1 Introduction

Social life cycle assessment (S-LCA) is a methodology designed to assess social performance and risks (De Luca et al. 2015), evaluating both positive and negative social impacts related to products, services, companies, or organizations (Goedkoop et al. 2020; UNEP et al. 2020). As an element of the life cycle sustainability assessment (LCSA) framework, S-LCA is linked to the social dimension, complementing the environmental and economic pillar, addressed by life cycle assessment (LCA) and life cycle costing (LCC), respectively (Kloepffer 2008; Sala et al. 2015). Among the three methods, which fall under the umbrella of life cycle thinking (De Luca et al. 2015), S-LCA is considered the least advanced (Pollok et al. 2021), with an ISO standard, ISO/AWI 14075 Principles and Framework for S-LCA, still under development (ISO 2023). S-LCA is rooted in the International Labor Organization Fundamental Principles on Rights at Work, established in 1998, including key rights, such as the elimination of child and forced labor (ILO 2022). Inspired by corporate social responsibility, guidelines for S-LCA were initially published in 2009 by the UNEP/SETAC Life Cycle Initiative (UNEP/SETAC et al. 2009; De Luca et al. 2015), now known as the UN Environment Life Cycle Initiative (UNEP et al. 2020), and provided a framework for assessing social impacts throughout the life cycle of products and services (UNEP/SETAC et al. 2009). The Product Social Impact Assessment (PSIA) Handbook, introduced in 2016 and updated in 2020, additionally placed a strong emphasis on methodological practicality and relevance to businesses (Goedkoop et al. 2020). Recent developments, including the 2021 UNEP Methodological Sheets, the 2020 UNEP guidelines for S-LCA of Products and Organizations and their 2022 extension, advanced S-LCA practices (UNEP et al. 2020, 2021; Traverso et al. 2022).

However, the field of S-LCA still faces several unresolved issues. These challenges include the need to identify complementarities between LCA and S-LCA (Sala et al. 2015), as there is an observed overlap in the areas of protection (AoP), which are defined to organize all primary impacts (Gaasbeek and Meijer 2013). LCA focuses on ecosystem health and natural resources (Dewulf et al. 2015; Verones et al. 2017; Taelman et al. 2020) alongside human health. The latter is also partially considered by S-LCA (Taelman et al. 2020), besides social or human well-being (UNEP et al. 2020; Lindkvist and Ekener 2023). Another challenge is the difficulty in using a consistent unit of measurement as endpoint level (Lindkvist and Ekener 2023), such as quality adjusted life years (Sala et al. 2015). Additionally, there are overlaps and divergences between different general guidelines, such as the UNEP guidelines and PSIA handbook, e.g., regarding the terminology concerning affected stakeholders (Mesa Alvarez and Ligthart 2021), which should be considered systematically when performing S-LCA.

Furthermore, there is a strong demand for enhanced integration of S-LCA, especially in the case of circular economy (Zanchi et al. 2018; Mesa Alvarez and Ligthart 2021; Luthin et al. 2023). However, social aspects currently receive limited attention in the context of circular economy, requiring further consideration in research (Luthin et al. 2023). This is particularly relevant in the context of the European Pillar of Social Rights of 2017 (European Commission 2017) and the European Green Deal of 2019 (European Commission 2019), where the EU articulated strategies aimed at fortifying the health and well-being of future generations, emphasizing sustainable resource management (European Commission 2019, 2021). A crucial part of this agenda is the Circular Economy Action Plan (CEAP). With the aim of promoting the necessary transformation outlined by the European Green Deal (European Commission 2020a), the CEAP identifies packaging and plastics as two of the primary six key product value chains (European Commission 2020b). Within Europe, 19.7 Mt of plastic is used in packaging applications, representing 39% of the total plastic demand (Plastic Europe 2022). Notably, certain types of packaging, particularly single-use food plastic packaging, possess limited lifespans, leading to accelerated waste generation (Ncube et al. 2021). In 2020, 61% of the 29.5 Mt plastic waste collected in the EU27+3 countries was attributable to packaging applications (Plastic Europe 2022). Although the European Commission is promoting the “Less Waste, More Value” strategy and thus the circular economy within production processes (European Commission 2020b), moving from a linear to a circular economy is challenging (Neves and Marques 2022).

Flexible packaging plastic waste, consisting of plastic films, bags, flexible food packaging, and other single-use flexible plastics (Ahamed et al. 2021), is one of the most difficult material streams to recycle (Ragaert et al. 2017; Van Belle et al. 2020; Lase et al. 2022). Opposite to rigid plastic recycling, which differs from flexible plastic recycling due to the low bulk density of the films (Horodytska et al. 2018), the following bottlenecks are identified for flexible packaging recycling: the need for enhanced selective collection and sorting processes, infrastructural modifications, advancements in recycling facilities and technologies, and designing packaging with recycling in mind. Only two-thirds of European countries collect flexible packaging with other dry recyclables (CEFLEX 2020), and efficient sorting infrastructure is lacking. Flexible film waste consists of mono- and multilayer structures, with the latter being further subdivided into mono- and multimaterial variants (Horodytska et al. 2018). The lack of identification systems of multimaterial multilayer plastic packaging (MMPP) during sorting (Koinig et al. 2022), combined with their complex multilayer structure comprising various materials in each layer (Tartakowski 2010), is a challenge for the recycling industry (Mulakkal et al. 2021; Soares et al. 2022). In Europe, only 40% of the collected 1.5 Mt of household flexible packaging waste was sent for recycling in 2020 (Plastics Recyclers Europe 2023), considering the high share of multilayer flexible packaging in household waste (Horodytska et al. 2018; Roosen et al. 2022). Due to the variety of materials used and the different processing characteristics of the materials (Tartakowski 2010), MMPP is in practice incinerated and/or sent to landfill (Ragaert et al. 2017; Lase et al. 2023). Hence, this waste stream is not contributing to the European target of 55% plastic packaging recycling by 2030 (European Commission 2015).

