 Azerbaijan 28 277,718 Very Low Confidence

Western Asia 48 Bahrain 26 41,949 Low Confidence

Western Asia 196 Cyprus 26 30,636 Low Confidence

Western Asia 268 Georgia 28 110,471 Very Low Confidence

Western Asia 368 Iraq 28 1,086,522 Very Low Confidence

Western Asia 376 Israel 27 233,752 Medium Confidence

Western Asia 400 Jordan 28 279,211 Very Low Confidence

Western Asia 414 Kuwait 26 107,534 Low Confidence

Western Asia 422 Lebanon 28 189,491 Very Low Confidence

Western Asia 512 Oman 26 127,161 Low Confidence

Western Asia 634 Qatar 26 72,389 Low Confidence

Western Asia 682 Saudi Arabia 26 875,905 Low Confidence

Western Asia 275 State of Palestine 28 137,686 Very Low Confidence

Western Asia 760 Syrian Arab Republic 28 471,817 Very Low Confidence

Western Asia 792 Turkey 28 2,305,992 Very Low Confidence

Western Asia 784 United Arab Emirates 26 249,735 Low Confidence

Western Asia 887 Yemen 28 806,034 Very Low Confidence

Western Europe 40 Austria 28 254,191 High Confidence

Western Europe 56 Belgium 20 227,371 Medium Confidence

Western Europe 250 France 24 1,594,579 Medium Confidence

Western Europe 276 Germany 21 1,718,433 High Confidence

Western Europe 438 Liechtenstein 26 971 Low Confidence

Western Europe 442 Luxembourg 21 12,868 Medium Confidence

Western Europe 492 Monaco 26 997 Low Confidence

Western Europe 528 Netherlands 26 437,003 Low Confidence

Western Europe 756 Switzerland 40 343,656 Medium Confidence

830 Channel Islands * * No Estimate
Other non-specified
* *
158 areas No Estimate
4.3 Retail
To briefly reiterate the methodology: the best available food waste data was collected, adjusted to
account for biases and improve comparability, and grouped into confidence ratings. Where
available, the average of these datapoints was applied for a country. Where not available, an
extrapolation was made based on the observed estimates. In Retail, for HICs, the average waste in
HICs was used. Due to insufficient regional or income group data in UMC, LMC and LICs, the global
average is used in all other cases. The High and Medium confidence refer only to estimates from the
data collected. HIC extrapolations are Low confidence and all others are Very Low confidence. This
methodology is detailed in Appendix Section 1.

Table 14: Level 1 retail food-waste estimates for each country

Retail estimate Retail estimate Confidence in
Region M49 code Country
(kg/capita/year) (tonnes/year) estimate

