Globally, it is estimated that 30–40% of all food produced is lost or wasted somewhere along the food supply chain (Gustavsson et al., Reference Gustavsson, Cederberg and Sonesson2011; World Wildlife Fund United Kingdom, 2021). Food loss is food that is ‘lost’ in the early stages of the food supply chain, i.e. on farm and at the manufacturer/processor, while food waste is food that is wasted at the latter stages of the food supply chain, i.e. retail/food service and consumer. After consumers, farms are the second largest contributor to food loss and waste (FLW) with an estimated 15% of all food produced being lost on farm (World Wildlife Fund United Kingdom, 2021).
The dairy industry faces growing pressure from political entities, local communities and consumers to enhance sustainable food production practices (Bojovic and McGregor, Reference Bojovic and McGregor2023); this includes addressing FLW. Despite the significant contribution of the dairy sector to global food systems, data on food loss during dairy farming and processing are scarce (World Wildlife Fund United Kingdom, 2021).
The few studies that have been conducted on production and processing of dairy food loss have estimated percentages of food loss ranging from 1% to 14% of total production (Redlingshöfer et al., Reference Redlingshöfer, Coudurier and Georget2017; Tostivint et al., Reference Tostivint, de Veron, Jan, Lanctuit, Hutton and Loubière2017; March et al., Reference March, Toma, Thompson and Haskell2019; Wesana et al., Reference Wesana, Gellynck, Dora, Pearce and De Steur2019a; Guzmán-Luna et al., Reference Guzmán-Luna, Nag, Martínez, Mauricio-Iglesias, Hospido and Cummins2022; Australian Dairy Products Federation Inc. and Stop Food Waste Australia, 2023; Elliot et al., Reference Elliot, Goldstein and Charlebois2025). The drivers of food loss from these studies ranged from dumping milk due to antibiotic use (Redlingshöfer et al., Reference Redlingshöfer, Coudurier and Georget2017; Tostivint et al., Reference Tostivint, de Veron, Jan, Lanctuit, Hutton and Loubière2017; March et al., Reference March, Toma, Thompson and Haskell2019; Guzmán-Luna et al., Reference Guzmán-Luna, Nag, Martínez, Mauricio-Iglesias, Hospido and Cummins2022; Australian Dairy Products Federation Inc. and Stop Food Waste Australia, 2023); a lack of automation during milking causing spills (Tostivint et al., Reference Tostivint, de Veron, Jan, Lanctuit, Hutton and Loubière2017; Wesana et al., Reference Wesana, Gellynck, Dora, Pearce, De Steur and Galanakis2019b); inefficient cleaning protocols (Australian Dairy Products Federation Inc. and Stop Food Waste Australia, 2023); poor standardisation of processes on farm leading to a reduction in the subsequent quality of products after processing (Wesana et al., Reference Wesana, Gellynck, Dora, Pearce and De Steur2019a); large volumes of waste whey from cheese processing (Hetherington et al., Reference Hetherington, Loch, Juliano and Umberger2025); and the use of quotas promoting overproduction and subsequently the dumping of surplus milk on farms (Elliot et al., Reference Elliot, Goldstein and Charlebois2025). However, these studies did not provide an understanding of the rationale behind such decisions, the destinations of dairy loss and the nuances in the relationship between farmers and processors with regard to food loss-related decisions and how these might prevent or cause greater quantities of food loss.
New Zealand’s dairy industry provides a valuable context to explore food loss dynamics and their broader implications providing highly relevant and novel perspectives to the global discussion on FLW in food supply chains. New Zealand’s dairy industry is unique compared to other countries as it is unsubsidised, has minimal government intervention (Anderson and Valenzuela, Reference Anderson and Valenzuela2021; Coltman, Reference Coltman2021), is dominated by a cooperative processing company (Fonterra) (Qian and Olsen, Reference Qian and Olsen2021) and is export-oriented (95% of dairy is exported). New Zealand is ranked the eighth largest milk-producing country, generating 21 billion litres of milk a year (Dairy Companies Association of New Zealand, 2021). In New Zealand, dairy farming is under increasing political and social pressure to change practices due to their estimated contribution to New Zealand’s greenhouse gases (i.e. enteric methane produced by cows) (van Selm et al., Reference van Selm, de Boer Ijm, Ledgard and van Middelaar Ce2021). Further, the comparatively high inputs (e.g. fertilisers, animal feed, calf milk powders) required for dairy farm productivity and the rising costs of these inputs (DairyNZ, 2023b) mean that the dairy sector is motivated to search for viable solutions that address industry-wide issues. Although one New Zealand study has assessed nutrient circularity by tracking nutrient flows through dairy farms (Burggraaf et al., Reference Burggraaf, Mazzetto, Romera, Mercer and Ledgard2022), no published data are available on dairy losses at either farm or the processor level when considered specifically through the scope of food loss. The aim of this study was to explore dairy food loss across stakeholders early in the food supply chain in a New Zealand context. A mixed-method approach was applied to two case studies (dairy farmers and a dairy processor) to quantify dairy loss using the FLW standard methodology, to determine the destination of dairy loss and to identify the social factors influencing food loss decisions during production and processing of dairy products in a New Zealand context.
Methods
The methods selected to explore dairy food loss involved a mixed-methods (quantitative and qualitative) approach using case study design (Cook and Kamalodeen, Reference Cook and Kamalodeen2020). This enabled the research to capture food loss quantities and destinations using the FLW standard, as well as use reflexive thematic analysis to interpret how the perspectives, decision rationales and stakeholder relationships of dairy farmers and the dairy processor contributed to prevention or caused food loss. Incorporating case studies from both the dairy processor and dairy farmers provided a holistic understanding of the factors that contribute to or prevent food loss within the dairy supply chain. Before recruiting participants, ethics approval for this research was granted by the University of Otago Human Ethics Committee (Ethics Committee reference number D22/019).
Sampling strategy
‘Data saturation’ methods were not used to inform sample size as it was incompatible with the chosen analysis (reflexive thematic analysis) (Braun and Clarke, Reference Braun and Clarke2021c). The number of participants included was instead informed using the ‘information power’ model (Malterud et al., Reference Malterud, Siersma and Guassora2016). This model emphasises data ‘richness’ over ‘data saturation’ (Malterud et al., Reference Malterud, Siersma and Guassora2016; Sim et al., Reference Sim, Saunders, Waterfield and Kingstone2018; Braun and Clarke, Reference Braun and Clarke2021c) and uses five aspects to determine sample size: the breadth of research aims, participant specificity (e.g. relevance of their expertise/experience), theoretical frameworks used, quality of dialogue and intended depth of analysis (within or cross-case). Using the information power model to determine sample size, the dairy processor case study included six participants who worked at the dairy processor, and the dairy farmer case study included eight participants who were dairy farmers, which represented approximately 10% of the dairy processors’ suppliers.
The dairy processor selected for this case study was similar to many of the processing plants in New Zealand in that it had sole provider agreements with its supplying farms, its main operations were to create milk powder products (i.e. whole milk powder, infant milk powder, etc.) and ultra heat-treated (UHT) milk primarily for export to Asian markets. The particular dairy processing site can be characterised as a medium-sized factory for New Zealand, with approximately 400 people working on-site. The participating dairy farmers had all provided milk to the dairy processor over at least the past 12 months prior to data collection, were over the age of 18 years and were either farm (land) owner/operators, sharemilkers, contract milkers or farm managers.
