Employee skills for circular business model implementation A taxonomy

.

One driver of socio-technical transitions is a shift in skills, as some skills relevant in the old system lose their value in a new system (Geels, 2002). Implementation of CE strategies requires businesses to extend their activities along the product life-cycle and integrate more deeply with business partners and the broader ecosystem in order to enable circular innovation (Bocken et al., 2016). While scholarship maintains that firms should develop new employee skills and organizational capabilities to implement CBM, research has only recently begun to examine such skills and organizational capabilities in detail. Some studies have focused on employee skills related to specific roles and circular skills (De los Rios and Charnley, 2017;Sumter et al., 2021), but only two have focused on more comprehensive overviews of general, sustainable, and circular skills for CBM (Ganiyu et al., 2020;Janssens et al., 2021). A modest body of research about specific organizational capabilities for CBM has emerged, focusing on environmental management (Scarpellini et al., 2020b), big data and business analytics (Kristoffersen et al., 2021), and integrative perspectives regarding these capabilities have also emerged, focusing on the intersection of skills, strategies, resources, and processes (Khan et al., 2020;Prieto-Sandoval et al., 2019;Santa-Maria et al., 2022).
This article seeks to contribute towards "our understanding on the implementation of the circular economy" (the aim of a special issue of the Journal of Cleaner Production 2 ) by offering a systematic examination of skills relevant to the implementation of novel CBM in start-ups. Besides incumbents innovating towards CBM (Santa-Maria et al., 2022), the uptake of circular start-ups 3 is driving the CE transition. Yet, start-ups have been largely overlooked by CE literature on skills and capabilities (more detail in Section 2.2). Skills 4 of individuals enable and enhance the processes, structures, and technologies needed for creating, deploying, protecting, and reconfiguring organizational capabilities required for CBM activities. As such, skills are a micro-foundation of firm-level capabilities 5 -both ordinary and dynamic (we define the terms 'capabilities' and 'skills' in the section 'Theoretical framing'). The relevance of individual employment skills to CBM implementation has yet to be fully researched or integrated into practice. The modest volume of literature offers some insights, but no published academic study, at the time of drafting this article, has developed a comprehensive skill taxonomy for CBM implementation in start-ups. To fill this gap, this study's research question is: What skills should be included in a comprehensive skill taxonomy for CBM implementation in start-ups?
To answer this question, this study analyzes self-reported skill profiles of 2407 staff working in circular start-ups and presents a novel skill taxonomy. Companies developing skill taxonomies often aim for a comprehensive set of high-level skills for successful (in this case, circular) business model implementationno general, sustainable, or circular skill would be missing, in concept. Accordingly, this study determined that a comprehensive skill taxonomy would be a helpful analytical device. The study focuses on skills for three reasons. First, the unit of analysis lends itself better to a targeted operationalization than does the sometimes ambiguous concept of 'capabilities' (Lankhorst and van Dijk, 2021). Second, the focus on skills enables a deeper connection with emerging research about skills as a principal microfoundation of capabilities, as even a macro-perspective on capabilities must consider constituent elements (Felin et al., 2015). Finally, skills are often a logical starting point for start-up entrepreneurs, as at smaller organizational scales individual skills are more immediately relevant and tangible than are capabilities.
The study finds that CBM implementation requires a set of general, sustainable, and circular skills; some of these skills, such as digital skills, have been neglected. Skills declared as specifically circular are not as prevalent in circular start-ups as the literature suggests. Given that CBM is not an entirely new concept, some skills identified in this study predate the CE concept. Thus, the novelty of skills for CBM implementation is apparent when the organizational context itself shifts; in circular startups, employees might apply existing skills to novel or differentiated circular ideas and need to develop an understanding and recognition of those skills in their circular context. Consequently, promoting circular narratives as a framing device for skill development can advance CE towards mainstream uptake. While the taxonomy aims to be an analytical device for both scholars and practitioners, it is not proposed as conclusive but rather as a prompt for further research.
The remainder of this article is structured as follows. Section 2 outlines the theoretical framing of the study, situating it within the literature on capabilities and skills. Section 3 describes methods, Section 4 presents the results and taxonomy, and Section 5 discusses the relevance of the taxonomy for scholarship and practice.

Capabilities and skills in organizational management
This study is theoretically situated within literature on organizational capabilities, focusing on a sub-strand addressing employee skills. Capability theory, having evolved over several decades, is expressed through various perspectives including the resource-based view and knowledge-based view (Felin and Hesterly, 2007;Helfat and Peteraf, 2003;Hoopes and Madsen, 2008;Langlois and Foss, 1997). Winter (2003) defines organizational capability "a high-level routine (or collection of routines) that, together with its implementing input flows, confers upon an organization's management a set of decision options for producing significant outputs of a particular type." Capabilities can also be described as firm-specific organizational knowledge (Dosi et al., 2000;Langlois and Foss, 1997) or competences (Teece et al., 1997) that enable an organization to perform activities and improve business performance (Helfat and Peteraf, 2003;Hoopes and Madsen, 2008).
Two types of capabilities are commonly discussed in the literature (Drnevich and Kriauciunas, 2011;Teece, 2014Teece, , 2018Winter, 2003). First, operational (or ordinary) capabilities enable a firm to maintain operations in the short-term (Winter, 2003) and ensure business efficiency. Second, dynamic capabilities are higher-order abilities to respond to opportunities and threats, to reconfigure business operations accordingly, and to maintain a sustainable competitive advantage (Teece et al., 1997). Both types of capabilities are essential and interdependent (Kraaijenbrink et al., 2009).
Capabilities are supported by microfoundations that include individuals and their skills, processes, technology, and structureas well as interactions among all four (Abell et al., 2008;Felin et al., 2012Felin et al., , 2015. While microfoundations are not limited in perspective to individuals (Barney and Felin, 2013;Dosi et al., 2008), research on microfoundations pays considerable attention to individualsa starting point for understanding organizational behavior and performance (Barney and Felin, 2013;Campbell et al., 2012;Felin and Foss, 2005;Felin and Hesterly, 2007). One example is employee mobility: 2 https://www.journals.elsevier.com/journal-of-cleaner-production/fort hcoming-special-issues/who-will-benefit-from-the-transition-to-the-circular-e conomy (accessed 28 Sept 2022) 3 This study adopts the definition of a circular start-up proposed by Henry et al. (2020, p. 2): "new, independent and active companies pursuing a [circular business model]." 4 Scholars also use the term 'competency' (e.g., Sumter et al., 2021). We view both terms (skill and competency) as synonymous: successfully performing a task on the individual level. For consistency, we use the term 'skills' throughout the article. 5 Some studies use the term 'skills' when referring to organizational skills as capabilities. For the remainder of this article, mentions of 'skills' refer to individual skills of employees, unless otherwise specified.
businesses can build new capabilities by recruiting individuals with particular skills (Felin et al., 2012;Felin and Hesterly, 2007). Fig. 1 depicts our conceptualization of capabilities and microfoundations. The 'collectivity' or aggregation of individuals' skills within an organization can be considered the 'skills of the organization,' and a key management function is to coordinate and utilize the collectivity of these skills effectively (Dosi et al., 2000). As outlined in Fig. 1, organizational capabilities or competences are not simply the sum of individuals and their skills but an aggregation of all microfoundations and their interactions (Barney and Felin, 2013). According to Dosi et al. (2008), it is appropriate to " [bear] in mind that the 'competence of company x in technology y' is something different from 'the ensemble of the individual skills in technology y of all the members of company x.'" Relatedly, synergies created by certain combinations of individual skills, processes, organizational structures, and technologies can generate firm-specific knowledge and capabilities attributable to the organization as a whole.
Building on the above line of reasoning, this study considers skills held collectively by a company (rather than by a single individual) as the starting point for understanding the adoption of processes, structures and technologies for creating, deploying, protecting, and reconfiguring ordinary and dynamic capabilities that enable an organization to perform business model activities (see Fig. 1). For example, microfoundations undergirding project management capabilities constitute not only supporting processes, structures, and technologies but also individual-level skills to develop and enhance these project management microfoundations altogether. There is no automated mechanism that translates the existence of certain skills into a competitive advantage, rather skills existing in a company need to be utilized (Dosi et al., 2000). The skills of an organization can thus be considered a necessary but not wholly sufficient condition for business performance.