Addressing identified bottlenecks is integral to the progress of flexible plastic packaging in the European circular economy. This encompasses the integration of various innovative technologies, including, e.g., tracer-based sorting (Kusch et al. 2021), selective dissolution, delamination (Kaiser et al. 2017; Ügdüler et al. 2022), and deinking and deodorization (Kol et al. 2021), in addition to laminate design including post-consumer recyclates (CEFLEX 2020), as part of this case study. To further drive this transition, comprehensive and prospective sustainability assessments are essential (Soares et al. 2022; Haase et al. 2022). Nevertheless, there is currently no suggested compilation of social impact indicators for evaluating the social risk specifically associated with flexible plastic packaging within the circular economy. While there is a study by Reinales et al. (2020) that addresses S-LCA in the context of plastic packaging in the circular economy, it takes a general approach and does not thoroughly examine the variations in subcategories concerning the social impacts of distinct types of plastic packaging.

Another relevant challenge in the context of S-LCA is the definition of impact categories, subcategories, and inventory indicators to comprehensively assess the social performance and risk of a system, as emphasized by Goedkoop et al. (2020) and UNEP et al. (2020). However, the selection of impact subcategories in S-LCA may appear arbitrary due to a lack of clear and robust selection criteria (Harmens et al. 2022). Although UNEP et al. (2020), the GRI Standard (2021), and the PSIA manuals (Goedkoop et al. 2020) provide insight into materiality assessment, comprehensive guidelines on procedural steps are lacking. To address these issues, Harmens et al. (2022) stress the importance of streamlining the approach for identifying key stakeholders, impact subcategories, and indicators for specific case studies. This can be achieved by better defining the underlying principles that influence the selection of impact subcategories and by harmonizing terminology to avoid confusion among practitioners.

Based on the principles and methodology of S-LCA, this paper is aimed at guiding the selection and inventory of social indicators by defining key impact indicators and data inventory parameters of flexible plastic packaging in the European circular economy. The study includes several key steps, which can be replicated for various industrial sectors. Initially, it involves a comprehensive screening process to identify relevant impact subcategories, indicators, and inventories related to circular flexible plastic packaging, utilizing a systematic literature review. This step is directed by research question (RQ) I: Which S-LCA indicators are used in the literature for the social risk/performance assessment of (food) plastic packaging and its current waste management or recycling? Subsequently, a multi-criteria decision analysis (MCDA) is applied to preselect a range of indicators guided by RQ II: Which S-LCA indicators are most material regarding, achievability, feasibility, easiness to interpret and relevance for assessing flexible (food and non-food) plastic packaging in the circular economy considering a scientific perspective? Participatory methods are then employed to engage stakeholders in the task of prioritizing key impact indicators aligning with RQ III: Which S-LCA indicators are most material regarding, achievability, easiness to interpret and relevance for assessing flexible (food and non-food) plastic packaging in the circular economy when involving stakeholder participation? Finally, a data collection questionnaire is developed to facilitate the subsequent collection of inventory data on the prioritized impact indicators. Ultimately, guidelines for the replication of the procedural steps are developed, based on the applied methodology and findings.

2 Materials and methods

The primary objective of conducting an S-LCA is to assess both the positive and negative social impacts of products and services throughout their life cycle. The ambition is to enhance social conditions and the performance of an organization, product, or service, considering all stakeholders involved (UNEP/SETAC et al. 2009; UNEP et al. 2020). To attain this objective, the methodological framework of S-LCA comprises four essential steps, based on the ISO 14040 (2006) and ISO 14044 (2006) framework for environmental LCA: (i) defining the objective and scope; (ii) life cycle inventory, data collection, and analysis; (iii) conducting the impact assessment; and (iv) interpreting the results. The methodology is iterative in nature, allowing for continuous improvements in the assessment process over time (Goedkoop et al. 2020; UNEP et al. 2020). On the one hand, the UNEP guidelines implement these methodological principles by involving four key elements, namely: (i) stakeholder categories, (ii) impact subcategories, (iii) inventory indicators, and (iv) reference scale or alternatively impact pathway assessment approach to assess the impact (UNEP et al. 2020). On the other hand, the PSIA method expands the methodology with an initial preparation phase consisting of three steps: (i) defining the communication context; (ii) assessing materiality to identify relevant impact subcategories; and (iii) preparing for data collection. Moreover, there is a phase in relation to the circular economy and another phase resulting from the division of the impact assessment phase in two groups: consumers and other stakeholders (Goedkoop et al. 2020; Mesa Alvarez and Ligthart 2021). Additionally, they refer to the aforementioned key elements as (i) stakeholder groups, (ii) social topics, (iii) performance indicators, and (iv) reference scales to assess impact (Goedkoop et al. 2020). The respective terminology associated with both frameworks will be further discussed in the following Subsection 2.1, as it is crucial to comprehend the foundation of S-LCA for the methodological approach employed and described in the subsequent Subsections 2.2–2.7.

2.1 Methodological terminology of S-LCA

Stakeholder categories or groups form the basis of an S-LCA, as social impacts are assessed in the context of different stakeholders, i.e., people potentially affected (directly or indirectly) by the life cycle of products or services (Goedkoop et al. 2020; UNEP et al. 2020). The way and extent of the interaction between the company or its product and its stakeholders determines which stakeholders are included in the scope of the assessment (UNEP et al. 2020). UNEP considers six stakeholders, namely, workers, local community, value chain actors, consumers, society, and children, while PSIA focuses on four stakeholders, i.e., workers, local communities, small-scale entrepreneurs, and users (Mesa Alvarez and Ligthart 2021). According to UNEP et al. (2020), alternative classifications of stakeholders are permitted but need to be documented and justified in a transparent manner. This study follows the UNEP stakeholder categories, as value chain actors are essential interaction partners not included in the stakeholder classification of PSIA.