Australia and New Zealand 36 Australia 9 238,248 High Confidence

Australia and New Zealand 554 New Zealand 3 14,923 High Confidence

Central Asia 398 Kazakhstan 16 290,121 Very Low Confidence

Central Asia 417 Kyrgyzstan 16 100,337 Very Low Confidence

Central Asia 762 Tajikistan 16 145,769 Very Low Confidence

Central Asia 795 Turkmenistan 16 92,927 Very Low Confidence

Central Asia 860 Uzbekistan 16 515,793 Very Low Confidence

Eastern Asia 156 China 16 22,422,617 Very Low Confidence

Eastern Asia 344 China, Hong Kong SAR 13 95,255 Low Confidence

Eastern Asia 446 China, Macao SAR 13 8,203 Low Confidence

Dem. People's Rep.
Eastern Asia 408 Korea 16 401,388 Very Low Confidence

Eastern Asia 392 Japan 9 1,095,308 Medium Confidence

Eastern Asia 496 Mongolia 16 50,438 Very Low Confidence

Eastern Asia 410 Republic of Korea 13 656,177 Low Confidence

Eastern Europe 112 Belarus 16 147,824 Very Low Confidence

Eastern Europe 100 Bulgaria 16 109,473 Very Low Confidence

Eastern Europe 203 Czechia 13 136,925 Low Confidence

Eastern Europe 348 Hungary 13 124,057 Low Confidence

Eastern Europe 616 Poland 13 485,329 Low Confidence

Eastern Europe 498 Republic of Moldova 16 63,232 Very Low Confidence

Eastern Europe 642 Romania 13 248,053 Low Confidence

Eastern Europe 643 Russian Federation 14 2,001,062 Medium Confidence

Eastern Europe 703 Slovakia 13 69,902 Low Confidence

Eastern Europe 804 Ukraine 16 688,006 Very Low Confidence

Latin America and the
* *
Caribbean 660 Anguilla No Estimate
Latin America and the
Caribbean 28 Antigua and Barbuda 13 1,244 Low Confidence
Latin America and the
Caribbean 32 Argentina 16 700,315 Very Low Confidence
Latin America and the
Caribbean 533 Aruba 13 1,362 Low Confidence
Latin America and the
Caribbean 44 Bahamas 13 4,989 Low Confidence
Latin America and the
Caribbean 52 Barbados 13 3,676 Low Confidence
Latin America and the
Caribbean 84 Belize 16 6,105 Very Low Confidence
Latin America and the
Caribbean 68 Bolivia (Plurin. State of) 16 180,051 Very Low Confidence
Latin America and the Bonaire, St. Eustatius &
* *
Caribbean 535 Saba No Estimate
Latin America and the
Caribbean 76 Brazil 16 3,300,555 Very Low Confidence
Latin America and the
Caribbean 92 British Virgin Islands 13 384 Low Confidence
Latin America and the
Caribbean 136 Cayman Islands 13 831 Low Confidence
Latin America and the
Caribbean 152 Chile 13 242,768 Low Confidence
Latin America and the
Caribbean 170 Colombia 16 787,246 Very Low Confidence
Latin America and the
Caribbean 188 Costa Rica 16 78,938 Very Low Confidence
Latin America and the
Caribbean 192 Cuba 16 177,242 Very Low Confidence
Latin America and the
Caribbean 531 Curaçao 13 2,093 Low Confidence
Latin America and the
Caribbean 212 Dominica 16 1,123 Very Low Confidence
Latin America and the
Caribbean 214 Dominican Republic 16 167,945 Very Low Confidence
Latin America and the
Caribbean 218 Ecuador 16 271,703 Very Low Confidence
Latin America and the
Caribbean 222 El Salvador 16 100,926 Very Low Confidence
Latin America and the Falkland Islands
* *
Caribbean 238 (Malvinas) No Estimate
Latin America and the
* *
Caribbean 254 French Guiana No Estimate
Latin America and the
Caribbean 308 Grenada 16 1,752 Very Low Confidence
Latin America and the
* *
Caribbean 312 Guadeloupe No Estimate
Latin America and the
Caribbean 320 Guatemala 16 274,953 Very Low Confidence
Latin America and the
Caribbean 328 Guyana 16 12,242 Very Low Confidence
Latin America and the
Caribbean 332 Haiti 16 176,141 Very Low Confidence
Latin America and the
Caribbean 340 Honduras 16 152,417 Very Low Confidence
Latin America and the
Caribbean 388 Jamaica 16 46,108 Very Low Confidence
Latin America and the
* *
Caribbean 474 Martinique No Estimate
Latin America and the
Caribbean 484 Mexico 16 1,995,124 Very Low Confidence
Latin America and the
* *
Caribbean 500 Montserrat No Estimate
Latin America and the
Caribbean 558 Nicaragua 16 102,364 Very Low Confidence
Latin America and the
Caribbean 591 Panama 13 54,395 Low Confidence
Latin America and the
Caribbean 600 Paraguay 16 110,169 Very Low Confidence
Latin America and the
Caribbean 604 Peru 16 508,424 Very Low Confidence
Latin America and the
Caribbean 630 Puerto Rico 13 37,576 Low Confidence
Latin America and the
* *
Caribbean 652 Saint Barthélemy No Estimate
Latin America and the
Caribbean 659 Saint Kitts and Nevis 13 676 Low Confidence
Latin America and the
Caribbean 662 Saint Lucia 16 2,859 Very Low Confidence
Latin America and the Saint Martin (French
Caribbean 663 part) 13 487 Low Confidence
Latin America and the Saint Vincent &
Caribbean 670 Grenadines 16 1,730 Very Low Confidence
Latin America and the Sint Maarten (Dutch
Caribbean 534 part) 13 543 Low Confidence
Latin America and the
Caribbean 740 Suriname 16 9,092 Very Low Confidence
Latin America and the
Caribbean 780 Trinidad and Tobago 13 17,869 Low Confidence
Latin America and the
Caribbean 796 Turks and Caicos Islands 13 489 Low Confidence
Latin America and the United States Virgin
Caribbean 850 Islands 13 1,340 Low Confidence
Latin America and the
Caribbean 858 Uruguay 13 44,343 Low Confidence
Latin America and the Venezuela (Boliv. Rep.
Caribbean 862 of) 16 445,952 Very Low Confidence