Capabilities and skills for CBM
CBM can be conceptualized as a subset of sustainable business models (Geissdoerfer et al., 2017; and is a way of operating a business that incorporates CE principles and strategies for slowing, narrowing, or closing resource loops Pieroni et al., 2019;Santa-Maria et al., 2021). Implementing CE strategies often requires businesses to extend activities beyond those of traditional business models (Bocken et al., 2016). These activities include, depending on the CE strategies, reverse supply chain and logistics activities, higher degrees of collaboration along the value chain, service design activities for product-service systems (Urbinati et al., 2017), and product design, manufacturing, and other enabling activities . CBM innovations occur either by creating a completely new model (e.g., as a start-up) or by reconfiguring elements of an existing one .
Accounting for these exigencies, scholarship has argued that firms should develop new employee skills and organizational capabilities; research on both is relatively new but growing within CE literature. This study reviewed articles on individual skills and organizational capabilities for CBM (more detail in the 'Methods' section) and summarized both in a comprehensive overview (Appendix A). CE literature most often focuses on organizational capabilities (Elf et al., 2022;Fernandez de Arroyabe et al., 2021;Khan et al., 2020;Kusumowardani et al., 2022;Marín-Vinuesa et al., 2021;Marrucci et al., 2022;Prieto-Sandoval et al., 2019;Santa-Maria et al., 2022;Scarpellini et al., 2020a,b;Stekelorum et al., 2021). Examples of such capabilities are big data analytics, customer service, (green) marketing, environmental management systems, reverse logistics, and circular patenting. Other research addresses individual skills for CBM implementation (Janssens et al., 2021;Sumter et al., 2020Sumter et al., , 2021, including data analytics, material analysis, problem solving, ethical and social principles, circular user engagement, and circular material use (in design).
Scholars have adopted various conceptual approaches and levels of aggregation in studying skills and capabilities for CBM, with some definitional overlaps. For example, Prieto-Sandoval et al. (2019) consider 'research and development' a dynamic capability, while Burger et al. (2019) consider 'science' an individual-level skill. Also, studies of organizational capability have proposed (operational) capabilities, dynamic capabilities, and aggregate microfoundations (avoiding differentiation of individual microfoundations outlined in Fig. 1; see Appendix A), while referring to the same ideas. For example, stakeholder collaboration has been proposed as an operational capability, a dynamic capability, and an aggregate microfoundation. Given these ambiguities, Wang and Ahmed (2007, p. 33) state that "a significant number of empirical studies pertinent to dynamic capabilities do not explicate the concept[s]." Further, Felin & Foss (2005) argue that capability research has faced problems of empirical operationalization, given its vague conceptual origins.
From this study's review of articles about skills and organizational capabilities for CBM, two additional findings deserve mention. First, skills and capabilities for CBM are often categorized as general, sustainable, or circular (Table 1 and Appendix A). Second, research either fails to distinguish CBM innovation between start-ups and incumbents or focuses only on incumbents (e.g., dynamic capabilities or aggregate microfoundations for CBM innovation within existing businesses; Santa- Maria et al., 2022). To the knowledge of the authors, no study of skills or capabilities has focused on start-ups. This trend has been observed in general CBM research  and is notable since start-ups are often considered to be drivers of innovation given their lack of organizational path dependencies . Start-ups in this context are typically small-scale companies that base their entire business models around CE principles; by contrast, larger incumbents  Extending the definition of general skill type, this study defines circular skills as skills that specifically address aspects of "cycling, extending, intensifying, and/or dematerialising material and energy loops to reduce the resource inputs into and the waste and emission leakage out of an organisational system" . often diversify rather than transform towards CBM entirely . As such, skills and capabilities in start-ups are often specific to a CBM, whereas larger incumbents also include (legacy) capabilities and skills serving traditional business models. This study develops a comprehensive overview of skills (as a microfoundation of capabilities) needed to implement CBM in start-ups (Fig. 2) by investigating existing skills in start-ups 6 that have adopted a CBM (see 'Methods' section).

Skill taxonomies
This study introduces a skill taxonomy at the organizational level. While not addressed extensively in the literature, skill taxonomies have received attention in practitioner venues including human resources blogs (AG5, 2021;AIHR, 2021;Creelman, 2021). A skill taxonomy can be defined as "a structured list of skills defined at the organization level" (AIHR, 2021). Such a taxonomy can support a unified language that informs human resources decisions and ultimately drives business performance (AIHR, 2021). Developed under the sponsorship of the Employment and Training Administration at the United States Department of Labor, the O*NET database provides the most commonly used generic skill taxonomy (Creelman, 2021;O*NET, 2022).
Existing taxonomies organize skills into various categories. For example, Burger et al. (2019) distinguish six groups of skills based on O*NET (2022): basic skills, complex problem-solving skills, resource management skills, social skills, systems skills, and technical skills. Janssens et al. (2021) distinguish three groups of skills: transversal skills, valorization skills, and technical skills. Likewise, Kirchherr et al. (2018a) distinguish three groups of skills: technological skills, basic digital skills, and classic skills. The categorization by Burger et al. (2019) forms the conceptual basis of this study's taxonomy. Additionally, there is no fixed number of skills appropriate for a taxonomy. Sumter et al. (2021) propose nine, Janssens et al. (2021) 37, andBurger et al. (2019) 35 skills. Thus, this study aimed to develop a comprehensive taxonomy encompassing 20 to 40 skills to provide a more practical and higher-level overview rather than a long list of detailed skills (this study identified roughly 700 self-declared skills; see Section 3.1).

Methods
The proposed skill taxonomy is based on an analysis that includes clustering and synthesis (Fig. 3) an approach commonly used in the social sciences (Ahlquist and Breunig, 2012;Fonseca, 2013). This is the first study in the CE literature on this topic to adopt a large-N clustering analysis (including natural language processing) that identifies skills for CBM implementation. Most existing research on this topic is qualitative (Khan et al., 2020;Sumter et al., 2020), with some quantitative exceptions or extensions (Burger et al., 2019;Janssens et al., 2021). The latter focus on the relevance of pre-imposed skills, whereas this study's analysis applies unsupervised learning to identify skills.