Another key concept, namely, Social topics or Impact subcategories, is described as social areas that require measurement and assessment of their social impact, risk, or performance including factors such as Child labor, Health and safety, Community engagement, and more. These impacts pertain to and can affect positively or negatively the aforementioned stakeholders along the life cycle of a product or service. They are represented by a cascade of indicators and inventory data (Goedkoop et al. 2020; UNEP et al. 2020). However, the UNEP/SETAC et al. (2009) guidelines also proposed a rational grouping of diverse impact subcategories (e.g., Child labor) into a restricted set of impact categories (e.g., Human rights). Due to the heterogeneous nature of impact categories, this process may involve a potential weighting and/or aggregation step (UNEP/SETAC et al. 2009), as frequently, potential social impacts are represented by multiple (inventory or performance) indicators (Goedkoop et al. 2020; UNEP et al. 2020). These indicators provide precise definitions of the essential inventory data required (UNEP/SETAC et al. 2009) and include quantitative, semi-quantitative, and qualitative units of measurement (UNEP/SETAC et al. 2013). Quantitative indicators can be measured directly in numerical terms (e.g., Occupational accident rate), whereas qualitative indicators assess information descriptively (e.g., Freedom of Peaceful Assembly and Association). Semi-quantitative indicators (e.g., Records on all workers stating names and ages) (UNEP/SETAC et al. 2013) involve categorizing qualitative indicators into a binary (yes/no) format or score them (e.g., ranging from 1 to 3) (Dreyer et al. 2010; Yıldız-Geyhan et al. 2017).

The purpose of the inventory analysis is to gather and examine pertinent data identified during the scope definition process (Jørgensen et al. 2008). There are two primary types of data sources, referred to as primary or secondary data by Goedkoop et al. (2020) in the PSIA manual, whereas the UNEP/SETAC et al. (2009) guidelines categorize them as site-specific data and generic data. The concept remains consistent: primary/site-specific data are specific to the company or product and can be directly obtained from the source, while generic/secondary data are not directly collected from the source (UNEP/SETAC et al. 2009; Goedkoop et al. 2020; UNEP et al. 2020). Primary data collection methods may involve document auditing, conducting interviews, administering questionnaires, and engaging in participatory evaluations. Examples of generic data sources include information from articles in newspapers; NGO reports; other external sources such as Datamaran, MapleCroft, and RepRisk (Harmens et al. 2022); or S-LCA databases such as Social Hotspots Database (SHDB) and Product Social Impact Life Cycle Assessment (PSILCA) database.

During the impact assessment phase, the inventory information is translated into actual impacts (Jørgensen et al. 2008). When assessing social impact, there is currently no universally agreed social life cycle impact assessment (S-LCIA) method available regarding the characterization procedure (Yıldız-Geyhan et al. 2017; Goedkoop et al. 2020; UNEP et al. 2020). The characterization models are not exclusively “mathematical” in nature but can involve consolidating textual or qualitative inventory information or summing up quantitative inventory data as basic aggregation steps. Furthermore, the models may involve the inclusion of additional information, such as performance reference points (PRPs) (UNEP/SETAC et al. 2009). Therefore, two different methodological approaches, reference scale assessment (RS S-LCIA) and impact pathway assessment (IP S-LCIA), can be applied, differing in their use of PRPs or social impact pathways (UNEP et al. 2020; Harmens et al. 2022).

RS S-LCIA relies on assumed causal links (Harmens et al. 2022) and involves the use of a scoring system to evaluate the inventory data, considering PRPs (UNEP/SETAC et al. 2009). In contrast, IP S-LCIA applies measured causal links elaborating the cause and effect chains of impacts (e.g., noise and odor). However, the impact pathways developed in current research often exhibit a lack of completeness as they do not comprehensively cover all relevant social aspects. Furthermore, their application is limited to only a few case studies, with characterization factors mostly being specific to each case. In this context, the RS S-LCIA approach has emerged as the most feasible and widely applicable method (Harmens et al. 2022).

2.2 Outline of research steps

As part of the preparation phase initiated by the PSIA handbook (Goedkoop et al. 2020), this study focuses on five research steps, spanning from (i) conducting a systematic materiality assessment for indicator selection (Steps 1–4) to (ii) laying the foundation to collect inventory data (Step 5). Analyzing materiality is essential for identifying critical factors, associated with all activities, products, services, and partnerships, internally or externally to the organization, that could encompass both opportunities and risks for the company (GRI Standard 2021). The objectives, research procedure, and outcomes for each of the five steps are visualized in Fig. 1. The first step consists of a systematic literature review. Consequently, in the second and third steps, a multi-criteria decision analysis (MCDA) is applied to preselect a variety of indicators in an orderly manner (Van Schoubroeck et al. 2019). In the fourth step, the materiality assessment results are used to prioritize indicators for the data collection of an entry-level assessment of social hotspots. Lastly, the fifth step aims to ensure organized data collection by creating a data collection form tailored to the specifics of the case study. Each of these steps is described in detail in Subsections 2.3–2.7.

Fig. 1
figure 1

The process of researching, selecting, and prioritizing specific social indicators suitable for assessing flexible plastic packaging within the circular economy adapted from Siebert et al. (2018)

Full size image

2.3 Step 1: literature review and sampling procedure in function of social indicator identification

For the literature review, the following methodological procedures were carried out based on the recommendations of Tranfield et al. (2003) and Duriau et al. (2007): (i) planning the review (research questions and search strategy), (ii) conducting the review, and (iii) reporting. During the review planning phase, the RQ I was formulated as follows: Which S-LCA indicators are employed in the literature to assess the social risk/performance of (food) plastic packaging and its current waste management or recycling? The main objective was to gather relevant information from previous studies to gain a comprehensive understanding of the core concepts of applied S-LCA. This involves identifying potential subcategories, indicators, and inventories used in S-LCA methodologies, as well as understanding how these elements interact with each other. This search strategy encompasses systems such as (food) packaging, plastics, recycling systems, and waste management systems (WMS), thereby including products and/or processes, that are part of the system under investigation.

As visualized in Fig. 2, the review initiated with the retrieval of two papers, Costa et al. (2022) and Taelman et al. (2020), which were already known to the research team through the informal approach (personal knowledge). The informal approach using personal contacts and academic networks, sources known to the research team, and “asking around” can substantially increase the yield and efficiency of the overall research efforts (Greenhalgh and Peacock 2005). Subsequently, the review was extended using a hybrid search strategy combining database search and snowballing (Mourão et al. 2020).