Melanesia 242 Fiji 16 13,919 Very Low Confidence

Melanesia 540 New Caledonia 13 3,623 Low Confidence

Melanesia 598 Papua New Guinea 16 137,247 Very Low Confidence

Melanesia 90 Solomon Islands 16 10,475 Very Low Confidence

Melanesia 548 Vanuatu 16 4,690 Very Low Confidence

Micronesia 316 Guam 13 2,143 Low Confidence

Micronesia 296 Kiribati 16 1,839 Very Low Confidence

Micronesia 584 Marshall Islands 16 920 Very Low Confidence

Micronesia (Fed. States
Micronesia 583 of) 16 1,780 Very Low Confidence

Micronesia 520 Nauru 13 138 Low Confidence

Northern Mariana
Micronesia 580 Islands 13 733 Low Confidence

Micronesia 585 Palau 13 231 Low Confidence

Northern Africa 12 Algeria 16 673,298 Very Low Confidence

Northern Africa 818 Egypt 16 1,569,947 Very Low Confidence

Northern Africa 434 Libya 16 105,992 Very Low Confidence

Northern Africa 504 Morocco 16 570,374 Very Low Confidence

Northern Africa 729 Sudan 16 669,546 Very Low Confidence

Northern Africa 788 Tunisia 16 182,891 Very Low Confidence

Northern Africa 732 Western Sahara * * No Estimate

Northern America 60 Bermuda 13 801 Low Confidence

Northern America 124 Canada 13 479,221 Low Confidence

Northern America 304 Greenland 13 726 Low Confidence

Saint Pierre and
* *
Northern America 666 Miquelon No Estimate

Northern America 840 United States of America 16 5,151,313 High Confidence

Northern Europe 208 Denmark 30 172,003 High Confidence

Northern Europe 233 Estonia 5 6,243 Medium Confidence

Northern Europe 234 Faroe Islands 13 624 Low Confidence

Northern Europe 246 Finland 13 70,865 Low Confidence

Northern Europe 352 Iceland 13 4,342 Low Confidence

Northern Europe 372 Ireland 13 62,543 Low Confidence

Northern Europe 833 Isle of Man 13 1,084 Low Confidence

Northern Europe 428 Latvia 13 24,424 Low Confidence

Northern Europe 440 Lithuania 13 35,349 Low Confidence

Northern Europe 578 Norway 14 74,088 Medium Confidence

Northern Europe 752 Sweden 10 100,364 High Confidence

Northern Europe 826 United Kingdom 4 283,627 High Confidence

Polynesia 16 American Samoa 16 865 Very Low Confidence

Polynesia 184 Cook Islands * * No Estimate

Polynesia 258 French Polynesia 13 3,578 Low Confidence

Polynesia 570 Niue * * No Estimate

Polynesia 882 Samoa 16 3,082 Very Low Confidence

Polynesia 772 Tokelau * * No Estimate

Polynesia 776 Tonga 16 1,634 Very Low Confidence

Polynesia 798 Tuvalu 16 181 Very Low Confidence

Wallis and Futuna
* *
Polynesia 876 Islands No Estimate

South-eastern Asia 96 Brunei Darussalam 13 5,550 Low Confidence

South-eastern Asia 116 Cambodia 16 257,829 Very Low Confidence

South-eastern Asia 360 Indonesia 16 4,232,252 Very Low Confidence

South-eastern Asia 418 Lao People's Dem. Rep. 16 112,122 Very Low Confidence

South-eastern Asia 458 Malaysia 79 2,518,199 Medium Confidence

South-eastern Asia 104 Myanmar 16 845,204 Very Low Confidence

South-eastern Asia 608 Philippines 16 1,690,811 Very Low Confidence

South-eastern Asia 702 Singapore 13 74,351 Low Confidence

South-eastern Asia 764 Thailand 16 1,088,859 Very Low Confidence

South-eastern Asia 626 Timor-Leste 16 20,222 Very Low Confidence

South-eastern Asia 704 Viet Nam 16 1,508,549 Very Low Confidence