Skills clustering
Skills data from employees were collected and clustered into sets. Borrowing from Bastian et al. (2014), Russell and Klassen (2019), and Bothmer and Schlippe (2022), this approach consisted of three steps: data-scraping of LinkedIn profiles, natural language processing, and skills clustering. The anonymized data 7 contained skills from LinkedIn profiles of staff employed in 113 circular start-ups ( Fig. 4; Appendix B includes a full list). Data analyzed were taken from the LinkedIn profile section labeled 'Skills & endorsement.' 2407 publicly available staff profiles were examined, with a total of 4830 self-declared skills. The list of circular start-ups was taken from , whose work may be considered the most exhaustive such effort.
It is prudent to note that skills self-declared on LinkedIn profiles are prone to subjectivity. There is a risk that employees might falsely report skills. Additionally, employees might declare skills across differing levels of granularity or abstraction. For example, one employee might declare programing tools (such as C or C#) while another might declare 'application development.' Fig. 5 presents a word cloud of the top-50 most frequently occurring skills (Appendix B presents skill frequencies).
Using the natural language processing technique 'Word2Vec' (Church, 2017;Mikolov et al., 2013), the researchers created context-based word embeddings of the scraped skills data serving as input for clustering. From the LinkedIn data, researchers first generated a comprehensive list of 715 self-declared skills and their frequency (i.e., how often each skill was declared across all start-up employees in the sample data; see Supplementary Materials); long-tail skills with a total frequency below five occurrences were excluded. Based on the researchers' experience, low-frequency skills on LinkedIn are not useful because they are either miss-spellings or not industry-standard. The researchers then generated mathematical vector representations (so-called 'word embeddings') with 100 dimensions using a Word2Vec model based on co-occurrence for every skill in the list. 8 Word embeddings can capture the context of words in a text (such as semantic similarity or co-occurrence) through mathematical representations of these words (in this case, 'skills'; Karani, 2018). The model was trained on co-occurrence of skills in LinkedIn profiles. For example, 'Python' or 'R' might be frequently mentioned along with 'SQL,' so vector representations (or word embeddings) of these skills would be mathematically closer.
Finally, the researchers clustered skills by applying an unsupervised learning algorithm (Hastie et al., 2009) on word embeddings, resulting in 50 clusters. The researchers applied k-means clustering (Hartigan and Wong, 1979) on word embeddings, one of the most common methods in such circumstances (Hastie et al., 2009;Pham et al., 2005). The k-means method requires users to assume and predefine a fixed number of k-clusters as an input parameter before clustering is performed. Determining the value of k is complex (Steinley, 2006) and the model's quality of fit with the data, based on the number of clusters, is a subjective decision (Feldman and Sanger, 2007;Pham et al., 2005). A trial-and-error approach (Pham et al., 2005) was used and clustering was run with 20, 30, and 50 predefined clusters. The researchers ultimately established the number of clusters at 50, perceiving that the majority of clusters contained clearly defined sets of skills. It was thus possible to aggregate clusters again in the subsequent step (skills synthesis) with the aim of 20-40 skills in the taxonomy.
The k-means clustering segmented the 715 self-declared employee skills into 50 distinct and non-overlapping groups based on their cooccurrence in LinkedIn profiles. An example of this effort is represented in Table 2, which presents a skill clustering frequency table showing skills assigned to one of the 50 clusters. These finance and accounting skills, which have been declared by employees with the indicated frequencies, typically co-occur in LinkedIn profiles. The resulting clusters contain 14 to 15 self-declared skills on average, ranging from three skills in the cluster with the least skills assigned to 30 skills in the cluster with the most skills assigned. Roughly 30 clusters contain between 10 and 20 self-declared skills. Full clustering results can be found in the Supplementary Materials. 6 For the remainder of this article, mentions of employees refer to those in start-ups, unless otherwise specified. 7 Data were legally acquired under official license agreements from a data vendor, and not directly scraped from LinkedIn.
8 The Word2Vec skills model used in this analysis has been developed in an industry context and groups skills in businesses. It has been trained on a large amount of LinkedIn skills data across a large number of businesses.

Skills synthesis
The skill taxonomy is not wholly mechanically developed but utilizes data from desk research about existing literature concerning skills and capabilities in a CE context and expert discussions; this approach seeks to further contextualize and inform the interpretation and aggregation of clusters. The researchers developed the skill taxonomy based on the results of a clustering exercise. K-means, as a traditional cluster analysis method, is an exploratory tool to understand underlying patterns in data but also requires human judgement in labeling and interpretation Type of capabilities and skills needed for CBM innovation in incumbents and start-ups (source: authors). Note: G/S/C denote the type of capabilities or skills: G = General | S = Sustainable | C = Circular. X indicates whether a type of capability or skill is needed for business model implementation.  ( Breunig, 2009, 2012;Feldman and Sanger, 2007). As such, there is no purely mechanical way to build such a taxonomy; the process is "a mix of mathematics and intuition" (Dave, 2019). In this case, the clustering effort grouped skills logically based on co-occurrence, and the results formed the starting point for the skill taxonomy. In a manual exercise, the researchers interpreted and labeled each cluster, then aggregated and combined related clusters iteratively towards a total of 20-40 high-level skills (i.e., consolidated sets of self-declared skills, as outlined in the theoretical framing). Some clusters did not contain a meaningful set of self-declared skills. 9

Review of capabilities and skills proposed in existing CE literature
The authors specifically searched Elsevier's Scopus database for literature proposing skills and capabilities in a CBM context. The search term included 'capabilities,' 'competences,' and 'skills' in conjunction with 'circular economy' and 'business' (as well as synonyms for 'business'). Appendix C provides details about the literature analysis and an overview of the literature reviewed. The researchers searched articles for skills or capabilities that were then used to contextualize and inform the taxonomy (see Section 2.2 and Appendix A for more details on proposed skills and capabilities). For example, the skill 'business propositions/strategy' was interpreted using the perspective of Sumter et al. (2021), who identify the skill as 'circular business propositions.'

Expert interviews
The researchers collected feedback about initial findings from scholars engaged in CE research and from practitioners working mostly in circular start-ups and in CE-related consulting (see Appendix D for more details, including an overview of the expert interviews). First, the researchers explained the idea of a skill taxonomy and how it is typically used. Presenting the initial draft of the skill taxonomy, the researchers asked interviewees the following questions: Is there anything you would want to add to this taxonomy? Is there any skill you would want to omit? Is there any skill you would formulate differently? Are there any other thoughts you want to share with us on this topic? The researchers deliberated on the comments received and accordingly revised the taxonomy. For example, the researchers included the skill 'Policy monitoring' as a skill on its own (rather than combining it with 'Market monitoring') after one interviewee, based on experience, shared how understanding the emerging (usually conducive) CE policy landscape can help a circular venture build a sustained competitive advantage.