Fig. 2
figure 2

Sampling procedure in function of social indicator identification adapted from Fischer et al. (2017); n, number of documents

Full size image

Snowballing has proven to be an attractive alternative or supplement to database searching (Jalali and Wohlin 2012) reducing the risk of missing relevant evidence (Mourão et al. 2020). In the snowballing process, three documents were identified. Subsequently, one document was also later found through the database search. As a result, more emphasis was placed on the database search due to its ability to efficiently yield additional relevant documents.

Scopus and Web of Science were chosen as databases for the database search as they were considered the most efficient digital libraries for the research area (Costa et al. 2022). For conducting a database search, it is essential to determine appropriate search terms and keywords and formulate search strings (Jalali and Wohlin 2012; Badampudi et al. 2015). In this study, the search term S-LCA and LCSA was applied, incorporating the main keywords: (Food) Packaging, Plastics, Recycling, and WMS. Considering the database search, a total of 159 documents were retrieved. The Supplementary Information (Annex 1) provides details on the respective applied search strings and results per search cycle.

In the next phase, both a partial and subsequently an in-depth screening were conducted (Fischer et al. 2017; Naidoo and Gasparatos 2018) to identify relevant articles and content that align with the RQ I.

The partial screening was initially performed to quickly identify potentially relevant articles based on their titles and/or abstracts. Then, an in-depth screening was conducted to thoroughly assess each article’s suitability in addressing impact subcategories and/or indicators. The decision to include or exclude the article was made after reading the abstract followed by other parts of the paper (Wohlin 2014). Following the guidance of Siebert et al. (2018), specific limiting criteria were established to aid in the selection process. The inclusion criteria were as follows.

  1. 1.

    The articles were published in English-language sources until 2010

  2. 2.

    The title and/or abstract suggest that the content of the article focuses on plastic, (food) packaging, waste management, recycling systems, and social indicators. This indicates that articles related to food production and/or food processing are excluded

  3. 3.

    The articles encompass a case study or some form of impact assessment and/or impact subcategories and/or indicators

Additionally, during the screening process, any duplicate entries from the search results were removed. Considering personal knowledge, snowballing, and database search, the initial set of potentially relevant articles (164 documents including duplicates) was narrowed down to 19 relevant documents (11%) that met the selection criteria of the screening process and were useful for further extraction and categorization of relevant information. The selected studies with indication of the journal in which they were published and the study objectives can be found in the Supplementary Information (Annex 2).

The subsequently applied content analysis is a systematic method for condensation of raw data into categories or themes based on valid conclusions and interpretations (Zhang and Wildemuth 2005; Baxter 2020). The respective literature was analyzed, guided by a categorization structure provided by (Taelman et al. 2020), which includes information on (i) impact categories/subcategories, (ii) indicators, (iii) units of measurement, (iv) description of impact categories/subcategories and/or indicators, and (v) reference method/model. In addition, the categories (vi) stakeholder, (vii) type of system studied, and (viii) sustainability dimension/discipline were added. The latter was added due to the overlap of S-LCA with the environmental and economic sustainability dimension (Sala et al. 2015; Taelman et al. 2020). Finally, 33 impact categories, 153 impact subcategories, and 296 indicators were extracted and compiled in tabular form.

2.4 Step 2: MCDA I

In order to address RQ II Which S-LCA indicators are most suitable regarding, achievability, feasibility, easiness to interpret and relevance for assessing flexible (food and non-food) plastic packaging in the circular economy considering a scientific perspective? MCDA I was applied. The purpose of MCDA is to address decision-making problems that require considering multiple perspectives (Van Schoubroeck et al. 2019). Therefore, MCDA I evaluates (i) achievability, (ii) feasibility, (iii) ease of interpretation, and (iv) relevance of each indicator as criteria, adapted from Taelman et al. (2020). A scale from 0 to 3 is linked to a specific reference point for each criterion (Table 1), utilized to justify the reduction of the extensive list of potential indicators from a scientific point of view. For example, under the criterion achievability, the indicator is rated highest (3) when it involves access to supplier-specific data retrieved by the company, while a rating of 0 is associated with no access and/or a too time extensive collection phase.

Table 1 Multi-criteria decision analysis (MCDA) I applied for the preselection of achievable, feasible, easy to interpret, and relevant indicators adapted from Taelman et al. (2020)
Full size table

Consequently, the indicators were rated considering their unit of measurement and their impact subcategory if the indicator alone did not contain enough information to ensure a conscientious rating. If only an impact category/subcategory was given but no indicator, or if the indicator was not linked to a subcategory but only to an impact category, the indicator was not considered to limit the progression of the use of non-uniform terminology. Thus, 287 indicators were rated based on the aforementioned reference points per criteria. In the context of defining relevance, socio-economic relevance at the European level and the severity of social impacts were considered, guided by Beaulieu et al. (2014). Subsequently, the average of the scores of the four criteria categories was calculated by applying equal weighting. This resulted in 56 highest-rated indicators, which were further narrowed down to 19 indicators as described in Subsection 3.1.

2.5 Step 3: stakeholder participation and MCDA II

To answer RQ III (Which S-LCA indicators are most suitable regarding, achievability, easiness to interpret and relevance for assessing flexible (food and non-food) plastic packaging in the circular economy when involving stakeholder participation?), a second MCDA involves the participation of various stakeholders, such as industry partners. It is essential to recognize that the importance of impacts may be perceived differently by various stakeholders. Therefore, in order to enhance the local relevance of S-LCA, participatory methods are beneficial (De Luca et al. 2015). Guided by Foolmaun and Ramjeeawon (2013), the study utilized a convenience sample for the selection of the participants, which falls under the category of non-probability sampling. Convenience sampling offers the benefit of being quick, cost-effective, and simple, as it involves selecting readily accessible individuals as participants for the study. Therefore, the preselected list of indicators was presented to the Circular Foodpack project value chain actors in a workshop funded by the Horizon 2020 (H2020) program (Grant agreement ID: 101003806).