Category inclusion criteria – FLW standard
The quantitative data were collected using the scope of a FLW inventory described by the FLW standard (Hanson et al., Reference Hanson, Lipinski, Robertson, Dias, Gavilan, Gréverath, Ritter, Fonseca, VanOtterdijk and Timmermans2016). This ensures that data collected are comparable and aligned with international reporting practices. The FLW standard requires data collected to have a defined timeframe (i.e. June 2021–June 2022 for farmer estimates, June 2017–June 2021 for dairy processor data), material type (i.e. dairy wastes/by-products), destination of waste and by-products (i.e. disposal, recycling, reuse) and boundary (i.e. data from the participating dairy processor and dairy farmers) when reporting volumes of FLW (Hanson et al., Reference Hanson, Lipinski, Robertson, Dias, Gavilan, Gréverath, Ritter, Fonseca, VanOtterdijk and Timmermans2016).
Material type
Dairy farm food loss includes milk retrieved from the cow on farm that is not sold on for human consumption. This included reject milk, ‘red’ milk and colostrum milk (see Table 1). While bovine colostrum for most farmers is harvested and then fed to calves, this colostrum milk has also been harvested and processed for human consumption when market conditions make this profitable (Linehan et al., Reference Linehan, Ross and Stanton2023); therefore, it is included in this study as a milk loss category. This definition does not include vat wash volumes, unmet yield potential or calf replacement powders to feed calves purchased from elsewhere. Dairy processor food loss includes all dairy products that arrive at the reception silo, through to end of processing (Table 1). This included dissolved air flotation (DAF) sludge to capture milk solids (MS) from cleaning in place (CIP) and other operational losses. Other categories of food reared/grown on farm (i.e. meat animals, food crops) and wastewater were excluded from this study.
Inclusions of waste categories in the dairy farm case study and the dairy processor case study
Table 1 Long description
Waste category terms are defined for two case studies: dairy farm and dairy processor. On farms, included categories are reject milk that fails processor quality or safety checks and is disposed of on-farm, colostrum milk from the first four days including transition milk, and red milk from antibiotic-treated animals kept out of the vat. At processors, included categories are rejected milk turned away at the reception silo, UHT milk waste lost or disposed of during processing, packaging, or storage, dissolved air flotation sludge from the on-site treatment system handling milk and cleaning wastes, and milk powder wastes from drying, canning, processing, or retention samples not sold for human consumption. Farm terms mainly describe milk withheld before it enters the supply chain, while processor terms include both incoming rejections and downstream processing and treatment residues. Definitions are qualitative and do not provide quantities, rates, or time periods, so comparisons are about scope and type rather than magnitude.
Quantitative data
Quantitative data for dairy farmers were gathered during semi-structured interviews in the form of farmer estimates focussing on the June 2021–June 2022 season. National statistics showed a decrease of 4.1% in MS compared to 2020/2021 season, but had substantially higher payouts for dairy farmers than the previous season (a 24% increase from the previous year) (Livestock Improvement Corporation Limited and DairyNZ Limited, 2022). During interviews, farmer participants did not report that they had changed their farming practices (or extra efforts to reduce milk losses) in response to this increase in payout. The region the farmers were from reported no extreme weather events for the 2021/2022 season, and an average somatic cell count of 159,000 cells/ml was reported for the region, which was slightly less than the previous year (163,000 cells/ml) (Livestock Improvement Corporation Limited and DairyNZ Limited, 2022). All farmer participants could give the number of cows that they milked; however, only participants 1, 3, 5 and 7 could give an estimate of the annual kg MS produced per cow (2001). Utilising farmer estimates and proxy data is in line with the recommendations of the FLW standard where primary data collection is not possible (Hanson et al., Reference Hanson, Lipinski, Robertson, Dias, Gavilan, Gréverath, Ritter, Fonseca, VanOtterdijk and Timmermans2016). Therefore, the following equations (Equations 1–5) were used to supplement these estimates, along with referenced proxy data (Table 2). Standardised equations were used to calculate MS, i.e. ‘the milk-fat and protein components of raw milk’ (Ministry for Primary Industries, 2001) of the different categories to enable consistent estimation of food loss across categories and participants.
Abbreviations and referenced proxy data for dairy farm equations
Table 2 Long description
Dairy farm proxy variables and abbreviations are listed with definitions, typical values, and literature sources for use in farm equations. Herd health and management inputs include clinical mastitis incidence at 14% per lactation and a 10-day milk withholding period after antibiotic treatment. Production timing is represented by 282 days in milk per year before the next calving. Colostrum parameters include an 8-litre yield, milk solids in colostrum at 18.7%, and a colostrum and transition period lasting 4 days, described as 8 milkings. Milk solids for regular milk are defined as the fat and protein components, with a proxy value noted as 9% of milk. The number of cows milked is defined but no numeric value is provided, so herd size must be supplied from another source.
For milk yield calculations, where per cow yields were not recalled or provided during farmer interviews, we imputed them using the mean kg MS per cow from respondents who did report. Therefore, for participants 1, 3, 5 and 7, estimates of kg MS produced per cow milk yield in tonnes of MS was calculated using Equation 1:
\begin{equation}{{Milk\,\,yield }}\left( {{{tonnes\,\,MS}}} \right) = \frac{{{{kg\,\,MS\,\,per\,\,co}}{{{w}}_{{ }}}\,\,\times\,\,{{Co}}{{{w}}_n}}}{{1000}}\end{equation}For farmer participants 2, 4, 6 and 8, an average of other farmers’ kg MS per cow was used to calculate milk yield using Equation 2:
\begin{align}
Milk\,\,yield \left( {tonnes\,\,MS} \right) & = \nonumber\\
& \!\!\!\!\frac{{Average\,\,kg\,MS\,\,per\,\,cow{_{F1,F3,F5,F7{{ }}}}\;\times\;Co{w_n}}}{{1000}}
\end{align}
The quantity of ‘red milk’ produced from antibiotic treated cattle was calculated using Equation 3:
\begin{align}
{{Red\;milk}}\,\,\left( {{{kg\,\,MS}}} \right)\,\,& {{ = }}\,\,\left( {{{Co}}{{{w}}_n} \times \;{{CMI}}\% } \right)\; \nonumber\\
&\quad \times\;\left( {\left( {\frac{{{{kg\,MS\,\,per\,\,cow}}}}{{{{Days\,\,in\,\,milk }}}}} \right)\;\times\;{{MWP}}} \right)
\end{align}
Colostrum quantities were calculated using Equation 4:
\begin{align}
{{Colostrum }}\,\,\left( {{{kg\,MS}}} \right) & {{ = Co}}{{{w}}_n}\; \times \;{{Yiel}}{{{d}}_{{{Col}}}} \times \;{{kg\,M}}{{{S}}_{{{Col}}}}\; & \nonumber\\
&\quad \times\;{{Days\,\,in\,\,colostrum}}
\end{align}
The percentage of total milk production attributed to losses was calculated using Equation 5:
\begin{align}
& \% {\text{ }}loss{\text{ }}of{\text{ }}Total{\text{ }}Production = \nonumber\\
&\quad \left( {\frac{{Red{\text{ }}Milk{\text{ }}\left( {kg{\text{ }}MS} \right) + Colostrum{\text{ }}\left( {kg{\text{ }}MS} \right)}}{{Milk{\text{ }}Yield{\text{ }}\left( {kg{\text{ }}MS} \right) + Red{\text{ }}Milk{\text{ }}\left( {kg{\text{ }}MS} \right) + {\text{ }}Colostrum{\text{ }}\left( {kg{\text{ }}MS} \right)}}} \right)\!\times 100
\end{align}
The participating dairy processor provided all the data they had on relevant categories within the timeframe (June 2017–June 2021) in Microsoft Excel workbooks. Categories included volumes of milk received from farmers, milk processed on-site, rejected milk, UHT milk waste and DAF sludge volume. Powder wastes were provided for the period January 2021–March 2022. The data were converted into tonnes MS using 9% conversion for UHT milk (Table 3) and 15.5% of the volume of DAF sludge (the mean MS % from tests of the DAF sludge provided by participating processor). Destinations of food loss were identified by participants for each specific category during interviews.