Results
The proposed skill taxonomy is presented in Fig. 6 (Appendix E lists skill frequencies). In total, the taxonomy includes 40 skills (definitions, in a circular business context, are presented in Table 3) that are grouped into six categories adapted from O*NET (2022). These categories are: • Business innovation skills: Developing and seizing innovative business propositions • Operational business skills: Solving business problems in real-world settings and allocating resources accordingly • Social skills: Working constructively with people to achieve goals • Systems skills: Understanding, monitoring, and improving sociotechnical systems • Digital skills: Developing and managing IT and data • Technical skills: Applying technical knowledge in relevant business domains Distinctive features of the skills in the taxonomy are presented in this section, based on differences across sectors and business model types, perspectives found in CE literature, and expert opinions. Similar to skills and capabilities proposed in the literature (Section 2.2 and Appendix A),

Fig. 5.
Word cloud of top-50 self-declared skills. Note: Basic digital tools found within the dataset (e.g., Microsoft Office), social media skills, and language skills were not considered in this figure.

Table 2
Example of cluster including frequency of self-declared skills. The researchers applied logical disaggregation of a cluster and reassigned skills into more logical skill sets. Additionally, the researchers translated non-English self-declared skills and excluded generic skills related to specific industries (hospitality/food, retail fashion, and non-renewablesas this study aims at a general taxonomy), generic Microsoft Office skills, and language skills.
general, sustainable, and circular skills are identified (although the scraping exercise did not identify skills specifically labeled 'circular'). Many general skills are found in circular and linear companies alike, as noted in our discussions with scholars and practitioners. Interpretation of skill frequency should proceed cautiously. Skills with a low frequency are not necessarily less important, as these might be needed only by relatively few employees (Kirchherr et al., 2018a). Additionally, frequency depends on the level of granularity of skills declared in LinkedIn. For example, software engineers declare many programing tools, adding to the frequency of the skill 'application design/development' (section 4.5); at the same time, many employees declare 'project management' once. Hence, a higher frequency for application design/development does not indicate higher importance than project management. In some cases, a particularly low frequency might indicate that a skill is not common.

Business innovation skills
Seven skills are listed in the category 'business innovation skills.' The first, 'sustainable purpose,' can be considered fundamental: it may be reflected in the value proposition and the priorities of the management (some start-ups examined are social enterprises or B Corps), and might draw from sustainability frameworks like the SDGs (Santa-Maria et al., 2022). Sustainable purpose helps employees define commitment towards sustainable aims (Kirchherr et al., 2017), and there is increasing focus on environmental commitment by businesses. However, this study indicates that start-ups in the fashion/textiles and transport sectors have particularly low frequencies for this skill.
Any CBM is implemented through 'project management' (Prieto--Sandoval et al., 2019), but implementing them may be as or more difficult than developing them (Janssens et al., 2021). Effective project management is essential, as the implementation of new strategies requires management of change (a skill found among start-up employees). Circular projects can be complex and involve many stakeholders (Köhler et al., 2022;Sanchez and Haas, 2018), often requiring new ways of approaching project management (Ismayilova and Silvius, 2020). Recognizing the importance of CE project management, the French industry organization AFNOR devised associated standards (AFNOR, 2018). For this study, one expert states that project management "is a skill that sounds dull, whereas so essential to get a business off the ground." Scholars frequently mentioned project management together with 'quality control & continuous improvement.' There exist few proven CBM, particularly regarding sustainability performance. The novelty of CBM as a concept requires continuous improvement (Prieto-Sandoval et al., 2019), including in collaborations within supply chains (Calicchio Berardi and Peregrino de Brito, 2021) and in broader ecosystems across the natural and built environments (Joensuu et al., 2020). According to Velenturf and Purnell (2021), "implementing a circular economy is a process of continuous improvement in which the [circularity and] sustainability of practices is continuously monitored, evaluated and adapted." For example, continuous improvement is found to be an effective enabler for waste elimination in agri-food supply chains (Kusumowardani et al., 2022).
Launching CBM, as with linear models, is dependent in part on 'investments & financing.' Rather than developing necessary capabilities internally, circular businesses might acquire other firms (Khan et al., 2020). While this CBM innovation approach is more common among incumbents , data indicate that employees exhibit these skills across business model types (with the exception of service-based firms). Financial markets might not yet provide sufficient support for circular businesses (Dewick et al., 2020), but investment and financing skills are already apparent.

Operational business skills
The category 'operational business skills' includes six skills. First, 'business/operations/product management' appear frequently in the data, encompassing basic operational business management skills like 'operations management,' 'negotiation,' 'forecasting,' 'product management,' and 'international business.' The data also show 'management' as the most frequent self-declared skill, indicating that employees need basic, transversal management skills to organize business operations. These 'ordinary' skills, in combination with other skills, can Note: There is no order of importance among skills in the taxonomy. Different sizes of boxes do not imply order. Sustainability/environmental skills are italicized with a long-dash outline; circular skills are italicized and bolded with a short-dash outline; all other skills are general.   Sustainability and environmental management skills are identified, but with relatively few occurrences compared to their frequent mention in CE literature on capabilities. Lack of environmental management skill constitutes a barrier to CE implementation in small-and medium-sized enterprises such as circular start-ups (Mishra et al., 2022). Environmental commitments should be operationalized, including in supply chains (where policy barriers exist; Kazancoglu et al., 2021). Adequate internalization of environmental management systems (EMS) can promote circular innovations (Geng and Doberstein, 2008;Marrucci et al., 2022;Scarpellini et al., 2020b). The term 'adequate' suggests that EMS should not be limited to achieving formal environmental certification but also include capabilities such as environmental accounting (Scarpellini et al., 2020a,b). For example, some new start-ups are offering carbon accounting solutions, an emergent concept (Planetly, 2022;Watershed, 2022). Another tool in environmental management is sustainability and ESG reporting (Santa-Maria et al., 2022;Lozano, 2020), which appears only modestly in the data. Establishing advanced formal environmental management operations is complex and costly, potentially explaining why start-ups engage less formally and only to the extent needed .
'Legal' skills ('corporate law,' 'legal advice,' and 'legal research') and 'governance' skills do not appear as frequently as the literature suggests. Internal governance and collaboration-based governance have been cited as important for CE implementation (Khan et al., 2020;Köhler et al., 2022;Scarpellini et al., 2020b), but formal governance structures are often less established in start-ups . Collaboration within ecosystems and along supply chains (section 4.4) raises issues like intellectual property and legalities around engagement. Scholars argue that intellectual property (e.g., circular patents; Marín-Vinuesa et al., 2021;Portillo-Tarragona et al., 2022) can impact implementation of CE strategies like remanufacturing (den Hollander et al., 2017). Required legal skills might be held by few employees (Janssens et al., 2021) or be outsourced.