The 18 experts were divided into the following two core sectors based in Europe: (i) representatives from the private sector, including industrial companies from domains such as waste management, laminate and packaging manufacturing, recycling technology production, a private non-profit organization, and a consulting firm; and (ii) members from the scientific sector, encompassing a university and a research institute. The Supplementary Information (Annex 3) provides further information on the organizations participating in the MCDA II.

The workshop was divided into a theoretical part, explaining the methodology of S-LCA and the 19 preselected indicators, thereby providing the participants with a common knowledge base. This was followed by a practical part, in which the experts could express their opinions using Wooclap, an interactive electronic platform used to create polls and questionnaires. Due to the challenge of assessing the criterion feasibility by a non-practitioner audience and in order not to distort the process with confusion among participants hence ensuring a comprehensible workshop setting, the number of MCDA I criteria was limited to three, excluding feasibility, while the reference points were simplified accordingly. Information on the utilized reference points for MCDA II can be found in Supplementary Information (Annex 4).

The survey thereby consisted of 57 questions on the relevance (C1), achievability (C2), and easiness to interpret (C3) of each of the indicators presented. In conformity with MCDA I, the attendees responded by rating each indicator on a scale of 0–3 (e.g., 0 = no relevance, 3 = high relevance) per criteria. Due to time constraints, 15 of the 57 questions could not be answered during the workshop; hence, a link for the second part of the survey was provided by email which was answered by 17 participants. The answers from the on-site and online survey were then combined, and in the first step, an average score per answer was calculated, followed by the average scores per indicator calculated with equal weighting, i.e., C1 (1/3), C2 (1/3), and C3 (1/3). A ranking was created based on the equally weighted average scores. Subsequently, to compare the influence of weighting on the rankings, alternative non-equally weighted average scores per indicator were calculated by giving a higher weight (0.5 instead of 1/3) to one of the individual criteria (C1, C2, and C3), e.g., C1 (0.5), C2 (0.25), and C3 (0.25). Finally, a corresponding ranking was generated based on the three non-equally weighted average scores as detailed in Subsection 3.3.

2.6 Step 4: prioritization of indicators

As emphasized by Van Schoubroeck et al. (2019), it is essential to apply the indicators to a specific case study to validate their applicability. Consequently, this involves establishing the communication context, as outlined in the PSIA methodology by Goedkoop et al. (2020), and defining the goal and scope, as well as the system boundaries (ISO 14040 2006). For this case study, the system boundary includes various processes: (i) collection, state-of-the-art sorting, and tracer-based sorting; (ii) pretreatment including oversorting, shredding, washing, grinding, and float-sink separation; (iii) purification with selective dissolution and/or deinking and delamination; (iv) posttreatment consisting of recompounding and deodorization; (v) laminate production; and (vi) packaging production in Europe. In this context, the objective is an initial entry-level assessment (Pihkola et al. 2022) of the emerging circular flexible plastic packaging value chain to provide feedback to value chain actors. This approach aims to introduce the value chain actors to the topic with a reduced scope of indicators, as resources for data collection are limited. As highlighted by Van Schoubroeck et al. (2019), ranking can be useful to prioritize indicators in the absence of resources such as data, time, and finances. However, there needs to be a rationale for how many indicators are collected in a first assessment that can potentially be expanded in a subsequent one. In the field of cognitive psychology, and more specifically information theory, Miller (1956) suggested that there is an upper limit to our ability to process information developing the idea of the “magical number seven.” Similarly, in the analytic hierarchy process, an approach for decision making in which several selection criteria are structured in a hierarchy (KaurSehra et al. 2012), Saaty and Ozdemir (2003) concluded that the number of elements should be limited to seven plus or minus two. In order to enhance accessibility of S-LCA for the target audience, who are non-experts in the field of S-LCA and considering their limited resources, a threshold was set. This threshold refers to the selection of the nine equally weighted, most relevant, achievable, and easily interpretable indicators for the data collection procedure in the context of the preliminary assessment.

2.7 Step 5: data collection preparation

In the final step, as shown in Fig. 1, a data collection questionnaire was developed based on the nine prioritized indicators. Initially, the primary data source was identified, prioritizing site-specific inventory data of the case study value chain, indicated in the introduction. Subsequently, the essential inventory parameters and unit of measurements for an initial entry-level assessment of the nine priority indicators were defined. Guided by Foolmaun and Ramjeeawon (2013), this process considered the targeted respondents within the circular flexible plastic packaging value chain and is presented in particular to waste managing companies, innovation technology professionals including research institutes and universities, and laminate and packaging producers. In addition, the inventory parameters were defined considering the chosen RS S-LCIA approach. Consequently, PRP categories were set as inventory parameters for certain indicators, e.g., Existence of certified environmental management system. The categories provide information about the quality of the respective parameter linked to an ordinal risk scale. Additionally, some general company-related information is requested in the questionnaire as described in Subsection 3.3.

3 Results and discussion

The subsequent sections present and discuss the outcomes of each research step. This is followed by the formulation of general guidelines derived from the methodological steps applied to the case study on flexible plastic packaging within the European circular economy and step-specific findings. Finally, the novelty of this study compared to existing research is discussed.

3.1 Step 1: results literature review

During the literature review, an inconsistency in terminology was found caused by incoherent assignment of impact categories, subcategories, and indicators; the use of synonymous terms; and the absence of their clear definitions. This data analysis confirms the observation from Harmens et al. (2022), namely, the lack of harmonized terminology. For example, the same indicator is used among different studies to assess different subcategories, as in the case of the indicator Absence of work(ing) accidents (Aparcana and Salhofer 2013; Mahdavi et al. 2022). Mahdavi et al. (2022) assign this indicator to the subcategory Health and Safety, while Aparcana and Salhofer (2013) present it as part of the subcategory Physical working conditions. Synonymity in naming subcategories and indicators further adds to the inconsistency. For instance, Foolmaun and Ramjeeawon (2013) assess the indicator Awareness on health and safety issues, while Yıldız-Geyhan et al. (2017) phrase it as Presence of health and safety awareness. Correspondingly, Foolmaun and Ramjeeawon (2013) use the subcategory Health and safety, while Yıldız-Geyhan et al. (2017) refer to it as Health and safe working conditions. These examples do not claim to be exhaustive, and a more illustrative overview is provided in the Supplementary Information (Annex 5).

