Harmonizing Social LCA in the Agri-Food Sector: A Literature Review

Abstract

The growing importance of social sustainability in the agri-food sector has increased attention toward Social Life Cycle Assessment as a tool for evaluating social impacts along the value chain. This review aims to explore the application of Social Life Cycle Assessment in the agri-food domain, with a particular focus on the dairy sector, to identify methodological approaches and recurring challenges. A review of the existing literature was conducted, analyzing case studies through a comparative lens based on key methodological dimensions such as stakeholder involvement, impact assessment methods, and database usage. The findings highlight a prevalent reliance on the Reference Scale impact assessment method, limited stakeholder engagement, and a general absence of materiality assessments. Furthermore, data availability and standardization remain critical issues, with most studies depending on generic databases such as SHDB. The review concludes that although progress has been made, further empirical validation and greater standardization are needed.

1. Introduction

Food systems are crucial for the well-being of humans, not only for nutrition but also for the livelihood of people worldwide. Population growth increases food demands, intensifying the sustainability concerns and emphasizing the balance of resource efficiency, social responsibility, and ecological protection [1]. The agri-food system comprises the entirety of activities related to agricultural production, industrial processing, distribution, and the consumption of food products [2]. As such, it encompasses all interconnected processes and actors involved in bringing food from the field to the consumer’s table [3]. This trend underscores the necessity of adopting a systemic approach to achieve sustainable production [4]. Indeed, preserving the natural environment and sustaining social equity are fundamental prerequisites for a well-functioning agri-food system. For the resilience of the food sector, a comprehensive view of all three pillars of sustainability (economic stability, ecological health, and social welfare) is crucial [5]. The 2030 Agenda currently serves as the main action plan for advancing sustainable development. It highlights the crucial role of these three dimensions of the Triple Bottom Line, emphasizing the need to balance and promote their integrated application [6]. Nevertheless, various social issues remain less integrated into sustainability assessments, such as workers’ equality, outward migration, labour conditions, consumer health, community well-being, and fair treatment throughout the supply chain [7]. Neglecting these issues undermines the efforts to develop a fair and resilient food system.

The dairy sector is a focal point due to its global importance and diverse social implications from farm to table [8]. It lacks key social aspects, such as migrant labour and children’s welfare [9,10]. This underlines the urgent need for a review to clarify the current state of research. The rationale for selecting this sector is also linked to current projections for global milk production, which is expected to increase by 1.4% compared to 2023, reaching nearly 979 million tons by 2024 [11]. In Europe, milk production is anticipated to rise marginally by 0.6%, reaching approximately 235 million tons [11]. This growth is primarily attributed to increased milk yield per animal and greater forage availability, which are expected to offset the decline in the number of dairy cows [11].

In this context, companies play a crucial role and must adopt appropriate methodologies and tools to support their social sustainability-related decisions [6].

Regarding the companies’ decision-making process, life cycle thinking is a valuable answer. It represents the basic qualitative concept of considering the whole product system life cycle [5]. It is made operational by the Life Cycle Sustainability Assessment (LCSA) tools [12]. These tools enable the assessment of environmental, economic, and social impacts associated with products and services [12].

The three impact assessment methodologies included in the LCSA framework reflect the three pillars of sustainability, according to the Triple Bottom Line logic, as follows: (a) Environmental Life Cycle Assessment (E-LCA) (or simply Life Cycle Assessment—LCA) to assess the environmental impacts associated with a product throughout its entire life cycle; (b) Life Cycle Costing (LCC) to evaluate the costs related to a product over its full life cycle; (c) Social Life Cycle Assessment (S-LCA) to assess social impacts, with the goal of evaluating the socioeconomic and social aspects of products and their potential positive or negative effects throughout their entire life cycle. The aspects considered in S-LCA include the indirect impacts on stakeholders resulting from company behaviour [13].

Although the number of E-LCA studies still exceeds that of S-LCA, the latter has been progressively developed and is now increasingly adopted for evaluating various products and organizations [4].

Research on S-LCA began in the mid-1990s and experienced substantial development starting in 2005 [14]. Ramos Huarachi et al. [15] identified four key phases in the evolution of this methodology: the preliminary phase (1996–2009), the period of uncertainty (2009–2012) marked by the release of the first S-LCA Guidelines in 2009 [16], the advancement phase (2013–2016) characterized by the publication of the initial Methodological Sheets in 2013 [17], and the current phase emphasizing standardization (from 2017 onward). The latter period has seen a significant rise in the number of publications [15].

It is also worth noting that the publication of case-study-based papers increased substantially after 2009, reflecting the scientific community’s growing interest in applying the S-LCA Guidelines [18].

The latter, initially released in 2009 as part of the UNEP/SETAC Life Cycle Initiative, have undergone significant refinement and widespread application in recent years. These advancements have contributed to a more detailed definition of the methodology, including its indicators and impact assessment procedures [19], culminating in the release of an updated version in 2020.

In addition to the 2020 S-LCA Guidelines [20], the S-LCA methodology also aligns with the steps outlined in ISO 14040 [21]. Specifically, the S-LCA framework can be regarded as an extension of environmental impact assessment practices, as it follows a similar four-phase structure defined in ISO 14040: (1) definition of goal and scope, (2) life cycle inventory analysis, (3) impact assessment, and (4) interpretation of results [21].

In response to the growing demand for conducting S-LCA, the ISO 14075 standard was introduced last year [22]. This new norm outlines the requirements and offers guidance for the development and implementation of methodologies aimed at evaluating social impacts. Notably, the analytical steps remain consistent with those established in ISO 14040.