Social skills
The category 'social skills' includes seven skills. The first is 'customer service,' as emphasized by literature on design thinking and CE (Andrews, 2015;Brown et al., 2021;Prieto-Sandoval et al., 2019). Extended customer eco-engagement can facilitate CBM implementation (Elf et al., 2022). This study distinguishes 'customer service' from 'marketing & sales,' as the latter concerns solicitation of new interest as opposed to serving existing customers. This is the most frequently appearing skill across sectors. Similar to basic management skills (section 4.2), these skills can be considered ordinary yet required in any business. Notably, while circular start-ups as new businesses operate in competitive markets alongside linear businesses, few references to marketing and sales appear in the CE literature on skills and capabilities.
'Storytelling' concerns internal and external business communication, individual communication and public speaking skills, and writing skills. While such skills are required in any business and frequently mentioned by employees in the data, only a limited number of employees declare 'environmental storytelling' skills like 'environmental awareness' and 'environmental education.' Circular storytelling, as proposed by Sumter et al. (2021), is not focused solely on selling. It also concerns visioning of circular futures Calisto Friant et al., 2020) in a way that fosters CE support beyond customers. According to one expert, circular storytelling entails explaining "circular ideas in a 'normal business sense,'" using 'linear language' to meet decisionmakers 'where they are' and ultimately shift priorities and ideas towards circularity.

Systems skills
Seven skills are classifiable as 'systems skills.' There is little evidence of 'systems thinking' skills in the data, with only five explicit mentions across 113 start-ups. On the other hand, systems thinking is frequently discussed by CE literature and sustainability literature more broadly: Vona (2021) classifies it as a key 'green' skill. The role of systems thinking in CE has also been highlighted by Blomsma and Brennan (2018) and mentioned as a skill by Kristoffersen et al. (2021), Santa--Maria et al. (2022) and Sumter et al. (2021). A circular entrepreneur interviewee stated that "[as a circular start-up founder] you're fighting two fights: one against other companies (like any other new company) and at the same time one against the economic system." Systems thinking entails understanding the currently dominant linear system while identifying opportunities for circular ventures and anticipating sustainability impacts within broader socio-economic and socio-technical systems. Other skills have a systems dimension (e.g., environmental science, supply chain collaboration, environmental engineering, information systems, and application design/development), underscoring the importance of systems thinking even if not explicitly declared by employees.
'Value chain collaboration' concerns developing value chain and supply chain bonds that help CE ventures succeed (Agyabeng-Mensah et al., 2022;Galvão et al., 2020;Geissdoerfer et al., 2018;Kanda et al., 2021;Stekelorum et al., 2021). This skill appears in the data in the context of strategic partnerships and stakeholder engagement. Köhler et al. (2022) highlight the link between cross-sectoral collaboration networks and the advancement of CE innovations in the construction sector. Such collaboration and co-creation involve problem-solving approaches, fair and transparent information-and burden-sharing, and trust-based relations (Agyabeng-Mensah et al., 2022;Köhler et al., 2022;Leising et al., 2018;Schönborn and Junge, 2021). Few employees mentioned this skill, potentially because the sample contained few service-based CBMs and because the start-ups covered seem to engage less formally in their supply chains .
'Ecosystem building' is frequently mentioned in the data (more so in customer-focused service-based start-ups) and in discussions with experts. This skill focuses on building networks beyond direct business interactions, and can include social networking, community-building, and event management (skills found among start-up employees). Related skills are 'circular storytelling' and the diffusion of circular futures. 10 Occurring also among linear businesses, it can be considered an ordinary, general skill.

Digital skills
Four digital skills are included in the taxonomy. While some CE literature addresses digitization for supporting CE (Alonso et al., 2021;Okorie et al., 2018;Pagoropoulos et al., 2017) and skills concerning data science (e.g. Kristoffersen et al., 2021), the scholarship largely overlooks digital skills explicitly. Scholar interviewees highlighted the need to include such skills in the taxonomy. Most companies require employees to hold basic digital skills like 'application design/development' and 'IT excellence' (both mentioned frequently in the data). These skills are not only enablers of other skills in the taxonomy but also drivers of digital business models, digital products/services, and smart/IT-based manufacturing activities (Rosa et al., 2020). Application design/development skills are found primarily in platform-based start-ups, but also in design-and service-based start-ups.
The third digital skill, 'data analytics/science,' is necessary for developing 'business analytics' capabilities and facilitating CBM implementation (Awan et al., 2021b;Kristoffersen et al., 2021). This skill is relevant for the types of complex supply chains in which many 10 Both 'value chain collaboration' and 'ecosystem building' rely fundamentally on social skills. Both are included in specific categories because social skills are more inward-oriented (from a company's perspective), whereas systems skills are more outward-oriented (the exception being 'storytelling'). circular ventures are involved (Stekelorum et al., 2021) and supports the quantitative CE metrics and models needed for circular impact assessments (Pauliuk, 2018;Walzberg et al., 2021). Notably, data revealed that few employees in waste-based start-ups declared data analytics/science skills.

Technical skills
The 'technical skills' category includes nine skills considered as essential enablers of CE implementation (Triguero et al., 2022). The importance of 'materials analysis' skills is recognized in scholarship (Allwood, 2014;De los Rios and Charnley, 2017;Janssens et al., 2021). This skill is found mostly among employees of agriculture/food and biotech (nature-based) start-ups, but scarcely mentioned in sectors like fashion/textiles and manufacturing/material engineering (where start-ups, mainly design-or waste based, do not seem to conduct deep material analyses themselves or consider new material inputs that need to fulfil sophisticated characteristics).
While design for multiple use-cycles and recovery is frequently proposed as a key CE topic (den Hollander et al., 2017) and as a skill (Sumter et al., 2021), experts noted that this type of design may not be relevant for some CBM. Data indicate that general design-related skills are most frequently declared by employees in the built environment/design and manufacturing/materials engineering sectors. Sustainable product/systems design skills were originally grouped into a single cluster, but the researchers distinguished product/systems design from sustainable design because the majority of skills refer to general design excellence skills (e.g., AutoCAD, Solidworks, and 'design thinking') while few employees declare their design skills as explicitly sustainable.
'Engineering excellence' skills were identified mainly in design-, waste-, and nature-based start-ups, and relates to processes, systems, manufacturing, and production. The less frequent connection to 'environmental engineering' (one of two circular skills identified) focuses on lower-level CBM strategies like recycling . While the start-ups in the sample also cover higher-level CBM strategies like reducing and reusing , employees did not explicitly declare related engineering skills (e.g., maintenance, and reverse re-manufacturing/repairing) -contrasting with trends in the literature (De los Rios and Charnley, 2017;Khan et al., 2020;Prieto-Sandoval et al., 2019;Sumter et al., 2021).
The second circular skill identified, 'Energy efficiency & sustainable energy,' is relevant to CE as "eco-innovations to support energy efficiency and the exploitation of renewables are considered important investments in the CE" (Scarpellini et al., 2020b). Energy management (along with an energy efficiency culture) enables other circular activities (Cavicchi et al., 2022) and is thus a key skill (Janssens et al., 2021;Mishra et al., 2022). While common in energy sector start-ups, the frequency of this skill is mixed across other sectors; only declared as a skill by a low number of employees, start-ups rather have energy experts than a widespread energy culture across their employee bases.
Finally, 'impact assessment' is infrequently identified in the data. Though highlighted by the literature (Janssens et al., 2021;Sumter et al., 2021), most companies have not fully developed this skill (Mishra et al., 2022). Methods for measuring the circular economy have risen in research salience (Corona et al., 2019;Moraga et al., 2019;Morseletto, 2020), and industries and institutional bodies continue to identify and elaborate standards and approaches. As such, the low frequency of this skill in the data is notable.