Southern Asia 4 Afghanistan 16 594,927 Very Low Confidence

Southern Asia 50 Bangladesh 16 2,549,842 Very Low Confidence

Southern Asia 64 Bhutan 16 11,934 Very Low Confidence

Southern Asia 356 India 16 21,369,097 Very Low Confidence

Southern Asia 364 Iran (Islamic Republic of) 16 1,296,672 Very Low Confidence

Southern Asia 462 Maldives 16 8,304 Very Low Confidence

Southern Asia 524 Nepal 16 447,405 Very Low Confidence

Southern Asia 586 Pakistan 16 3,386,815 Very Low Confidence

Southern Asia 144 Sri Lanka 16 333,476 Very Low Confidence

Southern Europe 8 Albania 16 45,054 Very Low Confidence

Southern Europe 20 Andorra 13 988 Low Confidence

Southern Europe 70 Bosnia and Herzegovina 16 51,624 Very Low Confidence

Southern Europe 191 Croatia 13 52,908 Low Confidence

Southern Europe 292 Gibraltar 13 432 Low Confidence

Southern Europe 300 Greece 7 77,332 Medium Confidence

Southern Europe 336 Holy See * * No Estimate

Southern Europe 380 Italy 4 219,552 High Confidence

Southern Europe 470 Malta 13 5,641 Low Confidence

Southern Europe 499 Montenegro 16 9,821 Very Low Confidence

Southern Europe 807 North Macedonia 16 32,583 Very Low Confidence

Southern Europe 620 Portugal 13 130,994 Low Confidence

Southern Europe 674 San Marino 13 434 Low Confidence

Southern Europe 688 Serbia 16 137,186 Very Low Confidence

Southern Europe 705 Slovenia 7 13,771 Medium Confidence

Southern Europe 724 Spain 13 598,681 Low Confidence

Sub-Saharan Africa 24 Angola 16 497,709 Very Low Confidence

Sub-Saharan Africa 204 Benin 16 184,556 Very Low Confidence

Sub-Saharan Africa 72 Botswana 16 36,027 Very Low Confidence

Sub-Saharan Africa 854 Burkina Faso 16 317,802 Very Low Confidence

Sub-Saharan Africa 108 Burundi 16 180,324 Very Low Confidence

Sub-Saharan Africa 132 Cabo Verde 16 8,600 Very Low Confidence

Sub-Saharan Africa 120 Cameroon 16 404,675 Very Low Confidence

Sub-Saharan Africa 140 Central African Republic 16 74,209 Very Low Confidence

Sub-Saharan Africa 148 Chad 16 249,390 Very Low Confidence

Sub-Saharan Africa 174 Comoros 16 13,307 Very Low Confidence

Sub-Saharan Africa 178 Congo 16 84,144 Very Low Confidence

Sub-Saharan Africa 384 Côte d’Ivoire 16 402,174 Very Low Confidence

Sub-Saharan Africa 180 Dem. Rep. of the Congo 16 1,357,298 Very Low Confidence

Sub-Saharan Africa 262 Djibouti 16 15,226 Very Low Confidence

Sub-Saharan Africa 226 Equatorial Guinea 16 21,206 Very Low Confidence

Sub-Saharan Africa 232 Eritrea 16 54,690 Very Low Confidence

Sub-Saharan Africa 748 Eswatini 16 17,955 Very Low Confidence

Sub-Saharan Africa 231 Ethiopia 16 1,752,773 Very Low Confidence

Sub-Saharan Africa 266 Gabon 16 33,977 Very Low Confidence

Sub-Saharan Africa 270 Gambia 16 36,715 Very Low Confidence

Sub-Saharan Africa 288 Ghana 16 475,699 Very Low Confidence

Sub-Saharan Africa 324 Guinea 16 199,726 Very Low Confidence

Sub-Saharan Africa 624 Guinea-Bissau 16 30,041 Very Low Confidence

Sub-Saharan Africa 404 Kenya 11 576,411 Medium Confidence

Sub-Saharan Africa 426 Lesotho 16 33,237 Very Low Confidence

Sub-Saharan Africa 430 Liberia 16 77,215 Very Low Confidence

Sub-Saharan Africa 450 Madagascar 16 421,767 Very Low Confidence