In the agri-food system, social impact has been examined to some extent but current assessments remain limited in scope and often fragmented [10,23,24,25]. This review contributes to the existing body of literature on the evaluation of the S-LCA methodology, building upon prior publications addressing specific aspects of its development, application, and limitations. Tragnone et al. [26], for instance, conducted a review focused specifically on the application of S-LCA in the agri-food sector. Ramos Huarachi et al. [15] provided an overview of the historical evolution of the method, identifying key research trends. Bonilla-Alicea and Fu [27] critically examined the methodology of Social Impact Assessment (SIA), aiming to highlight current trends, state-of-the-art practices, limitations, and knowledge gaps.

Petti et al. [28] reviewed the S-LCA methodology through its application in case studies, offering valuable insights into practical implementation. Rezaei Kalvani et al. [29] carried out a narrative review aimed at identifying existing gaps in product-based S-LCA and uncovering methodological weaknesses.

Pollock et al. [30] contributed by tracing past developments and pointing out methodological barriers to the wider adoption of S-LCA. Sureau’s analysis centred on the assessment criteria and indicators, focusing on their definition and selection processes [14]. Arcese et al. [31] aimed to systematize the diverse contributions to S-LCA, proposing a classification of emerging themes and methodological approaches. Finally, Rebolledo-Leiva et al. [32] explored the state of the art of the S-LCA methodology within the specific context of the bioeconomy framework.

Collectively, these studies underscored the ongoing challenges in achieving methodological clarity, consistency, and applicability in S-LCA publications.

Following previous research, this article seeks to advance the debate by analyzing empirical applications of Social Life Cycle Assessment (S-LCA) in the dairy sector.

This article reviews different studies to analyze the application of various methodological phases and approaches. Hence, the study aims to provide a better understanding of current practices and to underscore the areas where methodological development is further required. Our article has two research questions:

• What are the main phases of the S-LCA methodologies employed in the papers about the dairy sector?
• What are the differences between the publications in the aforementioned phases?

To achieve the research aim, a literature review is conducted by the authors, and a critical review by means of a descriptive and in-depth analysis is also carried out to synthesize the results.

The remainder of the article is structured as follows: Section 2 concerns the materials and methods employed in the paper. Section 3 presents the results of the analysis of all papers, with a specific focus on those related to the dairy sector. Section 4 discusses the results in light of the relevant literature in the field. Finally in Section 5, the conclusions, limitations, and directions for future research are presented.

2. Materials and Methods

The analysis of applications of the S-LCA methodology [20] in empirical studies related to the agri-food sector consists of a descriptive investigation of the various S-LCA steps followed by the authors in conducting their studies. To identify practical applications of S-LCA, a literature review was carried out to gather empirical cases.

In order to explore the differences and similarities among the selected papers, a critical review was also conducted. According to Grant and Booth [11], a critical review enables a comprehensive examination of the literature and an assessment of its quality. Rather than merely summarizing existing studies, it provides analytical depth and conceptual insight.

The case studies analyzed in this review were gathered from scientific articles and reports retrieved through the Scopus and Google Scholar databases, with a focus on the agri-food sector, in particular, on the dairy products. Adopting two databases is necessary for systematic reviews in order to not depend just on one database and include a wider range of publications in the analysis [33]. Since the authors adopted a less stringent methodology than that of a systematic review, only Scopus was used as the primary database due to its extensive use in academic research [34] and to its breadth and depth in indexing interdisciplinary peer-reviewed papers [33]. Owing to the limited number of articles obtained, it was decided to retrieve additional articles from the Google Scholar database using the cross-reference technique [35].

Unlike other databases, Google Scholar indexes the full text of papers, allowing it to retrieve documents in which the search keywords appear anywhere in the text [36]. The literature search was conducted using keywords selected to include agri-food-related terms (with a focus on dairy products). The used query was as follows:

(TITLE-ABS-KEY(“social life cycle assessment”) OR TITLE-ABS-KEY(“s lca”) OR TITLE-ABS-KEY(“s-lca”) OR TITLE-ABS-KEY(“so-lca”) OR TITLE-ABS-KEY(“social lca”) AND TITLE-ABS-KEY(agrifood) OR TITLE-ABS-KEY(“agri food”) OR TITLE-ABS-KEY(“agri-food”) OR TITLE-ABS-KEY(milk) OR TITLE-ABS-KEY(cheese) OR TITLE-ABS-KEY(dairy)) AND (EXCLUDE (SUBJAREA,”COMP”) OR EXCLUDE (SUBJAREA,“CENG”) OR EXCLUDE (SUBJAREA,“BIOC”) OR EXCLUDE (SUBJAREA,“NURS”) OR EXCLUDE (SUBJAREA,“MATE”) OR EXCLUDE (SUBJAREA,“EART”) OR EXCLUDE (SUBJAREA,“MEDI”) OR EXCLUDE (SUBJAREA,“MATH”) OR EXCLUDE (SUBJAREA,“CHEM”))

The initial search on Scopus yielded 29 publications. Subsequently, after removing not relevant subject areas (such as Computer Science, Chemical Engineering, Biochemistry, Genetics and Molecular Biology, Nursing, Materials Science, Earth and Planetary Sciences, Medicine, Mathematics, and Chemistry), only nine contributions were considered appropriate for the scope of the present study. Indeed, 12 papers were not selected from the Scopus database due to their research methodology (7 were literature reviews or conceptual rather than empirical papers); 2 articles had limited relevance to the research objective; 1 involved all the methodologies of the LCSA framework (and not just the S-LCA); 1 summarized case studies already specifically analyzed in their respective original publications; and 1 was not accessible. Subsequently, in order to obtain a broader set of case studies, the analysis was expanded by including publications identified in the Google Scholar database, which yielded an additional nine case studies.