Discussion
This section begins by contrasting the identified skills in the taxonomy with the skills and organizational capabilities proposed by CE literature. It then discusses why employees infrequently declare their skills as 'circular.' Further, pathways forward are proposed, including the need to adopt a more holistic perspective in recognizing the broader role of CE and how the taxonomy helps advance this effort. Finally, practical and scholarly implications of the skill taxonomy are discussed.

Comparison with skills and capabilities proposed in CE literature
The authors analyzed the skills in the taxonomy in two rounds of analysis against the comprehensive sets of (i) individual skills and (ii) organizational capabilities proposed in CE literature (Appendix A). Appendix F offers a detailed description of this comparison. While mapping taxonomy skills against literature skills was a straightforward process, the mapping of taxonomy skills (as a microfoundation of capabilities) against literature capabilities was also possible; some skills and capabilities 11 could be directly mapped (for example, project management) while a partial relationship was found for others (Table F3, Appendix F). This finding is consistent with microfoundations theory: capabilities are not simply the sum of individuals and their skills but an aggregate of all microfoundations (Section 2.1) and their interactions (Barney and Felin, 2013). Fig. 7 presents a heat map indicating to what extent taxonomy skills are supported across skills and capabilities proposed by CE literature. Many taxonomy skills have been proposed or there exists a correlation with skills or organizational capabilities proposed in the literature; taxonomy skills thus empirically confirm findings in the literature. Furthermore, this study identified new skills to implement a CBM that have not been proposed in the literature. Three taxonomy skills are not found in the CE literature. Two of these, application design/development and graphic design & multimedia, are digital skills. Given that IT excellence has been only partially identified, this study finds that the literature has neglected the need for digital skills (beyond data analytics) in implementing CBM. Data show that the need for digital skills extends beyond platform-based CBM (Section 4.5). The third skill, environmental science, is technical in nature and demonstrates that circular start-ups require a thorough understanding of complex systems in the natural environment. Additionally, eleven skills were only partially identified in CE literature; elements of these skills are newly introduced by this study (Tables F1 and F2, Appendix F).
This study's comparison also identified, from the literature, skills and organizational capabilities that could not be mapped to taxonomy skills. Table 4 shows employee skills that are proposed in existing literature but not found among employees in circular start-ups. Similarly, the table shows organizational capabilities proposed in existing literature, but no employee skills in the dataas a microfoundation of these capabilitiescould be mapped to them. This gap might be explained by sampling methods, the labeling of skills by employees on LinkedIn, and the focus of CE research on incumbents. For example, employees declare cloud computing skills (related to Industry 4.0 technology) but start-ups might not adopt other Industry 4.0 technologies like internet-of-things, as no further specialized Industry 4.0 skills (beyond cloud computing) (Wahl and Munch, 2021) could be identified. Research has examined the dynamic capabilities and aggregate microfoundations needed for the process of innovating and transitioning towards CBM within incumbents (rather than the comprehensive set of capabilities required to manage a CBM) (Khan et al., 2020;Santa-Maria et al., 2022); these capabilities are necessary as business model transformation beyond existing structures can be inhibited by a lack of flexibility and a change-resistant culture, also related to jobs that might become obsolete. Nevertheless, such limitations do not typically apply to start-ups adopting CBM, as they are inherently more flexible .
With the exception of the two circular skills identified (Section 4), many explicitly circular skills and capabilities proposed in the literature were not found in the data. This finding is unexpected given that the sampled start-ups engage in relevant circular activities like industrial symbiosis and service-based offerings . Whereas circular skills and capabilities relating to these two activities are proposed in the literature, employees do not declare them explicitly. While some circular skills and capabilities were not found (Table 4), many circular skills and capabilities proposed in the literature were found to be partially related to general or sustainable skills in the taxonomy but lacking circular 'framing' (Table F4, Appendix F). For example, the literature proposes the skill of circular storytelling, while employees declare general and environmental storytelling. Similarly, some sustainable skills and capabilities proposed in the literature were found to be partially related to general skills in the taxonomy (Table F5, Appendix F). For example, 'marketing,' as found in the taxonomy, is proposed in the literature as 'green marketing' in businesses adopting a CBM.

The role of circular skills in CBM implementation
This section discusses a rationale along five arguments why the majority of skills are not declared by employees as explicitly circular although they are utilized in a circular context. First, although there are some new circular skills identified in the study, a range of general, sustainable, and circular skills is needed for CBM implementationor, more specifically, to perform the activities necessary for implementing and running CBM in start-ups (including activities unique to CBM). Reverse logistics activities, for example, require general logistics skills (found in this study), i.e., proficiency in moving a good from source to destination, which can be applied to both forward and reverse logistics operations. Additionally, for CE value chain collaboration, employees in circular start-ups declare general collaboration/strategic partnership skills while involved in activities related to network and partnering operations (e.g., fostering industrial symbiosis in waste-based start-ups). As a final example, remanufacturing activities require general manufacturing skills but also skills to handle and integrate used parts in the rebuilding operation, such as quality control of used parts. Quality control (found in this study) is a general skill that is also needed to assess the quality of new parts and thus existed before concepts about circularity arose.
Second, although these business models have been given the 'circular' label, they may not be considered entirely new. As discussed in the literature, notions and variants of CBM have existed for decades (e.g., product-service offerings and waste recycling; Geissdoerfer et al., 2020;Linder and Williander, 2017). As such, skills existing in the workforce for decades may also be expected in this study. The two circular skills present among start-up employees concern sustainable energy and environmental engineering, relating to technical fields including renewables, energy efficiency, waste and water management, and recycling. Notably, environmental engineering skills have existed for some time, including in traditional waste and water management companies  adopting lower-level CE strategies.
Third, the use and degree of organizational embeddedness of general, sustainable, and circular skills are crucial determinants of their impact. Or put differently, it is more pertinent what the skills are applied to. For example, utilizing out-of-the-box thinking skills, an employee can determine how to build the next linear business or how to scale a circular start-up. Capability theory suggests that skills are needed in conjunction with other microfoundations (Barney and Felin, 2013) -namely processes, technology, and structuresto enable organizational capabilities and perform activities for CBM implementation. The mapping exercise showed that many capabilities correlate with one or more skills (as a microfoundation) needed to adopt other microfoundations (Table F3, Appendix F). Building on skills as the starting point and enabler (Section 2.1), organizational capability development depends also on the quantity and quality of skills (i.e., how many employees need them and at what level of proficiency) and on their positioning and configuration relative to other microfoundations.
Fourth, sourcing of skills is a relevant factor. Circular businesses deploying innovative business models might not possess all necessary skills to begin. As such, they must channel existing employee skills towards circular ideas and activities, through learning processes and continuous improvement (both identified as skills in this study). 12 This need is especially salient for start-ups that must find employees who are willing to join risky endeavors but might not have perfectly matching skill profiles.
Finally, there exists a crucial institutional dimension from a CE transition perspective. The data show that employees do not declare many of their skills as explicitly circular, even though they work on CBM implementation. For example, these employees declare general supply chain management skills that can apply to forward or reverse supply chains. They may not necessarily consider or interpret skills within the circular context, suggesting that circularity is often shaped by framing in mainstream practice. In strategic and engineering fields, employees may not harbor a 'circular perspective' with respect to their skills and may instead be focused only on the mechanics of operations as taught in mainstream business or engineering schools. Many employees working in other operations of a firm, including those with no role in strategy or engineering, may not consider their own work 'circular' or fail to see a need to reframe their skills.