The inconsistency in S-LCA terminology poses a challenge in establishing clear and standardized relationships between inventories, indicators, impact subcategories, and impact categories within the S-LCA framework, which ultimately hinders the comparability of results from different studies. The different use of terminology also complicates data analysis. To overcome these challenges, it is necessary to group, cluster, and standardize terms that determine the scope of subcategories, indicators, and inventories that can be considered for assessment. However, it should be noted that full consistency and standardization of terminology across studies can be challenging due to different research approaches and contexts.

3.2 Step 2: results MCDA I

Table 2 represents the outcome of the first MCDA, based on the rating of 287 indicators (including synonyms and duplicates) retrieved from literature review. Initially, 56 indicators received the highest overall rating of 3.00, while 184 indicators received a rating in the range of 2.00–2.75, and 47 indicators received a rating in the range of 0–1.75. The 56 highest-rated indicators were then reviewed in two bilateral discussions consulting five scientific experts in the field of S-LCA. By excluding duplicates and synonyms, reviewing and re-evaluation of the original scores, and prioritizing the most achievable indicators per impact subcategory, 19 indicators were preselected as a result of the peer discussions. This number was considered proportionate given that the original 19 reviewed papers assessed an average of 20 indicators. The 19 indicators are assigned to 13 impact subcategories, with End-of-Life (EoL) Responsibility, Equal Opportunities/Discrimination, Health and Safety, Local Employment, and Safe and Healthy Living Conditions represented by more than one indicator. Five stakeholder groups (i.e., consumer, worker, local community, society, and value chain actors) were covered, therefore, addressing almost all stakeholder categories proposed by UNEP et al. (2020), with the exception of children. However, it should be noted that children are included in the workers’ stakeholder group, i.e., through child labor. Subsequently, it was determined to align with the UNEP subcategory terminology, which was widely applied, except in relation to the indicator Existence of certified environmental management system (Reinales et al. 2020) and Extent of job creation potential (Pillain et al. 2019). Therefore, the respective subcategories have been adapted. A description of each of the indicators is illustrated in the Supplementary Information (Annex 5).

Table 2 Preselected social impact indicators per impact subcategory and their associated stakeholder group. Explanation of the abbreviations of certain units: yes/no (y/n); hours per year (h/y); number of full-time equivalents (#FTE)
Full size table

3.3 Step 3: results MCDA II

Table 3 shows the materiality assessment results as an outcome of the second MCDA including the rank positions of the equally and non-equally weighted scores, the latter for C1, C2, and C3. It can be observed that the weighting does not fundamentally influence the ranking in most cases. The first three ranks show consistency in both the equally and non-equally weighted rankings, except for weighting of C1. These positions are attributed to Existence of record of proof of age, Existence of certified environmental management system, and Percentage of workers who are paid a living wage or above. Except for weighting of C1, where Existence of health risk assessments regarding toxicity is placed in the third position, highlighting its relevance, and thereby moving Percentage of workers who are paid a living wage or above to the fourth position. In the middle ranks, there are slight variations among the rankings, with Percentage of male/female employees dropping to the 12th place in the C1 category, compared to positions of 4–8 in the respective other rankings. This suggests that this indicator is perceived comparatively less relevant. Even in the lower ranks, there is little variation between the rankings, where, for example, Existence of corporate social responsibility reporting remains consistently at rank 13, and Existence of clear information about EoL options on the packaging, Percentage of suppliers from countries with high estimated proportion of modern slavery, and Percentage of local suppliers show variation in rank only for weighting of C1. It can also be noted that the first 10 positions of equal weighting and the weighting of C2 only differ in rank position. Further information on the non-equally weighted scores of the 19 preselected indicators is provided in the Supplementary Information (Annex 6).

Table 3 Materiality assessment results based on equally and non-equally weighted ranks of the preselected 19 social indicators
Full size table

The ranking scores generally aligned with the researcher’s expectations. However, it is noteworthy that the indicator Percentage of suppliers from countries with high estimated levels of modern slavery is ranked 14th with an equally weighted score of 1.71, showing that it was considered relatively less crucial. This could be attributed to the tendency of value chain actors to overlook its importance due to the lack of transparency along the value chain. In the future, legal regulations can counteract this unawareness, much like the Supply Chain Act that has already been enacted in Germany (Lieferkettensorgfaltspflichtengesetz – LkSG. 2021).

In general, the ranking allows for the selection of specific indicators and corresponding criteria to prioritize, depending on the preferences and priorities of different stakeholders. Furthermore, a ranked set of indicators is useful for advancing the development of a potential weighting method. Nevertheless, it is valuable to explore a more extensive ranking process by involving a wider panel of experts and employing various ranking methodologies, as shown by Van Schoubroeck et al. (2019).

3.4 Step 4: prioritized indicators

As indicated in Subsection 2.6, Table 4 illustrates the nine priority indicators, their associated stakeholder groups, impact subcategories, and units of measurement considered for an initial entry-level assessment. Each impact subcategory is assessed using one or more indicators. However, it is evident that the stakeholder group of workers is comparatively overrepresented by means of various impact subcategories. Specifically, within the subcategory Health and Safety, three indicators contribute to this subcategory compared to only one indicator in the respective other subcategories. This indicates that workers are perceived particularly vulnerable to social risks, which can also be observed in the literature by Azimi et al. (2020) and Yıldız-Geyhan et al. (2017) but also raises the question of the necessity of indicator weighting within a subcategory. Moreover, implementing the threshold of nine indicators leads to a decrease in the number of covered stakeholders, excluding value chain actors.

Table 4 Prioritization of nine indicators based on the equal-weighted ranking applied in MCDA II
Full size table

Yıldız-Geyhan et al. (2017) emphasize that the quality of the indicators, rather than their quantity, is essential for a reliable S-LCA result. Moreover, the literature demonstrates cases where fewer than nine indicators have been employed to assess the social risk or performance of the system under study (Albrecht et al. 2013; Mirdar and Mansour 2022). Nevertheless, when setting a threshold for prioritization, the practitioner should be aware of the risk of incompleteness and reduced comprehensiveness of the assessment. To counteract these, it is more important than ever to point out the limitations of the study and to ensure a transparent presentation of the results, as highlighted by Van Schoubroeck et al. (2019).