Dairy farm participant farm demographics, milk production and estimated milk loss (n = 8)
Table 3 Long description
Eight dairy farms are compared on herd size, annual milk production, and estimated annual milk loss measured in milk solids. Herd sizes range from 335 to 1,578 cows, and annual production ranges from 166 to 761 tonnes of milk solids, with per-cow yield spanning 430 to 550 kg milk solids. Total annual milk loss ranges from 2,911 to 13,220 kg milk solids, corresponding to 1.55% to 1.98% of total production. Losses are dominated by colostrum milk, which is consistently sent to calves and ranges from 2,005 to 9,443 kg milk solids. ‘Red milk’ ranges from 884 to 3,777 kg milk solids and is typically diverted to on-farm manure management systems or to calves plus manure management. Reject milk is only reported for one participant at 1,530 kg milk solids. The mean values are 839 cows, 403 tonnes milk solids produced, 7,218 kg milk solids lost annually, and 1.78% annual loss, noting that figures reflect a single production season and some categories are not reported for all farms.
Based on 2021–2022 season (June 2021–June 2022) 282 days of milk production.
MS, milk solids. The mass of milk that includes milk-fat and protein) Approximately 9% of liquid milk.
‘Red milk’ refers to animals whose milk does not enter the vat for pick up by the processor due to animal being treated with antibiotics. Called ‘red milk’ because farmers mark these cows with red spray paint and bands to ensure workers know not to allow their milk into the vat.
MMS, on-farm manure management system – commonly referred to as effluent in interviews.
Qualitative data
Qualitative data were collected through 60-min-long semi-structured interviews and analysed using reflexive thematic analysis through the theoretical stances of critical realism (ontology) and contextualism (epistemology). Reflexive thematic analysis was selected for its more inductive analysis approach, which suited the exploratory nature of this research; and its theoretical flexibility enabled descriptive and interpretive examinations of the data (Braun and Clarke, Reference Braun and Clarke2021a). The data generated provided a more holistic overview of the social factors, leading to dairy food loss generation and prevention in these case studies.
This research did not involve vulnerable parties. Informed consent was obtained from participants before interviews, and steps were taken to ensure the confidentiality of participants (omission of names, businesses, etc.). Interviews were conducted in person and were recorded by audio devices and later transcribed into Microsoft Word non-verbatim. Transcripts were analysed within their respective case (dairy farmer or dairy processor) using the iterative six phases (familiarisation, data coding, initial theme generation, theme development, theme refining/defining and writing up) of reflexive thematic analysis to generate multifaceted themes (Braun and Clarke, Reference Braun and Clarke2006, Reference Braun and Clarke2019, Reference Braun and Clarke2021a, Reference Braun and Clarke2021b, Reference Braun and Clarke2021c; Clarke and Braun, Reference Clarke and Braun2021). Familiarisation and initial generation of semantic codes using NVivo software Lumivero (2020) were conducted. The phases of analysis were revisited at least three times create or refine codes and themes before generating the resulting ‘theme trees’.
Results
Dairy farmer milk food loss quantities
The total yield of MS (kg MS), milk yield per cow, waste milks (red milk and colostrum) and percentage loss of total production for each of the farmer participants are presented in Table 3. Given the farm demographics and milk yield per cow from the eight dairy farmer participants, the sample of dairy farmers broadly aligned with the average characteristics of herd size and milk production in the local region (791 cows and 422 kg MS per cow, respectively), but were higher than NZ national averages for herd size and milk production (449 cows and 386 kg MS per cow, respectively) (Livestock Improvement Corporation Limited and DairyNZ Limited, 2022). Participants in this study had an average herd size of 839 cows, which was 6% higher than the local regional averages and 87% higher than the national average. Participating farmers did report greater milk production per cow, producing an average of 482 kg annual MS per cow, which was 14% higher than the local region average and 25% higher than the national average for 2021 (Livestock Improvement Corporation Limited and DairyNZ Limited, 2022).
Overall, the mean (standard deviation [SD]) percentage loss of total milk production was 1.8 (0.12) %. The largest loss came from colostrum with a mean of 5018 kg MS per farm (5.98 kg MS per cow). Only two destinations, calves and on-farm manure management system (MMS), were used by the dairy farmers interviewed. Three of the dairy farmers chose not to feed calves ‘red milk’ because of calf health concerns and recommendations by their veterinarian. Sometimes, the calves fed red or colostrum milks were located on a different farm, with this milk being supplied to these other farms through informal distribution networks. If no calves were available to consume the waste milks, the milk was dumped into the on-farm MMS and eventually irrigated onto paddocks.
Dairy farmer themes
From the eight farmer interview transcripts, 11 codes and three themes were generated through reflexive thematic analysis (Figure 1).
Theme trees generated during reflexive analysis of the eight dairy farmer interviews. Presented are the 11 codes (green boxes) that were generated which subsequently informed the three multifaceted themes (black boxes). SCC, somatic cell count.
Figure 1 Long description
Theme trees generated from dairy farmer interviews illustrate three themes and eleven codes. Theme 1 states, ′Some dairy food loss does happen, but the bigger issue is other wastes.′ It includes codes: ′Farmers feel that big food loss events occur rarely,′ ′Some waste is inevitable when farming,′ ′Farmers do not feel that dairy food loss is an issue that needs solving,′ and ′Farmers’ on-farm waste issues are not dairy related.′ Theme 2 is ′A focus on current costs and future performance/profit drives decisions affecting food loss on-farm.′ It includes codes: ′Current food loss destinations work well because it is low/no cost,′ ′Profit and low SCC are the farmers’ measures of success,′ and ′Prevention actions mitigate possible risks of food loss.′ Theme 3 states, ′The dairy processor can be flexible, which prevents food loss, but does not uniquely affect on-farm practice differently than if supplying milk to another processor.′ It includes codes: ′The processor does not specifically affect what farmers do on-farm,′ ′The processor financially incentivizes quality targets (e.g., SCC),′ ′Farmers trust the processor to continue being good to deal with,′ and ′Farmers are unsure or don’t know what the processor does with the milk.′ The key indicates themes in black boxes and codes in green boxes.
Theme 1: Some dairy food loss does happen, but the bigger issue is other wastes
A theme that was generated from interview transcripts was that dairy farmers acknowledged that some dairy loss occurred, but they were more concerned about other wastes occurring on-farm (Theme 1, Figure 1), such as animal casualties and surplus (bobby) calves. When participants were asked about their on-farm dairy wastes, substantial one-off events were often framed as happening rarely, and routine food loss occurrence was dismissed as being insubstantial requiring no further action. Further, most participants commented or implied that some routine and one-off dairy food loss on-farm was inevitable and expressed these inevitable food loss occurrences were outside of their control.
P6: … But on a dairy farm. What could we do? Sort of a hard question to answer. Well ideally you could say we are not going to have any, but that is not going to happen.
This perception that dairy food loss did not need to be or could not be further managed may be why the farmers interviewed did not or could not provide a clear estimate on the volume of these routine dairy wastes occurring.
P3: But we do not put massive amounts out. I think we have got four red cows (treated with antibiotics, milk is discarded) at the moment. Like you would not even be able to pump that out to make any difference.