Pathways forward towards more circular skills
Findings suggest that skills framed specifically as circular may still be emerging in their practical conceptualizations, including among startups. Many skills are prevalent in their general framing among employees and can be applied in varying (linear or circular) contexts. The underdeveloped circular framing of skills constitutes a barrier to the wider dissemination of CE as a concept in business and society.
Given these circumstances, the researchers in this study call for reframing efforts, particularly as certain skills are becoming more mature and differentiated with increasing attention given to regularized and complex circular operations required for CBM implementation. Skills may be interpreted (and distinguished from applications in linear operations) in more nuanced ways in the circular context, as suggested in the literature (Sumter et al., 2021). Accordingly, the taxonomy skills have been defined in a circular context in Table 3 (Section 4). For example, employees in service-based circular start-ups might declare their general business proposition/strategy skills as circular or extend skill-framing to include a service dimension, as they mature in applying their general business proposition/strategy skills in the circular product-service systems context. Also, given that start-ups often already tell explicitly circular stories (e.g., on their websites), employees might declare their storytelling skills as circular.
A stronger effort by businesses is needed to identify and develop circular thinking among all employees. CBM implementation is influenced by decisions across all business functions including in strategic management, marketing, logistics, digital and financeand execution of these functions from upper management to the 'ground level.' This holistic perspective is under-recognized but has the potential to support novel thinking about CBM implementation and the employee skills needed for it. Circular narratives (through circular storytelling) can promote understanding and recognition of circular skills among the employee base and beyond, enabling wider CE transition towards the mainstream.
Given the ubiquitous relevance and need for mainstreaming CBM among businesses and CE in a wider societal context, research has highlighted the role of universities and education in transitioning towards a CE (Kopnina, 2021;Rokicki et al., 2020;Stevens et al., 2021). Aiming for mainstream, holistic uptake of circular thinking and skills in businesses and society, such educational approaches can go beyond the provision of degrees related explicitly to CE: one example is to require a module or course on CE and sustainability in all degree programs (e.g., business, engineering, and political science) or as a university-wide 'core' subject. Accordingly, Kirchherr & Piscicelli (2019)  The comprehensive skill taxonomy proposed in this study intends to provide a holistic conceptualization of skills required for CBM implementation.

Practical and scholarly implications of the skill taxonomy
After a skill taxonomy has been developed by a company, the taxonomy typically serves as a basis for quantifying the company's performance on internal skill development. Gaps identified through a taxonomy-based skill analysis can support efforts to drive business performance, whereupon a company may undertake targeted recruitment. Additionally, a company may choose to close skill gaps via selected upskilling and reskilling efforts (Fenton et al., 2021). Companies may also use skill taxonomies in performance management, with skills outlined in the taxonomy serving as a benchmark for performance assessment and pathway for career advancement. In these and other ways, a skill taxonomy can help HR activities meet broader strategic goals.
At the same time, merely closing identified skill gaps may not be sufficient for circular business model performance. As argued in Section 5.2, skills existing in a company need to be utilized. Effective utilization of skills includes and is dependent in part on developing and running processes, organizational structures, and technologies as complementary micro-foundations. Incumbent workers may have no current opportunities to use certain skills because leadership is not aware or fails to appreciate the value of these skills for CBM implementation. This limitation reflects the finding by Kirchherr et al. (2018b) that hesitant company culture is a principal barrier to implementing CBM. Existing skills that enhance CBM performance should be integrated into work streamsan effort that requires companies to coordinate microfoundations and (re)design processes, structures, and technologies. These are the types of systemic interventions, going beyond incidental 12 Another strategic approach is to recruit key individuals who possess an extensive set of skills in a particular domain, in order to add these skills to the organization's overall skill profile. However, such 'superstars' are limited in number and their employment is often intended for leadership roles (Felin et al., 2012;Felin and Hesterly, 2007). As such, these individuals alone cannot provide the comprehensive mix of skills needed but rather accelerate the development of skills among employees through leadership and coaching (both identified as skills in this study). personnel or technical adjustments, that are needed for meaningful CBM implementation. Accordingly, a skill taxonomy can be useful also for enterprise architecture functions (Kitsios and Kamariotou, 2019;Zhang et al., 2018), supplementing conceptual tools like capability maps and process maps.
The proposed skill taxonomy can also advance theory. For example, scholars studying CE can map identified skills onto CBM cases, aiming to further understand the role that skills play in implementation. The taxonomy can also serve as an analytical framework to identify capabilities and connect micro-level capacities (i.e., employee skills) with macro-level capabilities at the organizational level (Felin et al., 2015). The taxonomy may also guide further research on ECE, possibly utilizing it as an analytical device to test if the skills that are proposed in this article are included in the CE-oriented curricula and programs.

Conclusion
The CE concept has gained substantial momentum in the 21st century as a key facilitator of sustainability efforts. Given the influence of business decisions not only on environmental conditions but also on consumer preferences and habits, the private sector is recognized as a key catalyst for society-wide CE transition. Nevertheless, substantive progress towards CE transition remains limited. While refashioned strategies and supportive infrastructure provide businesses with some pathways, skills for CBM are an often-overlooked topic in the academic literature and in practice. A modestly sized literature offers some useful insights (e.g., Janssens et al., 2021;Sumter et al., 2021) but a systematic understanding about the relevance of employee skills to CBM implementation has yet to be fully researched or integrated into practice. This study has sought to fill this gap not only by outlining skills in circular start-ups but also by presenting a way to refine understandings about these skills and help businesses identify and cultivate them.
This study proposes the first comprehensive skill taxonomy for CBM implementation in start-ups in the literature, as far as the authors can determine. The taxonomy includes 40 skills for CBM implementation and finds that CBM implementation requires a set of general, sustainable, and circular skills. It also finds that some skills, such as digital skills, have been neglected. Skills declared as specifically circular are not as common in circular start-ups as the literature suggests. Given that CBM is not an entirely new concept, some skills identified in this study have existed in the workforce for decades. Thus, the novelty of skills for CBM implementation lies in the shifting context of their application and in their utilization as microfoundations of organizational capabilities. Circular start-ups might need to develop existing employee skills in novel or differentiated circular application contexts. Consequently, using circular narratives as a framing device for skill development can promote understanding and recognition of those skills in efforts to mainstream CE.
Ideally, the skill taxonomy can be used for activities such as skill mapping, targeted recruiting, upskilling and reskilling, performance and career management, and ECE. Additionally, this study merged the skill taxonomy concept with the theory of capabilities, which is largely considered a valuable contribution in understanding skills as key microfoundations.
This study has several limitations that suggest avenues for further research. First, the taxonomy provides an analytical lens to understand skill needs for CBM implementation but does not address organizationwide skill quantity and proficiency. Second, the study takes a supplyside (employee) perspective in examining skills and does not extend to analyzing whether and how these skills match labor demand (i.e., what businesses state that they need). Further research can develop an overview of needs based on activities and capabilities for CBM implementation and can compare those needs to the supply of skills available in the workforce. Such efforts might also consider how skills interact with other microfoundations. Third, this study did not compare skills across different types of businesses, (circular versus linear businesses; circular start-ups versus incumbents adopting CBM; degree of commercial success experienced by businesses) which could generate a more complete overview of skills necessary for CE transition. Fourth, given that this sample of start-ups considered only the Netherlands, UK, and Germany, future research should also examine whether and how skill sets vary by geography. Finally, the study considers only skills declared on LinkedIn, thus missing the skills of employees who do not use Link-edIn or list their skills there. In a self-reporting context like LinkedIn, individual subjectivity may also threaten validity. Regarding analysis of data, the methodological approach required some degree of researcher judgement, so any bias such as it might have arisen could be resolved through the adoption of more quantitative methodologies (e.g., topic modeling and model-based clustering). While the proposed skill taxonomy aims to be an analytical device for both scholars and practitioners, it is not proposed as conclusive but rather as a prompt for further research. It is anticipated that it will motivate more scholars and practitioners to examine skills for CBM implementation as a worthwhile topic.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
The authors do not have permission to share data.