Sub-Saharan Africa 454 Malawi 16 291,330 Very Low Confidence

Sub-Saharan Africa 466 Mali 16 307,427 Very Low Confidence

Sub-Saharan Africa 478 Mauritania 16 70,776 Very Low Confidence

Sub-Saharan Africa 480 Mauritius 13 16,264 Low Confidence

Sub-Saharan Africa 175 Mayotte * * No Estimate

Sub-Saharan Africa 508 Mozambique 16 474,887 Very Low Confidence

Sub-Saharan Africa 516 Namibia 16 39,011 Very Low Confidence

Sub-Saharan Africa 562 Niger 16 364,551 Very Low Confidence

Sub-Saharan Africa 566 Nigeria 16 3,142,824 Very Low Confidence

Sub-Saharan Africa 638 Réunion * * No Estimate

Sub-Saharan Africa 646 Rwanda 16 197,471 Very Low Confidence

Sub-Saharan Africa 654 Saint Helena * * No Estimate

Sub-Saharan Africa 678 Sao Tome and Principe 16 3,364 Very Low Confidence

Sub-Saharan Africa 686 Senegal 16 254,856 Very Low Confidence

Sub-Saharan Africa 690 Seychelles 13 1,252 Low Confidence

Sub-Saharan Africa 694 Sierra Leone 16 122,189 Very Low Confidence

Sub-Saharan Africa 706 Somalia 16 241,508 Very Low Confidence

Sub-Saharan Africa 710 South Africa 16 915,780 Very Low Confidence

Sub-Saharan Africa 728 South Sudan 16 172,998 Very Low Confidence

Sub-Saharan Africa 768 Togo 16 126,399 Very Low Confidence

Sub-Saharan Africa 800 Uganda 16 692,322 Very Low Confidence

Sub-Saharan Africa 834 United Rep. of Tanzania 16 907,135 Very Low Confidence

Sub-Saharan Africa 894 Zambia 16 279,324 Very Low Confidence

Sub-Saharan Africa 716 Zimbabwe 16 229,038 Very Low Confidence

Western Asia 51 Armenia 16 46,255 Very Low Confidence

Western Asia 31 Azerbaijan 16 157,134 Very Low Confidence

Western Asia 48 Bahrain 13 21,023 Low Confidence

Western Asia 196 Cyprus 13 15,354 Low Confidence

Western Asia 268 Georgia 16 62,505 Very Low Confidence

Western Asia 368 Iraq 16 614,757 Very Low Confidence

Western Asia 376 Israel 51 437,997 Medium Confidence

Western Asia 400 Jordan 16 157,978 Very Low Confidence

Western Asia 414 Kuwait 13 53,891 Low Confidence

Western Asia 422 Lebanon 16 107,215 Very Low Confidence

Western Asia 512 Oman 13 63,728 Low Confidence

Western Asia 634 Qatar 13 36,278 Low Confidence

Western Asia 682 Saudi Arabia 20 673,502 High Confidence

Western Asia 275 State of Palestine 16 77,903 Very Low Confidence

Western Asia 760 Syrian Arab Republic 16 266,955 Very Low Confidence

Western Asia 792 Turkey 16 1,304,737 Very Low Confidence

Western Asia 784 United Arab Emirates 13 125,157 Low Confidence

Western Asia 887 Yemen 16 456,056 Very Low Confidence

Western Europe 40 Austria 9 77,289 High Confidence

Western Europe 56 Belgium 10 112,100 Medium Confidence

Western Europe 250 France 26 1,667,568 Medium Confidence

Western Europe 276 Germany 6 498,244 High Confidence

Western Europe 438 Liechtenstein 13 487 Low Confidence

Western Europe 442 Luxembourg 7 4,454 Medium Confidence

Western Europe 492 Monaco 13 500 Low Confidence

Western Europe 528 Netherlands 11 188,068 Medium Confidence

Western Europe 756 Switzerland 13 110,053 Low Confidence

5 Appendix: Measurement methods appropriate for each sector
The following methods have been deemed appropriate for each relevant sector for Level 2 and Level 3.