All the papers in this review were included in the analysis due to their relevancy for the scope of the study.

In order to conduct the critical review, the authors collected all the papers included in the review in an Excel spreadsheet [37]. Every record in the sheet corresponded to a publication and the authors included information such as the scope of the study; the impact assessment method; software and databases used; Materiality Assessment; Social Hotspot Analysis (SHA); Social Life Cycle Inventory (S-LCI); and Social Life Cycle Impact Assessment (S-LCIA) as the actual phase of assessing social impacts, along with the results of the S-LCIA or SHA phase.

Table 1. Relevant information classified for the critical review of the papers.

Evaluation Criteria	Description
Company/topic	The name of the company (in the case of studies related to specific companies) or the analyzed topic (for publications related to countries or geographic areas)
Type of business	Explicitized just for studies on specific companies
Research objective	To identify the purpose of the case study
Scope of the study	It included the phase of “Goal and Scope” established by UNEP [20]. In particular, it comprised the definition of the functional unit, system boundaries, stakeholders and activity variables, and criteria for excluding processes from the product system
Impact assessment method	Reference Scale Approach or Impact Pathway Approach [20]
Software and databases	To gather and analyze secondary data on social impacts
Materiality Assessment	To identify the most relevant issues according to their impact on stakeholders and/or on the business [20]
SHA	To identify activities where social issues (as impacts) and/or social risks are likely to occur [20]
S-LCI	The data collection step necessary for the impact assessment phase
Source of inventory indicators	Which source (specific databases, Methodological Sheets, etc.) was used to select the indicators to be employed?
S-LCIA	The actual phase of assessing social impacts. In the Excel sheet, the methodology and results of the SLCIA phase are reported in two separate columns of the file
Results of the S-LCIA or SHA phase	The results of SHA were reported in cases where only the hotspot analysis is conducted
Source of the publication	Reference of the paper

contains a visual explanation of the information considered.

3. Results

In the present section, the bibliometric characteristics of the selected papers are presented in a descriptive analysis. Then, a critical review is conducted to identify: 1. the functional units (FUs) related to each paper and the corresponding sector; 2. the differences between studies related to the dairy sector based on specific methodological evaluation criteria.

3.1. Descriptive Analysis

The bibliometric characteristics of the papers investigated are related to the distribution trends per year and country. As shown in Figure 1, the distribution of publications by year is uneven. Although the first version of the S-LCA Guidelines by UNEP and SETAC was published in 2009, the first empirical publication dates back to 2012. In the following years, a modest upward trend can be observed, with a few exceptions in years with no publications. Finally, the last four years considered (2021–2024) represent the most consistently prolific period in terms of empirical case studies in the agri-food sector applying the S-LCA methodology.

Regarding the countries referenced in the papers for their case studies (

Table 2. Geographic distribution of publications included in review.

Country	Count	World Region
Italy	7	Global North
Canada	2
USA	1
Norway	1
Ireland	1
Indonesia	1	Global South
Costa Rica	1
Malaysia	1
More than one country	3

), Italy ranks first, with a total of seven published papers. The label “More than one country” refers to studies that analyzed case studies in different countries; the three papers in this category include two publications focused on Europe and one focused on Iran and Malaysia. Apart from Canada, which has two publications, Indonesia, Costa Rica, USA, Norway, Ireland, Brazil, and Malaysia (as a single country) addressed all just one case study related to the agri-food sector.

Empirical cases related to the S-LCA methodology in the agri-food sector are mainly concentrated in the West, with a few exceptions in the Global South (Costa Rica) and in Asia (Malaysia and Indonesia). This is also outlined by some authors [18,27,28], which confirmed that the S-LCA field of research became more popular in the past ten years, with a clear majority of European authors. This also corroborates what was found by Tragnone et al. [26], who noted that the field is still evolving but dominated by work from Western countries.

3.2. Functional Units and Commodity Sectors

The first thing to consider in order to connect each publication in the review to a commodity sector is the FU. The S-LCA Guidelines [20] identify the FU as a core element of the “Goal and Scope” phase of S-LCA. In this regard, among the publications included in this review, all contain a FU, with the exception of five papers, which differ from each other: three of them do not define either a FU or a Reference Flow [38,39,40]; Mancini et al. [41] do not define any criteria from the “Goal and Scope” phase (this is because it is a macroscale case study, which focuses on assessing social risks associated with food consumption in the EU); Benoit-Norris et al. [42] define the FU in monetary terms, since when using an S-LCA database, the FU is “typically expressed in monetary terms because of the use of trade models” [7]; for Møller et al. [4], instead of expressing the FU in terms of product weight, they define it according to nutritional criteria, namely as the “average amount of domestically produced animal protein per capita”; lastly, De Luca et al. [43] and Iofrida et al. [44] adopt a land-based FU, referring to the cultivated area of citrus rather than product weight. Based on the above considerations,

Table 3. Functional Units of case studies and related sector.