The responses to interview questions situated dairy wastes as ‘low priority’ and that other wastes and issues on-farm were viewed as more concerning or more relevant. Farmers implicitly trusted that the dairy processor would continue to have the capacity to take their milk, regardless of market or product fluctuations and therefore they were never at risk of overproducing and having to dump uncontaminated milk on-farm. Further, the two destination options for dairy food loss (calves and MMS) were actually viewed as resources benefitting farm performance, not waste destinations or issues that needed solving.
P4: We do not dump any milk. We use all the colostrum and all our penicillin milk is kept and fed to calves.
Theme 2: Current costs and future performance/profit drives decisions affecting food loss on farm
The second theme captures the main drivers of decisions influencing dairy food loss on-farm. Farmers considered food loss decisions through the focus of managing current costs and attempting to improve future performance and profits (Theme 2, Figure 1). Farmers were not motivated to find alternatives for their waste milks because utilising it as calf feed was viewed as a cost saving as it meant they did not have to purchase milk powder, and the MMS was already available as it is required infrastructure to process and dispose of animal manures from the dairy parlour.
P8: At the moment, it is pretty efficient because we have got enough milk to feed our replacements so there’s no loss and we are not having to buy in milk powder.
Most of the farmer participants discussed that pursuing lower somatic cell count (SCC), which is a quality metric for the dairy processor, in the milk was another metric of success due to the financial incentive from the dairy processor. While some farmers found it did not significantly affect their practice to achieve this, others discussed that they felt it led to more antibiotic treatments therefore having more ‘red milk’ to be disposed of or fed to calves.
P8: …It just seems to be where we tend to sit is around the 120–140 (thousand SCC) range. And if we do try and push for under 100, we do end up with a lot more cows under treatment (on antibiotics) and not really getting any advantage from it.
Participants were however proactive in seeking out and investing in prevention measures that would avoid or mitigate the risk of factors leading to food loss. These included investing in animal health products such as teat sealants, which are used at the end of the milking season to prevent mastitis during the following season (MSD Animal Health, 2024), and building additional storage tanks for waste milks.
P8: So obviously a big drive is to reduce the antibiotics on the farm as well. So we did some trial work in teat sealing some of our low somatic cell cows, so herd testing them, identifying that group and that seemed to work fine.
Another technology adopted by most farmers was cow collars. These are wearable monitors that provide farmers with individual cow data including their feed intake, body temperature, heart rate and movement. The data are interpreted by specialised software that alerts the farmer when a cow’s data differ from normal levels (Afimilk, 2024).
P5: … This is where the collars come into it. They tell us daily how the cows are performing. And health wise as well. One of our main drives for putting collars on is you get to pick up cows subclinically before they actually show signs.
When participants talked about possible milk contamination with ‘red milk’, farmers would pay for a test (i.e. a service offered by veterinarians) to check if antibiotic was present. The results of this test would dictate whether the milk was dumped to avoid the risk of a financial penalty from the processor. Because of the risk of contamination occurring, and the resulting significant loss of income from having to dump milk, and any associated financial penalty, farmers viewed insurance as a necessary financial risk mitigation method that would cover 80% for the first incidence of the value of the milk if it needed to be dumped.
P6: It is not cheap. It is not cheap to have the insurance but you have got to have it you know.
Theme 3: The dairy processor can be flexible which prevents food loss but do not uniquely affect on-farm practice
Other than the influence of incentivising lower SCC in the milk, the participants felt that the dairy processor did not directly guide on-farm decisions any different to the way they would if they were supplying other dairy processors in New Zealand (Theme 3, Figure 1). Instead, the influence of government policies, legal and consent requirements, and quality assurance rules were explicit in decision-making.
P6: Probably only what they (the milk processor) have got to do through MPI (Ministry of Primary Industries).… There would not be [dairy processor] rules. They are probably, I should not say it, but probably a bit, like [other dairy processor] you have got to tick the box on everything too much.… But still saying that, we had our shed checked yesterday and that is got nothing to do with [dairy processor] or [other dairy processor] or anything. AsureQuality (an independent food quality assurance provider) come do it and that is their rules we have got to abide by them.
Communication with the dairy processor was characterised as accessible, positive, personal and timely. Some farmers recalled that the communication with dairy processor staff led to some flexibility in the rules that dictated financial penalties, therefore avoiding food loss. For example, milk that was just outside of the temperature range was permitted to be picked up.
P4: No. [Dairy processor] were really good about it. They, because we normally get picked up between 6 and 8 o’clock in the evening. And they said, we will just pick it up at last pickup. So it managed to chill down.
Dairy processor quantities
The dairy processor provided monthly averages from June 2017 to June 2021 for milk inputs (milk from farmers and milk processed on-site includes milk from other processors), reject milk, UHT milk waste and DAF sludge. Milk powder wastes were average data from January 2021–April 2022 as presented in Table 4. The average percentage of milk wasted was 5.5% annually. Seasonal variation resulted in higher volumes being wasted during spring, summer and autumn (August–May), with lower volumes being wasted during the winter months when only a few farmers were providing milk. Percentage (%) loss of total milk processed was higher during May–July mainly driven by the milk powder wastes when retention samples (Three samples taken from each batch of product that is tested for microbiological and nutritional safety and specifications. Some samples are retained over the shelf-life of the product should retesting be required [Ministry for Primary Industries, 2022].) are disposed of. Also, if UHT milk needs to be disposed of, it is stored until lower production months to be given to pig farmers when there is less milk coming through the plant.
Dairy processor monthly losses and wastes
Table 4 Long description
Monthly milk inputs and on-site processing are listed alongside losses and wastes from reject milk, UHT milk waste, DAF sludge, and milk powder wastes, plus total waste and percent loss. Total milk wastes range from 58 to 130 tonnes of milk solids across months, with the highest tonnage in May at 130 and the lowest in July at 58. Percent loss varies much more than tonnage, rising sharply in low-throughput months: June has 72 tonnes wasted but the highest loss rate at 28.03 percent, and July is 17.64 percent. Higher-throughput months such as October and November process about 3,187 and 3,001 tonnes of milk solids and have lower loss rates of 3.84 and 3.40 percent. DAF sludge is the largest waste stream most months, typically around 20 to 65 tonnes of milk solids, while reject milk is minimal at 0 to 4. Annual totals are 21,807 tonnes processed, 1,191 tonnes of milk solids wasted, and an overall loss rate of 5.46 percent. Interpret comparisons cautiously because milk powder waste values are recorded in cubic metres and come from a different time window than the other waste categories.
All data except milk powder wastes were based on data from June 2017 to June 2021. Milk powder waste data were from January 2021 to March 2022. Tonnes MS (metric tonnes of milk solids) represents the mass of milk after water has been removed (i.e. protein, fat, lactose, minerals), which is approximately 9% of liquid and UHT (ultra heat-treated) milk, and 15.5% of DAF (dissolved air flotation) sludge.
a Destination: pig farm/land application.
b Destination: pig farmer.
c Destination: pig farmer/land application/composting.
d Destination: stock-feed.
e Destination: higher quantities of milk powder wastes during the winter months are due to the disposal of retention samples.
DAF sludge (2.7%) and milk powder wastes (2.1%) were the main contributors to total food loss, with reject milk (0.6%) contributing the least. Differences between milk from farmers and milk processed on-site can be accounted for in the trading of raw milk with other milk processor companies. Milk is sold on or purchased depending on product specification plans as different milk powders require different volumes and time to process.
Pig farms were the main destination for liquid wastes with land application and composting as secondary, less preferred options. Powder wastes were either recycled through the processing facility through reworking (not included in the quantities of this study), or on-sold to a local stock-feed manufacturer.