Table 15: Measurement methods for Manufacturing

Waste stream Appropriate measurement methods Appropriate means for national government to obtain the
measurements from companies

Use of records specifying volume or mass e.g., from

Food waste in a container (single waste contractor
stream – not mixed with other wastes) Volume assessment
Weighing, of whole containers or samples
Weighing, via waste composition analysis or trial
Food waste in a container (mixed with weighings
other wastes) Use of nationally held records e.g., regulatory returns
Volume assessment
Audit (face-to-face survey) to take measurements
Uncontained food waste (not mixed Weighing, of samples or entire stream depending on
feasibility Self-completion or telephone survey – to request/require provision of
with other wastes and not discharged
measurement data
to sewer) Volume assessment
Data provision as part of a framework to tackle food waste (e.g., a
Use of biological / chemical oxygen demand (BOD and voluntary agreement)
COD), suspended solids (SS). For further advice see:
Waste discharged to sewer (for Level 3)
https://www.wrap.org.uk/sites/files/wrap/food-waste-
in-effluent-guidelines_1.pdf

Waste coefficients applied to material flow

All waste streams
Mass balance (i.e., inputs minus outputs)

It is possible that food manufacture companies keep records of their waste already. Companies may call it something other than waste e.g., leakage,
slippage, residue, etc. Therefore, a degree of relationship building and understanding between governments and food manufacturers/processors in the
country may need to be built before either understands whether it is possible to use company records to build a national picture.

Informal food processing may not be at the scale necessary to quantify under 12.3.1(b) but this should be an informed decision. It is possible that informal
processing occurs on farm or in some households as local business in rural areas. Food removed from the human supply chain in those cases may either be
picked up in 12.3.1(b) or as part of in-home consumption under ‘household’ studies. If the latter, it may be useful to use diaries or surveys to determine
how much food waste is likely to be discarded for that reason.
Table 16: Measurement methods for Retail

Waste stream Appropriate measurement Appropriate means for national government to obtain the measurements from
methods companies

Use of records specifying

volume or mass e.g., from
waste contractor (direct
measurement)
Food waste in a container (single stream Waste composition analysis
– not mixed with other wastes) Scanning items as they are
wasted Use of nationally held records e.g., regulatory returns
Volume assessment Audit (face-to-face survey) to take measurements
Weighing, of whole Self-completion or telephone survey – to request/require provision of measurement data
containers or samples Data provision as part of a framework to tackle food waste (e.g., a voluntary agreement)
Use of records specifying
volume or mass e.g., from
waste contractor (direct
Food waste in a container (mixed with measurement)
other wastes)
Waste composition analysis
Scanning items as they are
wasted

The methods appropriate for formal and informal retail differ slightly. First, informal retail is unlikely to keep records so weighing or volume assessments
are necessary. Secondly, the manner of scaling any measurements for informal retail is likely to be difficult. If informal retail is a large proportion of food
retail in a country, an effort will have to be made to quantify the number and type of informal food retailers across different geographic areas. This will help
to determine a sample frame for the measurement studies and provide the basis for scaling. However, it is likely that the study on number and type of
informal retailers will need to be repeated as a country’s retail market changes between reporting periods.
Table 17: Measurement methods for Food Service

Waste stream Appropriate measurement methods Appropriate means for national government to obtain the
measurements from companies
Food waste in a container (single Use of records specifying volume or mass e.g., from
stream – not mixed with other wastes) waste contractor
Scanning items as they are wasted

Food waste in a food waste-only Volume assessment

container shared with other businesses Weighing, of whole containers or samples
or households Intercepting waste when shared with other businesses or Use of nationally held records e.g., regulatory returns
households
Audit (face-to-face survey) to take measurements
Self-completion or telephone survey – to request/require provision of
Food waste in a container (mixed with Weighing, via waste composition analysis or trial measurement data
other wastes) weighing Data provision as part of a framework to tackle food waste (e.g., a
Volume assessment voluntary agreement)
Food waste in a container mixed with
other wastes and shared with other Intercepting waste when shared with other businesses or
businesses or households households

Uncontained food waste (not mixed Weighing, via waste composition analysis or trial
with other wastes and not discharged weighing
to sewer) Volume assessment