Reference	Functional Unit	Sector
Pelino et al. [45]	1 batch of Confetti (60 kg)	Confectionery
Tragnone et al. [46]	1 batch of Confetti (60 kg) and 1 batch of Tenerelli (130 kg)	Confectionery
Saputra et al. [47]	1 kg of live broiler chicken	Poultry/livestock
Petti et al. [48]	1 kg of tomato “cuore di bue”	Agriculture
D’Eusanio et al. [6]	1 jar of honey (500 g)	Beekeeping
Tragnone et al. [49]	1 jar of orange blossom honey (250 g)	Beekeeping
Brenes-Peralta et al. [50]	1 kg of green coffee, 1 kg of raw milk, 1 kg of lettuce	Multisectoral (agriculture, dairy)
Benoit-Norris et al. [42]	USD 1 million of strawberry yogurt	Dairy
Møller et al. [4]	The average amount of domestically produced animal protein per capita in Norway	Multisectoral
Revéret et al. [51]	1 kg of milk	Dairy
Chen & Holden [52]	1 kg of milk	Dairy
De Luca et al. [43]	1 ha of clementine orchard	Agriculture
Iofrida et al. [44]	1 ha of citrus orchard	Agriculture
Tallentire et al. [53]	1 kg of chicken meat	Poultry/livestock
Kalvani et al. [38]	/	Agriculture (rice)
Muhammad et al. [39]	/	Agriculture (palm oil)
Graham et al. [40]	/	Poultry/livestock (beef meat)
Mancini et al. [41]	/	Multisectoral (44 products)

presents the FUs identified for each case study, together with their respective commodity sectors.

The last four sectors in Table 2, referring to studies in which the FU is not specified by the authors, refer to the production of rice [38], palm oil [39], beef meat [40], and 44 representative products in the context of the European Union [41].

Three studies adopt a multisectoral approach, considering several FUs. In this regard, Brenes-Peralta et al. [50] consider different FUs related to different sectors, namely agriculture and livestock, associated with coffee, milk, and lettuce. The animal proteins considered by Møller et al. [4], on the other hand, include milk, beef, pork, poultry, and eggs. Finally, the 44 products referenced by Mancini et al. [41] belong to various commodity sectors: agriculture (e.g., apples, bananas), dairy (e.g., milk cream, cheese), livestock and poultry (e.g., pig meat, cattle meat), aquaculture and fishery (e.g., salmon, shrimps), and beverages (e.g., wine, beer). The three studies which do not specify a FU [38,39,40] addressed, respectively, rice, palm oil, and beef meat.

3.3. Dairy Sector

Considering what is mentioned in the previous subsection and the focus of the authors on the dairy sector, Figure 2 shows the number of papers that address this topic.

In total, 3 publications out of the 18 included in this review explicitly address the dairy sector [42,51,52]. Although the studies by Brenes-Peralta et al. [50] and Møller et al. [4] are categorized as “multisectoral”, they are nonetheless relevant to the dairy context, bringing the total number of studies dealing with dairy products to five. Mancini et al. [41] also included dairy products among the 44 products analyzed as representative of European consumption, but their study was not considered specific enough to be classified within the dairy sector.

S-LCA Methodology in Dairy Sector

In line with the “Scope of the Study” evaluation criteria and the “Goal and Scope” phase outlined in the S-LCA Guidelines [20], empirical S-LCA studies are required to define key methodological elements such as system boundaries, stakeholders, activity variables, and cut-off criteria.

System boundaries, as defined by ISO 14040 [21], “specify which unit processes are part of a product system” and thus determine which portions of the product system are included in the assessment [20]. Ideally, these boundaries should encompass both foreground processes (those directly linked to the FU) and background processes (those further from the FU, identified as either upstream or downstream processes, for which primary data are often unavailable) [20]. Although a cradle-to-grave approach is recommended, in practice, the scope of the analysis is frequently constrained by the study’s objectives, available resources, and data limitations [20]. Complementary to the system boundaries are the cut-off criteria, which allow for the exclusion of specific portions of the product system. In particular, they enable the definition of thresholds for excluding certain unit processes or parts of the product system from the study, based on the amount of material or energy flow or their level of significance [21].

ISO 14040 refers to the so-called “interested party”, defined as an “individual or group concerned with or affected by the […] performance of a product system, or by the results of the life cycle assessment.” However, the S-LCA Guidelines (2020) use the term “stakeholder” to refer to an “individual or group that has an interest in any activities or decisions of an organization”, in accordance with ISO 26000 [54]. The S-LCA Guidelines [20] also specify distinct stakeholder categories, defined as clusters of stakeholders who are expected to have similar interests due to their similar relationship to the investigated product system. These categories are Children, Local Community, Workers, Value Chain Actors, Consumers, and Society.

Finally, the activity variable represents a measure of the scale or intensity of a process that can be linked to its output. When scaled according to the output of each relevant process, activity variables help quantify the proportion of a specific activity attributed to each unit process [20]. Typically, the activity variable refers to the working hours related to each process unit, thus allowing researchers to determine which unit process is likely to be more affected by social impacts.

Table 4. Identification of system boundaries, cut-off criteria, stakeholders, and activity variables related to dairy sector papers.

Criteria	Benoit-Norris et al. [42]	Revéret et al. [51]	Chen & Holden [52]	Brenes-Peralta et al. [50]	Møller et al. [4]
System boundaries	Cradle-to-grave	Cradle-to-gate	Cradle-to-gate	Cradle-to-farm gate	Cradle-to-grave
Cut-off criteria	Social impacts that account for less than 0.1% of the total working hours	Production factors and expenses not directly related to milk production	Processes excluded for lack of data	/	Working hours and hotspot analysis
Stakeholders	/	Workers, Local Community, Society, Value Chain Actors	Workers, Local Community, Society, Value Chain Actors	Workers, Farmers, Local Community	Workers, Local Community, Society, Value Chain Actors, Children, Consumers
Activity variable	Working hours	/	/	/	Working hours, distribution of proteins

presents the system boundaries, cut-off criteria, stakeholders, and activity variables of the studies considered relevant to the dairy sector.