Dairy processor themes
Of the six dairy processor staff interview transcripts, 13 codes and four themes were generated through reflexive thematic analysis (Figure 2).
Theme trees generated during reflexive analysis of the six dairy processor interviews. Presented are the 13 codes (blue boxes) that were generated which subsequently informed the four multifaceted themes (black boxes).
Figure 2 Long description
Four themes related to food loss in processing companies are presented. Theme 1: ′Food loss is inevitable (when producing food), but the processing company is doing a good job of not being wasteful.′ Codes include preventative action to reduce food loss, the company′s process affecting food loss and participants′ views on food loss. Theme 2: ′Food loss is generated as a risk avoidance practice.′ Codes include legal requirements, reputation concerns, destination choices for being ′less risky,′ and food loss to avoid contamination. Theme 3: ′Financial outcomes are the processing company′s measure of success/driver in destination choice and prevention decisions.′ Codes include decisions based on costs and profit opportunities. Theme 4: ′Relationship management with stakeholders is essential for preventing food loss and access to food loss destinations.′ Codes include customer specifications, concerns about external food loss destinations, current destinations working well and prevention through stakeholder communication.
Theme 1: Food loss is inevitable (when producing food) but [processor] is not wasteful
The theme ‘Food loss occurrence was inevitable but also that the processor was not wasteful in the generation of products’ (Theme 1, Figure 2) captured that some dairy processor participants shared they did not like that waste was happening and were frustrated about their lack of control over preventing some of it, but most were content with current waste reduction efforts. Although disposal was viewed as a negative option in the processing of dairy products, food loss was seen as embedded, and necessary in the process of making food. Despite this, the quantities of waste occurring were understood by participants to be proportionally small compared to the volume of product generated and were further reduceable through incremental technology changes, and big waste events were expressed as occurring rarely.
P6: Proportionally, it is like a per cent. While the numbers seem big when you talk about it. When you average it out over the course of the year. It is small in comparison.
Theme 2: Food loss is generated as a risk avoidance practice
Concern about the reputation of the company that owned the dairy processor with its customers and local community were evident during interviews with processor participants, which is captured in the theme ‘Food loss is generated as a risk avoidance practice’ (Theme 2, Figure 2). Anxiety was expressed about the occurrence of anything that would increase the risk of quality or safety breaches for product to be exported to China. Some participants referred to previous well-publicised scandals such as the 2013 Fonterra recall (Fonterra Co-operative Group Limited, 2017) and the 2008 Chinese milk scandal (Ramzy, Reference Ramzy2008) to justify the choice to dispose of food products that varied from approved standards. Further, ensuring that product meets high-quality standards was essential for continued operation of the business to maintain the reputation of the company and its associated brands.
P2: We are going through that full process at the moment so, those regulations cause a lot of waste but if they were not there and were not so tight then we would have what has happened in the past, bad product going out to China, and having mass recall events, people getting sick. We work in an industry where people die. So that is why they are so tight.
Reputation protection which led to waste was also enforced by New Zealand government policies. Specifically, participants discussed rules enforced by MPI and how maintaining MPI compliance influenced the dairy processors’ rules and implementation of financial penalties to discourage poor quality or unsafe milk arriving on-site from their supplying farmers. It was understood that these rules were focussed on ensuring food safety, but also achieving overseas market access requirements.
P5: Because all the regulations that govern raw milk, are almost all driven by our oversees market access requirements. [The requirements are …] quite strict on temperature, very strict on temperature. So 2018 those new regs (regulations) came in, and that meant that at 14 degrees (milk temperature when picked up from farm) we should be rejecting it.
Destination decisions were influenced by both local community and consumer perceptions of practices. This was associated with preferring destinations for food loss that decreased the risks of negative outcomes for the reputation of the business. For example, land application for the waste was identified as a less preferred option to the products being fed to pigs due to its visibility to the local community and its perceived negative attributes that would be linked to the business. There was also a reflection by participants that the perception of the entire dairy industry can be affected by the waste practices and disposal destinations chosen by the processors within it.
P5: Oh totally. That is reputational damage, oh [dairy processor] is dumping product, .. So we pretty much, we have not gone to land (land application) all of this season.
Theme 3: Financial outcomes are the measure of success and driver in food loss choices and prevention
Unsurprisingly, improved financial outcomes were the focus when considering destination choice or dairy food loss prevention strategies by participants (Theme 3, Figure 2). During interviews, participants discussed how incremental improvements in their processing efficiency were focussed on increasing profitability rather than food loss prevention objectives.
P1: … there is no point dumping product that could be good product. If you can process it and it is still going to be good and within your specs (product specifications) and all the rest of it then you are better off to do so.
However, destination choices – while influenced by the avoidance of risk – were compared and selected based on the least costly option for the dairy processor. This disincentivised investment into diversifying destination options.
P3: But like most companies they go for the least expensive option. So all of a sudden we picked up another pig farmer, then all of a sudden the DAF was not quite the problem that it was going to be, so then the (alternative destination) processing thing dropped off.
Theme 4: Relationship management with stakeholders is essential for prevention and access to food loss destinations
Fostering positive relationships and interactions was important for preventing food loss occurrence and maintaining access to current food loss destinations (Theme 4, Figure 2). Participants were highly knowledgeable about the stringent product specifications and why they were required by their customer stakeholders, and participants shared they continued to refine their planning and communication with customers to decrease the number of ‘CIP’ occurrences between product specification changes, leading to less food loss.
P4: [Customer] will turn around and say look we are doing S26 or whatever it is or S48 that goes on the production and they will get a sheet and the sheet will turn around and say, hey look you can run this but after this you are required to do a CIP before you go onto this next product.
The dairy processor incentivised farmers to meet quality and safety standards by applying significant financial penalties through supplier contracts. For example, antibiotics in milk above an accepted limit (i.e. 0.006IO/ml benzylpenicillin) could cost up to 40,000 NZD for the farmer. However, the dairy processor reduced the need to use these penalties by relying on positive relationships and open communication with the dairy farmer, enabling some flexibility around stringent rules as incidents were assessed on a case-by-case basis. A participant recalled preventing food loss occurring through assessing and discussing the event directly with the farmer.
P5: I had one (a farm whose milk was too warm when collected) last week, I have not had one for ages, and their primary cooler went out. And he (the farmer) said: no, the milk was down (cooled temperature), it was definitely 5°C when we went to the shed this morning. He was an early pick up (milk collection by the processor) at 8 o’clock in the morning, and it was 12°C. So it was not far out (of required temperature) at 2°C out, but it would be if he did not have primary cooling in that initial phase. So we were collecting at 8 o’clock. It was being collected into cold milk, it would come back here and the tanker would be well below 10°C.
The dairy processor further prevented waste of milk through the establishment of a network with other dairy processors that facilitated purchasing and selling of surplus raw milk between companies preventing seasonal surplus from farmer suppliers. It was also shared that there was a memorandum of understanding (MOU) agreement whereby one-off waste events would be prevented by communicating with other dairy processor companies to pick up or process milk in case of equipment failure, which prevented processing waste.
P5: If there was an absolute disaster, so one of the things that we have got in place is that with the independent companies we have a memorandum of understanding that if we break down we would all help one another out. So if we have a break down, we can ship raw milk straight off farm. So when they had the big floods in [town], we only ended up having to take two or three tanker loads of [other dairy processor] milk because of the bridges (being closed),.… We have a memorandum that basically says, if we have a breakdown we will (dairy processor network) help out.