3.5 Step 5: data requirements

Considering the communication context in this study, the S-LCA focuses on feedback to and from the value chain actors. The objective is to evaluate social risks and performance through a hotspot analysis, offering insights to decision-makers on enhancing ongoing developmental processes. Regarding the structure of the data collection questionnaire, which is presented in the Supplementary Information (Annex 7), general company-related data (i.e., total number of employees) and data inventory parameters per indicators are requested. The total number of employees is essential in relation to two indicators, namely, Percentage of male/female employees and Percentage of workers who are paid a living wage or above. The latter is assessed based on employee numbers per living wage categories, categorized into A, B, C, and D for different family and household types, calculated with upper and lower limits per country (Guzi 2021). Meanwhile, the indicator Existence of certified environmental management system (EMS) assesses the existence and quality of EMS divided into four categories, including Eco Management and Audit Scheme (EMAS) and ISO 14001 certifications. The categories function as PRPs and enable the conversion of, e.g., dichotomous yes/no answers into polytomous answer options assessable on an ordinal scale. Conversely, Percentage of male/female employees, Extent of safety training, and Percentage of (fatal and non-fatal) accidents/injuries in the organization are evaluated by numerical values, the latter in relation to 1000 employees in the company/y. Existence of research and development (R&D) and Existence of record of proof of age are initially assessed by a yes/no response followed by a numerical value. The values are subsequently linked to certain PRPs during impact assessment. Lastly, Existence of health risk assessments regarding toxicity and Existence of extended producer responsibility (EPR) policy are assessed solely with yes/no responses as categorization is impractical.

To complete the assessment and compensate for any weaknesses in the ranking, it can be discussed to complement the data collection with general data for specific segments of the value chain. Background data from PSILCA could be included for assessing Percentage of suppliers from countries with high estimated levels of modern slavery. Thus, the foreground value chain is assessed based on the primary data collected for the foreground system, while the background system, e.g., the upstream value chain, is assessed based on the “frequency of forced labor using cases of forced labor per 1,000 inhabitants in the region” as a unit of measurement (Maister et al. 2020). Since using an existing database reduces the effort of data collection, this can be considered as an approach to expand the number of indicators if necessary and address the recommendation of Ciroth and Franze (2011) to mix various forms of data to ensure a reliable study.

According to Yıldız-Geyhan et al. (2017), the most time-consuming and demanding part of social assessments is the inventory analysis, especially data collection. When sharing the data collection questionnaire with the respective value chain actors, several bottlenecks and opportunities for improvement could be identified. First, data collection poses greater challenges for multinational corporations due to the complexity of managing multiple outputs. Furthermore, the availability of country-specific disaggregated inventory data may be limited. Hence, maintaining a coherent functional unit (e.g., treatment of 1 ton PE-rich sorted flexible packaging waste resulting in flexible plastic packaging) and ensuring compliance with the system boundary (as described in Subsection 2.6), particularly in terms of the geographical scope, can be demanding. In such cases, a company perspective is recommended as an alternative to a product-specific approach. Moreover, data on salaries are not accessible in some cases as they tend not to be released through head office and the works council. Meanwhile, it is difficult to assess the potential social risks or benefits of new recycling sectors, given that the processes are not yet operating on industrial scale (Pillain et al. 2019). Additionally, due to the involvement of research institutes and universities, which are not companies selling products, the human capital requirements for operating this potential sector are not yet properly defined and will change as the sector matures (Pillain et al. 2019). Consequently, combining innovative processes with existing industrial waste management is a practical and pragmatic approach. In this way, an actual production line can be simulated that combines all processes. Finally, implementing a prospective aspect would require an expert estimation of at least the indicators Percentage of workers who are paid a living wage or above and Extent of safety training, as these two indicators are particularly dependent on the implementation of advanced technologies, according to waste management and innovative recycling technology experts.

3.6 Guidelines on the procedural steps for social indicator selection and inventory

While the indicators selected and prioritized are tailored to the European case of circular flexible plastic packaging, the employed methodological steps possess a generic nature as they are not limited to a single case study and integrate various methodological approaches (as outlined in Subsections 2.2–2.7). Simultaneously, their applicability was validated through their implementation in the case study. Consequently, Fig. 3 provides an 11-step guideline, which S-LCA practitioners across various industry sectors can follow to establish a prioritized set of indicators and inventory. The first step is the definition of the value chain and system boundary to gain a comprehensive understanding of the assessment context. The context of the system under study forms the basis for collecting information on stakeholders, impact subcategories, indicators, and units of measurement, sourced from scientific literature in databases and/or personal networks. The data is then compiled into a tabular format in the third step. In the fourth step, MCDA I is applied for the preselection of indicators guided by four criteria, namely, achievability, relevance, feasibility, and easiness to interpret provided by Taelman et al. (2020). This step should be conducted in collaboration with specialists in the field of S-LCA to achieve consensus. The timeframe for the subsequent stakeholder ratings should also be considered, as the number of indicators that can be rated in an on-site workshop is approximately 15 indicators per hour, according to the researchers’ experience. Individual ratings per participant may result in a higher number of indicators being rated per hour.

Fig. 3
figure 3

Guidelines on the eleven procedural steps for the selection and inventory of social indicators

Full size image

In the seventh step, the process involves the selection of stakeholders with different expertise, e.g., from different departments of the company such as sales and marketing, R&D, sustainability, design, and engineering. Subsequently, MCDA II is applied to rate the indicators focusing on three criteria, namely, achievability, relevance, and easiness to interpret adopted from Taelman et al. (2020). The challenge of assessing feasibility by a non-practitioner audience necessitates a reduction in criteria to maintain the integrity of the rating process. In the ninth step, the most material indicators are selected initially through either equally or non-equally weighted ranking. The non-equally weighted ranking assigns specific weight to each MCDA II criterion, allowing for the prioritization of indicators from a stakeholder perspective, such as choosing the most achievable indicators. The final indicator prioritization is based on a three-level approach including entry, intermediate, and advanced levels (Pihkola et al. 2022). The entry level involves selecting five to nine of the most material indicators, while the intermediate and advanced levels encompass more than nine indicators. The intermediate level includes additional compensatory indicators to balance any potential weaknesses identified in the stakeholder ratings. The eleventh and final step comprises the preparation of the data collection questionnaire considering two crucial aspects: the source of data, either primary company-specific data and/or secondary data from databases, and the chosen assessment method, based on either PRPs or cause-and-effect chains.