With regard to system boundaries, most of the case studies did not cover the full life cycle of the products under analysis but limited their analysis to the “gate” of the company.

Revéret et al. [51] explicitly specify the excluded processes and costs, including production factors associated with farm buildings, cow replacement, electricity use, and other expenses not directly related to milk production. Chen and Holden [52] justify their exclusions based on data limitations, identifying this as a cut-off criterion. In their study, some of the processes were excluded due to the lack of available data. Other authors excluded certain processes from the product system based on their limited contribution to the life cycle—both in terms of quantity and labour hours as well as on their low relevance identified through hotspot analysis [4,42].

Among the stakeholder categories considered, only one case study [4] included all six categories specified in the S-LCA Guidelines [20]. Benoit-Norris et al. [42] did not specify any stakeholder group due to the use of SHDB. Brenes-Peralta et al. [50], instead, introduced another category of stakeholders, which is “farmers”, because of the importance that medium or small family farmers hold for rural activities in Costa Rica. This is comparable to the “smallholders including farmers” subcategory stated in the S-LCA Guidelines [20].

In total, only two studies employed an activity variable. This is consistent with the findings of Tragnone et al. [26], who observed that only half of the studies analyzed employed an activity variable. Benoit-Norris et al. [42] used working hours as an activity variable, drawing on the Social Hotspot Database (SHDB) to conduct an S-LCA of strawberry yogurt. Møller et al. [4] also used working hours; however, this metric was considered inappropriate for evaluating the subcategory “animal welfare” within the Society stakeholder group. To address this, they applied the distribution of protein production, which reflects the proportional contribution of each animal species. In this study, for the stakeholder groups Value Chain Actors, Consumers, and Children, no activity variable was required [4].

The software and databases used in studies related to the dairy sector may be of particular interest. It should be noted that, in this study, only the databases specifically intended for the calculation of social impacts using the S-LCA methodology, as well as the corresponding software (when available), are included in

Table 5. Software tools and databases specifically intended for S-LCA studies within dairy sector publications.

Criteria	Benoit-Norris et al. [42]	Revéret et al. [51]	Chen & Holden [52]	Brenes-Peralta et al. [50]	Møller et al. [4]
Software	/	/	Gabi 6	/	/
Specific S-LCA database	SHDB	SHDB	Life Cycle Working Environment (LCWE)	/	SHDB

. The studies in this review employed SHDB and LCWE. The latter works in combination with GaBi software (version 6).

SHDB is the first commercial S-LCA database [32] released by New Earth B. It consists of CSS indicator tables that support the identification of social hotspots (i.e., countries and sectors of concern within Value Chain Actors) based on potential social impacts. It is also integrated with a model designed to determine the countries and sectors with the highest share of working hours, known as the Worker Hours Model. CSS are identified as hotspots when they account for a significant share of total working hours, are classified as high-risk according to the Hotspot Index, and have been recognized as critical by multiple sources in the literature [42]. As for the software, no specific tools used with SHDB are mentioned in the papers (e.g., SimaPro, OpenLCA, or Quantis Suite) [55].

The other mentioned specific S-LCA database is the LCWE. This database was developed by the Chair of Building Physics at the University of Stuttgart to incorporate social dimensions into LCA using the GaBi software. This approach enables the integration of high-quality, process-specific social data into product LCA by drawing on statistical data sources [56].

The S-LCA Guidelines [20] state that the impact assessment method chosen for the analysis should be defined in the “Goal and Scope” phase. These Guidelines mention two methods of assessment: Reference Scale (RS) S-LCIA and Impact Pathway (IP) S-LCIA. The RS S-LCIA (also known as the “Type I” approach) evaluates the social performance of organizational activities by comparing them against defined benchmarks known as Performance Reference Points (PRPs) using a Reference Scale (RS). This approach comprises the definition of impact subcategories related to different stakeholders. The IP S-LCIA approach, instead (known as the “Type II” approach) employs a process known as “characterization” to establish the link between organizational activities and their potential or actual social impacts. This approach focuses on identifying and tracking the outcomes of such activities, including their prospective long-term implications [20]. This method implies the use of impact categories connected to their related stakeholders.

As reported in

Table 6. Impact assessment methods within dairy sector publications.

Criteria	Benoit-Norris et al. [42]	Revéret et al. [51]	Chen & Holden [52]	Brenes-Peralta et al. [50]	Møller et al. [4]
Impact assessment approach	RS S-LCIA	RS S-LCIA	RS S-LCIA and IP S-LCIA	RS S-LCIA	RS S-LCIA

, all the publications included in the “dairy sector” adopted the RS S-LCIA approach, with the exception of Chen and Holden [52], who applied a mixed method. Indeed, according to Pollock et al. [30], database methods (such as SHDB methodology) can be included in the RS S-LCIA approach. Regarding Chen and Holden [52], Group 1 impact subcategories (quantitative and attributable to the functional unit) were treated as IP S-LCIA impact categories, whereas Group 2 (quantitative but not attributable to the functional unit) and Group 3 (semi-quantitative) impact subcategories were considered RS S-LCIA impact categories. The Subcategory Assessment Method (SAM), which was used by Brenes-Peralta et al. [50], is considered part of the RS S-LCIA approach. It uses a four-level measurement scale (A, B, C, and D) for each impact subcategory. Level B represents the compliance level, meaning that the organization, with respect to that subcategory, meets the Basic Requirements (BRs), which are defined, for instance, in accordance with international agreements, national laws, or best practices [57].