Relationship management with waste destination stakeholders was essential for the dairy processor. This was exemplified when there was a risk that pig farmers would no longer accept DAF sludge. This was due to pig abattoirs reporting that pig carcasses fed large quantities of dairy DAF sludge led to substantial increases in hard fat causing abattoir staff to suffer repetitive strain injuries due to having to use extra force to cut the pork meat. To ensure continued access to pig farms as a waste destination, the dairy processor pays to have the DAF sludge composition tested so pig farmers can feed it in metred quantities, which decreases the occurrence of hard fat.
P5: And the inclusion of it (DAF sludge) is quite important, because they get a hard fat in the pigs if they put too much dairy product into them. And that is if their inclusion rates get too high. So they have all got nutritionists. We supply them the data, their composition, and they adjust their inclusion rates based on the composition of the DAF sludge.
Discussion
To our knowledge, this is the first study to quantify dairy food loss early in the food supply chain in New Zealand. Despite the many challenges facing dairy farmers, on-farm dairy food loss was low at 1.8%. Likewise, the loss measured at a dairy processor was also low at 5.5%. Although losses are low, given that approximately 21 million tonnes of milk (1.87 billion kg MS) (DairyNZ and Livestock Improvement Corporation, 2023) is produced annually in New Zealand, a 1.8% milk loss on farm across New Zealand equates to the annual production of approximately 86,000 cows (33 million kg MS) while a 5.5% dairy loss at the processor is equivalent to the annual production of 264,000 cows (102 million kg MS). While in other food loss quantity studies, it is argued that animal-sourced products, like dairy and meat, have proportionately low food loss quantities. These food losses still account for greater greenhouse gas emissions than other food types, primarily due to the enteric methane produced during farming of ruminant animals like cows and sheep. These studies argue that prioritising the reduction of dairy-associated FLW has the potential for greater reduction of negative environmental impacts, compared to similar reductions of other plant-sourced food types that have less greenhouse gases associated with their production (Al-Obadi, Reference Al-Obadi2021; World Wildlife Fund United Kingdom, 2021). Further, deciding the most appropriate destination of food loss is context and priority specific. While governments might prioritise destinations that reduce overall greenhouse gas emissions (New Zealand Government, 2019), businesses prioritise reducing their costs of operations (Figure 1, Theme 2; Figure 2, Theme 3). In FLW research, the FLW hierarchy prioritises preventing FLW and ensuring consumption of food by people before less prioritised destinations that reuse, recycle (animal feed) and dispose of FLW (Hanson et al., Reference Hanson, Lipinski, Robertson, Dias, Gavilan, Gréverath, Ritter, Fonseca, VanOtterdijk and Timmermans2016).
Based on the percentages of dairy loss identified in the current study, it can be argued that there is little food loss to address early in New Zealand’s dairy food supply chain, particularly on farm as 1.8% of milk loss would be difficult to improve beyond the current efforts of dairy farmers. Rather, the present study provides insights of the specific factors and drivers that have reduced dairy food loss occurrence, and decision rationales that led to ‘recycling’ destinations of this food loss. It was also identified that dairy processors and dairy farmers are incrementally improving their productivity through technology adoption and refinement, which in turn will likely reduce current food loss. The main concern that this study highlights is the lack of alternative destination options. Although dairy losses represent a small percentage of overall production, this still equates to substantial volumes that would need an alternate destination method that is at the same level or higher on the FLW hierarchy should current factors and drivers change due to economic, environmental, operational or social influences. For example, a change in policy which may mean farmers cannot feed milk with antibiotic residues to calves due to biosecurity risks, or development of land around dairy farms into residential areas make spreading of effluent and waste milks onto paddocks no longer feasible due to negative odour impacts on neighbours, or pig farmers may decide DAF sludge is affecting their operations negatively and stop receiving it from the milk processor. This lack of food loss destination options reduces system resilience and highlights a vulnerability for dairy farmers and processors to respond to changes in their operating landscape.
Quantities
The percentage of food loss occurring on farm (1.8%) was in line with previous studies that identified low levels (1–2.15%) of milk food loss occurring on farm (Tostivint et al., Reference Tostivint, de Veron, Jan, Lanctuit, Hutton and Loubière2017; March et al., Reference March, Toma, Thompson and Haskell2019; Guzmán-Luna et al., Reference Guzmán-Luna, Nag, Martínez, Mauricio-Iglesias, Hospido and Cummins2022) but less than the estimated milk loss on Canadian dairy farms (7%) (Elliot et al., Reference Elliot, Goldstein and Charlebois2025). Across production and processing, the results from the current study (5.5%) were similar to dairy food loss identified in France (5.6–8.2%) (Redlingshöfer et al., Reference Redlingshöfer, Coudurier and Georget2017) but much lower than the losses estimated in Uganda (14%) (Wesana et al., Reference Wesana, Gellynck, Dora, Pearce and De Steur2019a). While the proportions of food loss estimated in this research were small compared to the amount of dairy produced, the volumes of dairy food loss are substantial given the scale of production and processing in New Zealand. Figure 3 provides more detail on the quantities and destinations.
Dairy food loss and waste categories, quantities and their destinations identified from participating dairy farms and dairy processor. Quantities represent average cow number (n) and annual T MS (tonnes of milk solids) of the eight participating dairy farmers for each category, and the annual total quantities for the dairy processor. Some values may be affected by rounding error. Figure created using BioRender (O’Connor, 2025).
Figure 3 Long description
At the top, various waste categories are shown: DAF sludge (579 T MS; 2.7 percent), rejected milk (18 T MS; 0.1 percent), UHT milk waste (138 T MS; 0.6 percent), waste powders (456 T MS; 2.1 percent) and milk products (20,616 T MS; 94.5 percent). These are directed towards composting, land application, pig farmers, animal stock feed and food. An onsite waste treatment system is linked to the dairy processor. Below, the dairy farm section shows manure management/land application, red milk (2 T MS; 0.5 percent), colostrum (5 T MS; 1.2 percent) and milk (403 T MS; 98.2 percent) being fed to calves or processed. Cattle count is 839.
Factors that reduce dairy food loss
The low percentage of dairy food loss can be attributed to several factors including the properties and economic value of the products that are produced by liquid milk – in this instance dry powders and UHT milk with long shelf-life, few cosmetic specifications required for dairy products for market acceptance compared to fresh produce and the dominance of the cooperative farmer-owned business Fonterra, which accounts for approximately 79% of national milk supply (Fonterra Co-operative Group Limited, 2024). This has led to high competition with other dairy companies in New Zealand for milk supply, and as a result, processors must now offer supplier agreements and relationship management practices that are comparable or better than those provided by Fonterra to attain and retain farmer suppliers meaning dairy farmers have more equitable market power with the dairy processor. Thus, greater vertical collaboration (Bhattacharya and Fayezi, Reference Bhattacharya and Fayezi2021) between farmers and processors is seen across New Zealand’s dairy industry. This was exemplified by the dairy farmer Theme 3 (Figure 1), where the farmers shared they had positive relationship with the processor who could be flexible on rules and worked with them to prevent food loss, and the farmers did not see this as exceptional, but rather an expectation from any dairy processor they could supply to. Therefore, compared to other food types such as fruits, vegetables and arable crops, the dairy farmers in this study were less exposed to market-driven practices (Messner et al., Reference Messner, Johnson and Richards2021; Office of the Prime Minister’s Chief Science Advisor, 2024). This was due to greater vertical collaboration and more equitable sharing between farmers and processor of the uncertainties and risks of food production like variable environmental conditions (e.g. adverse weather), changing economics (e.g. change in value of dairy MS at global auctions) and operational dysfunction (e.g. machinery breakdown) (Bhattacharya and Fayezi, Reference Bhattacharya and Fayezi2021). The sharing of these risks reduces dairy food loss because it prevents farmers from having to take all of the responsibility for mitigating these risks – which often occurs through food disposal on farm (Gille, Reference Gille2012; Messner et al., Reference Messner, Johnson and Richards2021; Croad et al., Reference Croad, Campbell and Mirosa2024). This contrasts with a Canadian study by Elliot et al. (Reference Elliot, Goldstein and Charlebois2025), which discussed the role of the national dairy supply management system in generating on-farm milk loss. The system described used quotas for Canadian dairy farmers to match milk supply with domestic demand. While this may benefit Canadian dairy farmers as it stabilises prices, this also means Canadian farmers are largely responsible for dealing with the variable uncertainties and risks of milk production. It is argued by Elliot et al. (Reference Elliot, Goldstein and Charlebois2025) that the quotas incentivise Canadian dairy farmers to produce excess milk to mitigate the financial risks of underproducing and not meeting their quotas. But, as there was little opportunity for selling the surplus milk, it was disposed of on farm – generating food loss.