The diversity of entity types within this sector is such that records are unlikely to cover them all. Larger public establishments like hospitals or schools may
have records or can be more easily regulated than private organisations. The restaurant sector is likely to be diverse and made up of majority small and
medium enterprises, many of which may be informal in certain countries. Appropriate methods for measurement are therefore likely to be volume
assessments or weighing in a sample study over a series of site visits. The same challenges for scaling such measurement studies apply here as for informal
retail; getting as accurate an understanding of the quantity of waste-producing entities as possible is as important as the measurement study and not likely
to be easy. This is directly linked to SDG 11.6.1 and could be measured as part of a waste composition analysis.
Table 18: Measurement methods for Household Sector

Waste stream Appropriate measurement methods Appropriate means for national government to obtain the
measurements from relevant organisations

Use of records specifying volume or mass e.g., from

waste contractor
Food waste in a container (single
Volume assessment
stream – not mixed with other wastes)
Weighing, of whole containers or samples
Food waste diaries

Food waste in a container (mixed with Weighing, via waste composition analysis or trial Commission organisation to conduct studies and scale up on behalf of
other wastes) weighing (linked with SDG 11.6.1) governments
Weighing, via waste composition analysis or trial Directly commission studies and maintain oversight of estimates
Uncontained food waste (not mixed weighing (linked with SDG 11.6.1)
with other wastes and not discharged
Diaries
to sewer)
Volume assessment
Waste discharged to sewer (for Level 3)
Diaries
and food home composted, animal
Diversion and weighing
feed
Methods most appropriate for household food waste vary by the destination of that waste. If generation and collection are equivalent, then a synthesis of
waste composition analyses of samples of collected waste from around the country with the total waste collected figure can give a relatively accurate
picture of food waste generated in the home without conducting a household study. However, this will ignore the amount of waste composted at home.
These amounts, if likely to be a smaller part of the waste stream, are likely best quantified by a diary study and scaled via population demographic statistics
e.g., number of households. If they are likely to be a larger part of the food waste generated from households, a direct measurement study may be more
appropriate using in-home observers or measurement devices. This is directly linked to SDG 11.6.1 and could be measured as part of a waste composition
analysis.
6 Appendix: Destinations and food waste
Table 19: Definitions of food-waste destinations

Classified as
food waste for
Destination Definition
the purposes
of the FWI
Diverting material from the food supply chain 9 (directly or after
Animal feed N
processing) to animals.
Converting material into industrial products for food and non-food
purposes. Examples include creating fibres for packaging material;
Bio-based creating bioplastics (e.g., polylactic acid); making “traditional”
materials/ materials such as leather or feathers (e.g., for pillows); and rendering
N
biochemical fat, oil, or grease into a raw material to make products such as soaps,
processing biodiesel, or cosmetics. “Biochemical processing” does not refer to
anaerobic digestion or production of bioethanol through
fermentation.
Breaking down material via bacteria in the absence of oxygen. This
process generates biogas and nutrient-rich matter. Co-digestion refers
Codigestion/ to the simultaneous anaerobic digestion of FLW and other organic
anaerobic material in one digester. This destination includes fermentation Y
digestion (converting carbohydrates – such as glucose, fructose, and sucrose –
via microbes into alcohols in the absence of oxygen to create products
such as biofuels).
Composting/ Breaking down material via bacteria in oxygen-rich environments.
aerobic Composting refers to the production of organic material (via aerobic Y
processes processes) that can be used as a soil amendment.
Sending material to a facility that is specifically designed for
Controlled combustion in a controlled manner, which may include some form of
Y
combustion energy recovery (this may also be referred to as incineration or
thermal treatment).
Land Spreading, spraying, injecting, or incorporating organic material onto
Y
application or below the surface of the land to enhance soil quality.
Sending material to an area of land or an excavated site that is
Landfill Y
specifically designed and built to receive wastes.
Not
Leaving crops that were ready for harvest in the field or tilling them
harvested/ Not applicable
into the soil.
ploughed-in
Abandoning material on land or disposing of it in the sea. This includes
open dumps (i.e., uncovered, unlined), open burn (i.e., not in a
Refuse/
controlled facility), the portion of harvested crops eaten by pests, and Y
discards/ litter
fish discards (the portion of total catch that is thrown away or
slipped).
Sewer/
Sending material down the sewer (with or without prior treatment),
wastewater Y
including that which may go to a facility designed to treat wastewater.
treatment
Sending material to a destination that is different from the 10 listed
Other N
above. This destination should be described.

9
Excludes crops intentionally grown for bioenergy, animal feed, seed, or industrial use