Although the steps required to conduct an S-LCA are outlined in both the 2020 S-LCA Guidelines [20] and ISO 14040 [21], each of the case studies reviewed applies a different methodology.

Table 7. S-LCA methodology steps identified in case studies.

	SHA	S-LCI		Source of Indicators	S-LCIA Method
		Source of Data	Type of Data
Benoit-Norris et al. [42]	The assessment is conducted using SHDB	SHDB draws secondary data from additional databases.	Secondary	SHDB	SHDB methodology.
Revéret et al. [51]	Potential Hotspot Analysis (PHA), including a literature review, consultation of websites, and SHDB	Data gathered through questionnaires for the specific analysis (SA).	Primary and secondary	Indicators for the SA were defined based on frameworks proposed in the literature and expert judgement	RS S-LCIA approach. Aggregation of indicators within impact subcategories and comparison with a PRP.
Chen & Holden [52]	/	Data were obtained from reports, databases, and websites.	Secondary	Indicators were retrieved from databases and reports	RS S-LCIA approach and IP S-LCIA approach.
Brenes-Peralta et al. [50]	/	Primary data gathered from focus groups, questionnaires, and on-site observations; secondary data are derived from primary documentation, reports, and databases.	Primary and secondary	Indicators are retrieved from the Methodological Sheets [17]	SAM [57]. The Basic Requirements (BRs) are established based on national legislation and context-specific practices.
Møller et al. [4]	SHA used for identification of impact subcategories	Secondary data are obtained from databases, statistics, indexes, surveys, and websites.	Secondary	Indicators are developed based on the Methodological Sheets [58] and impact subcategories were selected based on a stakeholder survey and a SHA using SHDB	RS S-LCIA approach with four levels. For each indicator, a Reference Scale was developed using global standards or indexes.

summarizes the main phases identified in the papers (excluding the already discussed “Goal and Scope” phase and the “Interpretation” phase). Additionally, the SHA is not explicitly defined as a formal stage of the assessment in either the S-LCA Guidelines [20] or the ISO standard [21] but is frequently adopted in the studies.

Concerning the SHA, only three out of the five publications included this methodological step. The study by Benoît-Norris et al. [42] explicitly applied the SHDB methodology, thereby making the entire study a hotspot analysis. The Potential Hotspot Analysis (PHA) employed by Reveret et al. [51], which is comparable to SHA, involved the use of tools such as literature review, website consultation, and SHDB. Finally, Møller et al. [4] applied SHA to determine which impact subcategories would be relevant for the case study analyzed.

Regarding the type and source of data for the S-LCI phase, all studies utilized secondary data, though to varying extents and from different origins. Benoît-Norris et al. [42], Chen and Holden [52], and Møller et al. [4] relied exclusively on secondary data derived from structured databases such as SHDB and other secondary sources (i.e., reports, databases, and websites). In contrast, Reveret et al. [51] and Brenes-Peralta et al. [50] complemented secondary sources with primary data collection, including questionnaires, focus groups, and on-site observations.

Concerning the source of indicators, Benoît-Norris et al. [42] based their indicators directly on SHDB, whereas Chen and Holden [52] sourced them from other existing databases and reports without using SHDB. Reveret et al. [51] defined their indicators using literature-based frameworks combined with expert judgement, showing a higher degree of customization. Brenes-Peralta et al. [50] and Møller et al. [4] developed indicators based on the Methodological Sheets, with the latter also incorporating stakeholder surveys and SHA results for the selection of impact subcategories.

With regard to the S-LCIA method, there is methodological heterogeneity. Benoît-Norris et al. [42] adopted the SHDB methodology, while Reveret et al. [51] and Chen and Holden [52] applied the RS S-LCIA approach, with the latter grouping impact subcategories into three classes. Chen and Holden [52] also combined RS S-LCIA with the IP S-LCIA approach. Brenes-Peralta et al. [50] employed the SAM, aligning Basic Requirements with national legislation and context-specific practices. Finally, Møller et al. [4] applied the RS S-LCIA approach with four levels, developing a reference scale for each indicator based on global standards or indexes.

4. Discussion

This critical review has provided insights into the practical implementation of the S-LCA methodology.

Regarding the FU, 78% of all publications considered in the review have defined it, supporting the trend identified by previous authors regarding the definition of the FU [28]. In this percentage, all the studies related to the dairy sector have established a FU.

Nevertheless, while the S-LCA Guidelines recognize the use of FUs as advantageous for comparing results, they also acknowledge that “some potential social impacts do not depend on physical flows and the nature of unit processes, but rather on the behaviour of the companies and stakeholders involved in the life cycle” [20].

Indeed, the inherently qualitative nature of social phenomena often poses significant challenges in linking social impacts (or their indicators) to the FU or in normalizing data accordingly [30].

In this review, “Agriculture” emerged as the most investigated sector, with a total amount of five papers. This is supported in the literature: even in studies where the agri-food sector was not the central focus [27,28,29], “Agriculture” consistently emerges as the most extensively examined sector with respect to the FU defined in each publication.