In this study, it was evident that the processor and farmers shared a culture of communication and collaboration between each other, further preventing dairy food loss. This was illustrated in the dairy processor Theme 4 (Figure 2) where relationship management between stakeholders (dairy farmers, pig farmers, testing agencies, other dairy processors) was essential in preventing dairy food loss. This aligns well to the food loss reduction critical success factor theme ‘communication and collaboration’ previously identified by Thorsen et al. (Reference Thorsen, Croad, Vincent and Mirosa2022), and the argument by Bhattacharya and Fayezi (Reference Bhattacharya and Fayezi2021) that food loss reduction can be realised by promoting collaboration across vertical (e.g. farmer to processor) and horizontal (e.g. other dairy processors, pig farmers) stakeholders. Communication was evident in the prevention of farm food loss as participants noted that the direct relationship between farmers and dairy processor facilitated decisions as to whether to dump milk on a case-by-case basis. Additionally, collaboration among processors prevented dairy food loss through the trading of excess raw milk between dairy companies, and the establishment of a collaborative MOU further ensured that milk was not wasted during infrastructure disruptions caused by weather events, power outages or mechanical failures. The social factors (more equitable sharing of risks between farmers and processors, a culture of collaboration between stakeholders) preventing dairy food loss identified in this research may be suitable for application to other country’s dairy food production and processing, and other food types. Adopting business structures that promote a more equitable market power share between farmers and other stakeholders in the supply chain, and collaboration between companies in the food system, could mitigate market-driven risks and subsequently avoid burdening farmers with the responsibility to dispose of food on farm. Establishing collaboration between similar companies through MOU’s between processors may prevent food loss or facilitate the utilisation of more preferred destination choices improving overall food loss outcomes.
Dairy production and processing benefited from a network of stakeholders that influenced food loss prevention, waste and destination availability. For example, veterinarians can prevent dairy food loss on farm, and subsequently at the reception silo of the processor, by providing testing services for antibiotic traces in the milk, informal networks between farmers enable use of waste milks for calves, pig farmers accepting DAF sludge for animal feed, composting companies offering services for DAF sludge disposal and livestock feed companies taking waste milk powders which then provide calf milk replacement products for dairy farmers. These relationships with stakeholders are beneficial as it keeps these nutrients within the food system (Kirchherr et al., Reference Kirchherr, Reike and Hekkert2017; Ellen MacArthur Foundation, 2019, 2021). On the other hand, some of these practices could continue the lock-in of surplus food waste in food production processes by disincentivising identification or development of alternative destinations (Papargyropoulou et al., Reference Papargyropoulou, Lozano, Steinberger, J and Ujang2014; Messner et al., Reference Messner, Johnson and Richards2021). For example, pig farmers taking DAF sludge as an end-of-pipe solution disincentivise the processor from reducing DAF sludge production, or investing in developing destinations that are more preferable in FLW hierarchies.
Drivers and destinations
Similar to the observation by Gille (Reference Gille2012) of food production in high-income countries, decisions to generate dairy food loss for both case studies were largely driven by risk avoidance or cost avoidance as seen in dairy farmer Theme 2 (Figure 1) ‘A focus on current costs and future performance/profit drives decisions affecting food loss on farm’; and in dairy processor Theme 2 ‘Food loss is generated as a risk avoidance practice’ and Theme 3 (Figure 2) ‘Financial outcomes are the processing company’s measure of success/driver in destination choice and prevention actions’. For farmer participants, the decision to divert colostrum and ‘red milk’ from the vat (exiting the food supply chain) were routine decisions to avoid financial penalties from the dairy processor, who sets these standards and enforces the penalties for the farmer, and who themselves are limited by policy and requirements placed on dairy products for export by the regulator (MPI) (Ministry for Primary Industries, 2022) and their customers. For the dairy processor, milk powder wastes during the winter months (May–July, Table 4) were largely due to the disposal of retention samples. These samples are a requirement by the regulator (MPI) to ensure microbiological and nutritional safety. Some samples are retained (retention samples) in storage over the shelf-life of the product, so that the company can retest the product should a contaminant issue be discovered later in the supply chain. The frequency of testing and the volume of the samples taken during production are based on regulatory and customer requirements. Reducing the volume of food loss caused by these samples would require regulatory changes. However, this is currently unfeasible as reducing sample volumes could increase food safety risk and consequently expose New Zealand’s dairy industry to greater reputational risks. The New Zealand dairy industry’s success in profitably exporting 95% of its dairy products relies on maintaining stringent food safety standards, which are essential for preserving customer and consumer trust. This is particularly important in the after-math of previous dairy-related scandals, such as the 2013 Fonterra recall (Fonterra Co-operative Group Limited, 2017) and the 2008 Chinese milk scandal (Ramzy, Reference Ramzy2008). For dairy processor participants, risk avoidance was the main driver for their food loss decisions with emphasis on choices that were perceived to protect the reputation of the dairy company to maintain a social licence to operate, and for sustaining trust in their products by their export customers.
This study identified that cost considerations and avoidance of risks were determining motivators that directed destination decisions of food loss for both case studies. The consensus in this study was that these destinations worked well because they were low cost (Dairy farmer code in Theme 2, Figure 1 ‘Current food loss destinations work well because it is no/low cost’), there was low risk of negative reputational consequences (Dairy processor code in Theme 2, Figure 2 ‘Reputation concern driven decisions’) or they were not waste destinations as the material was utilised for other beneficial purposes (Dairy farmer code in Theme 1, Figure 1 ‘Farmers do not feel that dairy food loss is the issue that needs solving’; Dairy processor Theme 3 ‘Financial outcomes are the company’s measure of success/driver in destination choice and prevention actions’).
However, issues were identified with relying on these limited food loss destinations (calves and MMS for dairy farmers; pig farmers, stock food, composting, land application for the processor), that may exclude their future use. For example, some dairy farmers opted to not feed calves ‘red milk’ to prevent the risk of future negative animal health outcomes such as spreading diseases (like the bacterial wasting disease Johne’s (DairyNZ, 2023a)) from their cows to their calves, and processor participants shared concerns about the future viability of pig farmers to take DAF sludge. Most participants expressed low motivation for exploring or adopting a more diverse array of food loss destinations or attempting to use destinations that were higher up in the FLW hierarchy (Ministry for the Environment, 2023).