3.7 Limitations of the study

Considering the ranking of indicators (MCDA I and MCDA II), it is crucial to acknowledge that decision-making can be influenced by various factors, as noted by Feng et al. (2022), including attention, memory, thinking, emotion, and sentiment. Several potential limitations were identified in relation to MCDA II and these factors, such as distractions during the workshop, technical issues like multiple login attempts to the interactive electronic platform, and potentially reduced focus due to the workshop's timing towards the end of the event. Even though the indicators were introduced in advance, it is not guaranteed that all participants have fully memorized them. Additionally, participant engagement varied, despite encouragement for questions and clarification provided during the workshop. This variation could be attributed to the limited time frame and the requirement for swift decision-making. Additionally, setting a threshold for prioritization of indicators could risk incomplete assessments, highlighting the importance of transparently communicating study limitations, as noted by Van Schoubroeck et al. (2019).

3.8 Further discussion

To distinguish the novelty of this study from existing contributions, the following chapter provides a comparison with related studies, within or outside the field of plastic packaging. In the case of indicator selection for wood-based products by Siebert et al. (2018) and for household food by Li et al. (2023), both approaches involve four distinct steps, which deviate from the proposed methodology in this study in terms of scope, preselection procedure, criteria for the final selection, and the incorporation of harmonized terminology. Unlike the current study, which utilizes MCDA guided by Van Schoubroeck et al. (2019), Siebert et al. (2018) applied semi-structured stakeholder interviews, while Li et al. (2023) used an expert opinion questionnaire. Furthermore, Siebert et al. (2018) do not include the selection of prioritized indicators, while Li et al. (2023) lack explicit exclusion principles.

Conversely, Reinales et al. (2020) applied a stakeholder ranking approach to assess the materiality of subcategories for plastic packaging, however, focusing exclusively on assessing relevance. Their intention was to complement the identification of relevant subcategories, which corresponds to the second step of the guide (Fig. 3), without the aim of prioritizing a smaller group of indicators. Similarly, Van Schoubroeck et al. (2019) rated indicators for biobased chemicals but did not select a set of prioritized indicators, although emphasizing the importance of ranking in the context of prioritization.

In this study, as shown in Fig. 3, a preselection process is integrated as the fourth to sixth step focusing on four distinct selection criteria including relevance. Moreover, a strong emphasis is placed on harmonized terminology and the prioritization of indicators in the ninth and tenth steps. In addition, the procedure was extended by an eleventh step, outlining how and which data inventory to collect. These procedural steps were applied in a case study on flexible plastic packaging but can be adapted to various industries, including further validation and development.

4 Conclusions and perspectives

Providing a guideline on uniform procedural steps holds importance for practitioners involved in S-LCA across diverse industry sectors, as it contributes to establishing a standardized approach for prioritizing indicators and inventory. These procedural steps were formulated using a case study on flexible plastic packaging within the European circular economy to illustrate their practicality. A combination of a literature review and two cycles of MCDA resulted in a materiality ranking of 19 social indicators specifically tailored to circular flexible plastic packaging. This ranking was performed in collaboration with stakeholders from both the private and academic sectors which increases the local relevance of the indicators.

For the prioritization of indicators, a threshold was established to guide the selection for an entry-level assessment, aiming to ensure the accessibility of S-LCA for non-experts with limited resources. This threshold refers to the selection of the nine, most relevant, achievable, and easily interpretable indicators including: Existence of record of proof of age, Percentage of workers who are paid a living wage or above, Existence of certified environmental management system, Existence of research and development department, Percentage of accident/injury in the organization, Extent of safety training, Existence of health risk assessments regarding toxicity, Percentage of male/female employees, and Existence of extended producer responsibility (EPR) policy. These nine prioritized indicators serve as the basis for the collection of inventory data, which can subsequently be utilized for an entry-level S-LCA while allowing for potential expansion in subsequent assessments. In future research, these procedural steps can be utilized to prioritize social indicators and inventory in various industrial sectors.

Prioritizing indicators is essential for entry-level and prospective assessments, particularly when resources such as time, data, or expertise of the target audience are scarce. Utilizing primary inventory data enables the assessment of the foreground system and its direct impacts on the stakeholders, producing robust results that promote accountability and corporate social responsibility. However, it is crucial to emphasize transparency and acknowledge the associated limitations when communicating the results, especially when narrowing the scope. While a ranked set of indicators is practical for further development of a potential weighting approach, a more comprehensive ranking procedure involving a broader expert panel and different ranking methods should be considered. Additionally, these ranked indicator sets can serve as a basis for establishing category rules for a social product footprint, complementing environmental and economical assessments for a holistic evaluation of flexible plastic packaging. In future research, these indicators can be applied to assess the social performance of various case studies within the value chain of flexible plastic packaging in both food and non-food applications. These assessments can be compared to baseline scenarios, such as the current European waste management system or alternative packaging solutions. Expanding the assessment to cover all 19 indicators and incorporating secondary data for evaluating the background system using S-LCA databases like SHDB and PSILCA can further enhance the comprehensiveness of the assessment.

Synonymity and inconsistency in S-LCA terminology and challenges in data collection due to inaccessible inventory data are recognized as two main obstacles in streamlining S-LCA. Moreover, future research should place a particular focus on establishing methodological guidelines for the social assessment of emerging technologies and address the challenges associated with collecting data for prospective assessments, as such estimates are inherently subject to uncertainty.

Finally, assessing the sustainability of current and emerging technologies remains an essential element for informed decision-making as we progress in the transition towards a more sustainable, socially just, and equitable circular economy.