With specific reference to the literature pertaining to the dairy sector, the majority of the studies conducted the analysis from a “cradle-to-gate” perspective. This observation is also supported by Tragnone et al. [26], who found that the majority of empirical studies adopted a “cradle-to-gate” approach rather than a “cradle-to-grave” perspective, despite the S-LCA Guidelines [20] indicating a preference for the latter. Nevertheless, the selection of a particular perspective (along with the definition of cut-off criteria) should be thoroughly justified by the authors to ensure methodological transparency and rigour. This does not always appear to be the case, as previously highlighted by Tragnone et al. [26]. In this study, Brenes-Peralta et al. [50] did not explicit the cut-off criteria. The S-LCA Guidelines [20] allow considerable flexibility for researchers to establish their own cut-off criteria to justify the exclusion of certain processes or stakeholders. This often leads to a wide range of approaches or, in some cases, a lack of clear explanation [30]. Moreover, there is no consensus within the S-LCA community regarding how to define the socially relevant system boundaries and apply appropriate weighting systems [29].

Among the five studies addressing social impacts in the dairy sector, four explicitly identified the stakeholder groups considered in their assessments [4,50,51,52]. The stakeholder category most extensively examined was that of Workers, with a study extending this group to include “farmers” [50]. Other stakeholder groups, listed in descending order of representation, included Local Communities, Society, Value Chain Actors, Consumers, and Children. Notably, only one study [4] incorporated all six stakeholder categories within its analytical framework. Previous literature has affirmed that Workers represent the most extensively investigated stakeholder group in S-LCA applications [18,29], possibly due to the fact that, in terms of social impacts, labour exploitation remains one of the primary concerns [26]. The findings of this review are consistent with the existing literature. Stakeholder coverage in S-LCA studies remains uneven [30], with certain categories, particularly Consumers, often underrepresented. While the production phase is frequently prioritized, the consumption stage and its associated social impacts receive comparatively limited attention [26,29]. In support of this, in our review, just one publication addressed the Consumers category [4]. It is notable that the stakeholder group Children was included in only one study [4], a limitation that can be attributed to the structure of the S-LCA Guidelines themselves. Specifically, in the earlier version [16], only five stakeholder categories were defined, with Children being introduced as an additional category in the revised 2020 edition [20]. In general, it can be affirmed that few studies comprehensively address all relevant stakeholder groups [29].

Nevertheless, it is important to note that the S-LCA Guidelines [20] do not require authors to include all stakeholder categories and impact subcategories. On the contrary, it is essential to clearly identify which relevant stakeholders to include, ensuring transparency in the selection process and providing justification. If necessary, additional stakeholder categories not explicitly listed in the S-LCA Guidelines may also be identified [20].

With regard to activity variables, only two out of the five studies identified working hours as an activity variable. However, in accordance with the S-LCA Guidelines, value added may also serve as a valid choice [20].

As suggested in preliminary research [26] and even in this study, PSILCA (Product Social Impact Life Cycle Assessment database) was not used as an S-LCA-specific database. Instead, databases such as the SHDB and the LCWE are employed. The SHDB has emerged as the most widely utilized resource, establishing itself as a valuable instrument for conducting S-LCA. Consequently, it has been employed across a broad range of studies spanning diverse products and industrial sectors [15]. The PSILCA database is also a valuable alternative and has begun to be applied in some studies [15]. It is important to emphasize that S-LCA-specific databases (e.g., SHDB, PSILCA) cannot be applied at local or regional scales [29], and are therefore primarily used for generic S-LCA and hotspot analyses.

The majority of the case studies in this research utilized the RS S-LCIA approach. This is confirmed by Tragnone et al. [26], who highlighted the limited adoption of IP S-LCIA, thereby underscoring the ongoing challenges associated with the development and implementation of impact pathways for the evaluation of social impacts. IP S-LCA makes it possible to predict effects or to elucidate causal mechanisms; instead, the RS S-LCIA approach offers a descriptive representation of the existing conditions [44]. According to Pollock et al. [30], one of the primary challenges associated with the adoption of IP S-LCIA lies in its reliance on quantitative data to define impact pathways (which is frequently unavailable due to the limited accessibility to causal relationships underpinning social and socioeconomic impacts).

Lastly, Sureau et al. [14] suggested to combine the RS S-LCIA and IP S-LCIA approaches, as carried out by Chen and Holden [52]. Overall, the ongoing need for standardization in the S-LCA methodology underscores the recommendation to use well-established frameworks (e.g., SAM), unless the validity of a novel model is demonstrated through empirical application.

Among the studies examined within the dairy sector, none incorporated a materiality assessment into their methodological framework.

For the agri-food sector, materiality assessment is recommended due to the complexity of this sector. The selection of social themes and indicators should be clearly justified and, preferably, derived from a stakeholder-engaged materiality assessment. Without this process, studies risk overlooking the most relevant social issues in the specific context examined. Yet, only a small proportion of studies actively involve stakeholders in determining the material issues to be addressed [26]. In this review, Møller et al. [4] engaged stakeholders in the identification of impact subcategories by asking them to evaluate the relative importance of the various subcategories.

Although stakeholder engagement is widely acknowledged as desirable during the data collection phase [26], in practice, some stakeholders may be reluctant to disclose sensitive information (e.g., the number of undocumented workers). Furthermore, both geographical and cultural factors must be considered when selecting impact subcategories and conducting primary data collection through interviews [30].

The process of data collection constitutes the most resource-intensive and time-demanding phase in the implementation of S-LCA [29].