For dairy farmers in this study, this perspective can be attributed to two reasons. Firstly, biologically, milk is produced by cows to feed calves; however, in the strict sense of food loss definitions, feeding this milk to calves is considered waste as it was produced with the intention of being consumed by humans (Hanson et al., Reference Hanson, Lipinski, Robertson, Dias, Gavilan, Gréverath, Ritter, Fonseca, VanOtterdijk and Timmermans2016; Ministry for the Environment, 2023). Secondly, the use of waste milks as a resource by farmers to rear replacement calves is arguably the best option (environmentally and financially) as when waste milks are not available, farmers buy calf milk replacers from stock food companies (who are supplied the milk powder by dairy processors). Therefore, there is a weak value proposition for dairy farmers to decrease or prevent current levels of routine waste milk occurrence based on financial metrics of success.
When some dairy processor participants were directly asked about whether the option to explore donation or other avenues for milk powders that did not meet specifications but could be considered safe to eat such as retention samples, the strictness of regulations set by MPI was identified as prohibitive and exposure to reputational risk, without a strong value proposition for reducing costs or adding profits made it an unattractive option. Again, this reflects that the generation of waste can occur due to a desire to avoid risks, and an unwillingness for the dairy processor to increase their risk exposure by providing food products that may be considered unsafe (Gille, Reference Gille2012).
Limitations
Like previous studies on food loss, the results of this research must be interpreted with the caveat that farm data was not directly quantified due to technical constraints and farmers not currently collecting this type of data for their businesses. This is an ongoing issue and is due to a combination of factors. One factor is that the perception of the term ‘food loss or food waste’ is attributed to societal shame which can lead to knowledge hiding (Bradshaw, Reference Bradshaw2018; O’Connor et al., Reference O’Connor, Skeaff, Bremer, Lucci and Mirosa2023). Another factor is participants misinterpreting whether a category ‘counts’ as food loss if its destination was not landfill. When interviewing participants, this issue was navigated by the interviewer who instead of using the term ‘waste’ or ‘loss’, also used terms participants would more readily identify as ‘by-products’ such as ‘red milk’ and ‘colostrum’. Another factor contributing to the absence of FLW data, suggested by Gille (Reference Gille2012) is that there is low motivation for stakeholders to identify and measure wastes as a risk avoidance tactic, where identifying and providing reliable quantitative FLW data could mean they would be forced to manage their wastes by regulatory authorities. This may partially explain why both case studies generated themes that expressed that dairy food loss was an unfortunate, inevitable, but not a substantial problem for the participants (Dairy farmer Theme 1, Figure 1; and Dairy processor Theme 1, Figure 2), which is similar to the perspectives previously identified in New Zealand kiwifruit production food loss (Croad et al., Reference Croad, Campbell and Mirosa2024). More pragmatically, the lack of dairy food loss data collection by farmers reflects that participants do not see dairy food loss as a priority issue for them, where their concerns were other wastes such as plastics, casualty animals and surplus calves. In contrast, the dairy processor collected food loss data regularly and invested in incremental improvements to improve the efficiency of their process. This was directly associated with increasing product yield and profitability while also ensuring they adhered to rules and reduced the costs of disposing of dairy wastes and by-products. Although some critique may be applied to sample size of this study as a limitation, this study applied the information power model (rather than data saturation) to determine the number of interviews within each case. Given the specificity of the topic, this ensured interviews conducted provided relevant data from participants who had insights and knowledge about decisions pertaining to dairy losses. A potential critique of sample size could relate to the number of dairy farmers included in the quantitative analysis. However, given that quantitative data for farmers were collected through interviews and the sample of eight farmers represented approximately 10% of the processor’s supplying farms, this was deemed sufficient for this study.
While greenhouse gas mitigation is a major incentive of FLW prevention research (IPCC, 2019), emissions and other environmental impacts were outside the scope of this study. Future work should utilise life-cycle assessment methodology to estimate environmental impacts and compare the influence of interventions on overall food supply chains. This research needs to take into account that milk losses in some countries and regions, may equate to different environmental impacts, for example New Zealand dairy farms are reported to have a 48% lower carbon footprint than other major dairy producing countries, due to management practices and context-specific environmental data (Mazzetto and Ledgard Reference Mazzetto, Falconer and Ledgard2020) Therefore, milk losses in NZ will have different environmental impacts compared to milk losses in other countries.
Future directions
Expanding FLW quantitative efforts through exploring technology or methods that enable accurate quantification would be highly beneficial in understanding dairy food loss quantities and help identify where prevention strategies could be most effective and/or to encourage reuse/recycle destination choices that are higher in the FLW hierarchy. Farmers are currently adopting data collection technologies such as cow collars and milk meters to improve their business outcomes that could also be applied to providing data for farm food losses. This would contribute to improving the accuracy of milk loss data and provide the opportunity for longitudinal data collection of seasonally relevant losses. For example, mastitis infections are most prevalent during calving and decrease during later stages of lactation (McDougall et al., Reference McDougall, Williamson, Lacy-Hulbert, Steele and Eastwood2025), which may mean that later in the season, when there are no calves on farm, this is when alternative destinations or solutions to dairy losses are most relevant. Further, the collection of data from more farms will enable the impact of factors such as farm size, herd numbers, or dairy parlour design on FLW to be assessed, and provide the information required for full food supply chain assessments or life-cycle assessments of FLW. Importantly, data collection needs to be paired with data sharing services, such as distributed ledger technologies like blockchain, to enable the sharing of routine FLW, one-off FLW events and evaluation of impacts to stakeholders across the supply chain. This approach would increase the visibility of FLW occurrence across the food system highlighting where FLW solutions are most needed, and where opportunities for entrepreneurs may be realised. This will also inform policy makers of practices to be encouraged or altered.
Policies could also facilitate destination prioritisation for companies to keep food loss from waste destinations through encouraging donation of out-of-specification milk powders for human consumption from low-risk sources such as retention sample excesses. Exploring these avenues is imperative to addressing inefficiencies across our food systems that negatively impact our environment, our access to food and our management of finite resources.
Conclusions
This study provides the first quantification of early-stage dairy food loss in New Zealand. It shows that losses are low: 1.8% on farm and 5.5% at the processor. However, when scaled to national production, even small percentages represent substantial volumes – equivalent to the annual milk production of approximately 86,000 cows on farm and 264,000 cows at the processor.
Our findings suggest that, from a percentage perspective, there are limited opportunities for further reducing or preventing of food loss on farm. Instead, the value of this research lies in identifying the factors, decision rationales, relationships and technology used that have already minimised losses. These included a more equitable sharing of market power and risks of food production between the processor with farmers, trading of excess milk between dairy processor companies, an MOU between dairy companies to prevent milk waste on farms and incremental technological advancement by both farmers and processor. These strategies could conceivably be adopted by other food commodities to address food loss occurrence.
The key concern about the dairy food loss occurring in this study was the limited range of alternative destination options for food loss. Although proportionally small food loss is generated compared to total milk production, this still equates to substantial quantities which could be problematic when economic, environmental or operational conditions shift leading to changes in farmer and processor drivers or destination availability. Strengthening destination pathways through developing alternatives that align with more preferable options on the FLW hierarchy is an important future opportunity for maintaining sustainability across New Zealand’s dairy sector.
Acknowledgements
This work was funded by the Circular Bioeconomy program via the AgResearch SSIF Integrative Solutions Fund.
Competing interests
The authors declare none.
Declaration of Generative AI and AI-assisted technologies in the writing process
During the preparation of this work, the author used OpenAI ChatGPT in order to reduce the word count for the abstract and Microsoft Co-Pilot to propose a title. After using this tool/service, the first author reviewed and edited the content as needed and takes full responsibility for the content of the publication.
Open access