It is well-established that directly collected (primary) data are more appropriate for assessing foreground processes, while secondary data are typically employed for background processes [30]. As demonstrated in this review, however, there are case studies in which results are derived exclusively from secondary data [4,42,52], and no study relied solely on primary data. The approach of combining primary and secondary data may allow for richer context-specific insights, while using only secondary sources favours replicability and methodological standardization.

Due to the demanding nature of collecting primary data (both in terms of effort and resources), this step frequently influences the delineation of system boundaries, stakeholder selection, and indicator definition [26].

With regard to the social implications related to S-LCA, the authors believe that in publications related to the dairy sector, some potential hotspots are overlooked.

In general, the Methodological Sheets [58] do not distribute impact subcategories evenly across all stakeholders: the groups Worker, Local Community, and Society present the highest number of subcategories, whereas Consumers, Children, and Value Chain Actors are the most overlooked.

With regard to the Worker stakeholder, the Methodological Sheets [58] make no mention of workers’ psychological well-being, which could be of particular interest in primary sectors such as livestock farming. According to the European Parliament, agricultural workers have significantly higher mortality and injury rates compared with other sectors, face multiple health risks, and perform physically demanding tasks. They are also more likely to experience stress and have higher suicide rates [59]. None of the five studies examined investigates the psychological well-being of employees in livestock farms or cheese factories.

The situation of migrant workers is also a topic of interest. Although the subcategory “Delocalization and Migration” is included under the Local Community stakeholder, the Methodological Sheets [58] do not specify any indicators for this category of workers within the Worker stakeholder, which could be more appropriate. In 2021, it was estimated that 71% of non-EU seasonal agricultural workers in the EU were not employed under the Seasonal Workers Directive and, as such, were not entitled to its provisions concerning rights and social protection. Seasonal and migrant workers frequently lack access to essential social benefits, including healthcare and unemployment assistance. Their temporary or migrant status may exclude them from eligibility for these benefits, leaving them vulnerable to financial insecurity and social exclusion, particularly during periods of unemployment or illness [60]. Of the five dairy sector studies reviewed, only Brenes-Peralta et al. [50] examined the psychosocial conditions of migrant workers, assessing their integration into the local community and the existence of company policies aimed at ensuring decent living conditions.

On the topic of equal opportunities for workers, beyond the gender balance already mentioned in the Methodological Sheets [58], it would also be appropriate to investigate the presence of workers with disabilities in farms and cheese factories, an aspect neglected both by the S-LCA Guidelines [20] and by publications on the dairy sector.

Regarding the Society stakeholder category, it includes the impact subcategory “Ethical Treatment of Animals”, but it would be preferable to recognize animals as stakeholders in their own right rather than as part of another stakeholder group. This approach would better highlight the prevention of abusive treatment and the adoption of practices and policies consistent with animals’ nature and well-being. In such cases, species-specific indicators should be developed, as already recommended in the S-LCA Guidelines [7]. It should also be noted that food safety is directly linked to animal welfare, particularly for animals raised for food production, given the close relationship between animal welfare, animal health, and foodborne diseases. Therefore, failure to respect animal welfare can represent a risk to consumers [61]. Among the five dairy sector publications considered in this paper, Chen and Holden [52] and Brenes-Peralta et al. [50] make no reference to the concept of animal welfare.

As for the Children and Consumers stakeholders, it could be useful to assess, respectively, within the subcategories “Health Issues for Children as Consumers” and “Health and Safety”, the nutritional quality of the product sold by the company and its potential impacts on human health. At present, this specific indicator is not included in the UNEP (…) Methodological Sheets, and none of the dairy sector studies have proposed its use.

Finally, as previously mentioned, in the S-LCA Guidelines [20], the impact subcategories relating to Value Chain Actors are few in number and differ from those concerning the Worker stakeholder, which examine the conditions of employees of the company under analysis. When data permit, it would be valuable to investigate suppliers’ equal opportunity policies or their treatment of animals (in cases where the main company is a processing facility rather than a livestock farm).

5. Conclusions

This review has highlighted the growing interest in applying S-LCA to the agri-food sector. Despite the increase in research in recent years, the methodology still faces substantial challenges related to data availability, the lack of standardized procedures, and the heterogeneity in the selection and application of indicators, stakeholder categories, and impact subcategories.

The findings reveal that most of the studies in the dairy sector often rely on secondary data, such as databases, reports, and websites, and that stakeholder engagement remains limited and often superficial. Moreover, SHA, which could be the best use of an S-LCA-specific database, is not always employed. A further point of concern is the underutilization of materiality assessments.

Furthermore, the outcomes associated with social impacts are likely to exhibit substantially greater variability than those related to most environmental impacts, since they are influenced primarily by organizational behaviour and the geographical context rather than by the intrinsic nature of the underlying process [20]. Because of the highly context-dependent nature of S-LCA, a limitation of this study is that the papers analyzed pertain only to a specific part of the world, namely Western countries, and this may compromise the generalizability of the findings. Consequently, more research is needed on social impacts in Eastern contexts.

Moreover, future studies on S-LCA empirical analysis should prioritize primary data collection, stakeholder participation, and context-specific indicators. Efforts should be made to enhance methodological harmonization. Indeed, as underlined in previous research, academics accept the general structure of the S-LCA Guidelines and norms but do not necessarily act in line with them [30].

Furthermore, the scientific community could further investigate the potential social hotspots overlooked both in the S-LCA Guidelines and in the dairy-related papers.

Finally, investigation is also needed to explore the integration of the S-LCA methodology with environmental and economic assessments in the LCSA framework.