Appendix A
Skills of individuals and organizational capabilities proposed in the literature are summarized below. Researchers have used various concepts in the context of organizational capabilities: operational capabilities, dynamic capabilities, and aggregate microfoundations of dynamic capabilities. These are listed in separate tables. Additionally, as the literature was reviewed, three types of skills and organizational capabilities became apparent: General skills or capabilities refer to skills or capabilities that can be found in traditional businesses (such as project management or teamwork). Sustainable skills or capabilities refer to skills or capabilities that are related specifically to aspects of the triple bottom line, but not to circularity (such as sustainable mindset or environmental commitment). Thirdly, circular skills or capabilities refer to skills or capabilities that are related specifically to aspects of circularity (such as reverse logistics or industrial symbiosis). Some scholarly publications do not distinguish between sustainable and circular notions, for example when attributing aggregate microfoundations to business model types (Santa-Maria et al., 2022); yet we deem it important to distinguish them. Accordingly, skills and organizational capabilities are grouped by these skill types in the tables.

Table A1
Operational capabilities for CBM as proposed by CE literature       (2017)      identify five business model types for circular start-ups. Short definitions of start-ups, taken from , are provided below: • design-based: adopting circular innovations mostly in the pre-market phase through source material minimization, product design or production process efficiency • waste-based: seeking to extract value from unexploited external waste streams • platform-based: pursuing sharing/trading business models built around B2B, B2C, or C2C marketplaces • service-based: embedding products in service-systems to increase usage efficiency • nature-based: increasing the delivery of (products and) services based on nature-based systemic solutions The 50 most frequently declared skills on LinkedIn profiles are listed below. Basic digital tools found within the dataset (e.g., Microsoft Office), social media skills, and language skills were not considered.

Appendix C
We reviewed the CE literature on skills and capabilities to identify which skills and capabilities have already been proposed in the context of CBM implementation. Following Henry et al. (2021), we used Elsevier's Scopus database due to its larger coverage compared to Web of Science. We searched for relevant articles with the following search term: TITLE-ABS-KEY("circular economy" AND (skill* OR capabilit* OR competenc*) AND (business* OR firm* OR enterprise* OR start?up* OR corporate* OR organi?ation*)) The search returned 339 articles, which were then further assessed for their relevance by the authors. First, by reading the title and abstract of each article, we identified whether the article studies skills or organizational capabilities in the context of CBM implementation. Following this criterion, a subset of 82 articles was created. The exclusion of many articles in this step is due to the frequent and broad usage of the term 'capability' beyond the context of organizational capability theory. Next, we examined each of the 82 articles in detail and identified those proposing specific skills or capabilities in the context of CBM implementation. A final set of 57 relevant articles was created (see Supplementary Materials). Lastly, we extracted and summarized skills and capabilities that were proposed in these articles (Section 2.2 and Appendix A), using this information to contextualize and inform the taxonomy.

Appendix F
In two rounds of analysis, the employee skills identified in this study were compared against the lists in CE literature of (i) employee skills and (ii) organizational capabilities (operational capabilities, dynamic capabilities, and aggregate microfoundations) (Appendix A). In particular, mapping skills against capabilities must consider the conceptual relationship of employee skills as one of multiple microfoundations of organizational capabilities, which makes it less likely to identify a one-to-one match. Subsequently, when referring to the mapping of taxonomy skills against skills and capabilities in the literature, the conceptual difference of organizational capabilities and skills as a microfoundation of these capabilities is assumed. Consequently, skills and capabilities must not be considered as synonyms. Every skill in the taxonomy and every skill or capability in the CE literature was then classified as found (if a perfect match was identified), partial correlation (if a connection was identified based on the researchers' judgement) or not found. Classifying any skill-skill or skill-capability comparison as found required a perfect match. For example, the skill 'project management' in our taxonomy is proposed as a skill as well as an organizational capability in the CE literature. Partial correlation classification was applied if an explicit connection was observable but no perfect match. For example, the skill 'environmental storytelling' in our taxonomy was compared with 'circular storytelling,' proposed as a capability in literature, and classified as a partial correlation. We introduced the partial correlation classification to avoid exaggerating any contributions of this study by identifying skill as strictly found or not found. We aimed at making 'fair' arguments and acknowledge that these classifications are judgements of the researchers.
We then summarized our findings into five tables. Table F1 gives an overview of the skills in our taxonomy mapped against skills proposed in the CE literature. Similarly, Table F2 shows an overview of the skills in our taxonomy mapped against capabilities and microfoundations proposed in the CE literature. Tables F1 and F2 also serve as the basis for the heat map in Section 5.1. For each skill in the taxonomy, the mapping against skills and capabilities in CE literature was synthesized through a logical OR evaluation. For example, the skill 'science' is indicated as found in the CE literature on skills and as not found in the CE literature on capabilities. Based on the logical OR evaluation, the synthesized classification is found, given that the skill has been proposed in CE literature on skills before.
We found it straightforward to map taxonomy skills against literature skills, but also found it possible to map taxonomy skills against literature capabilities. Some skills and capabilities could be directly mapped (such as the project management example mentioned above), while for others a partial correlation was found while a simple aggregation of the skills to match this capability was not possible. Table F3 provides an overview of literature-based capabilities, where related skills exist with partial correlation while a simple aggregation of the skills to match this capability was not possible.
Furthermore, we summarized circular skills and capabilities proposed in CE literature for which a related general or sustainable skill exists, yet without a circular framing (Table F4). Similarly, we summarized sustainable skills and capabilities proposed in CE literature for which a related general skill exists (Table F5).    Engineering excellence partial correlation capability (related to restructuring of production/manufacturing) found in literature that correlate with this skill Environmental engineering partial correlation higher-ranked circular production/manufacturing capabilities found in literature that correlates with this circular skill (which has a focus on lower-ranked circular strategies Energy efficiency and sustainable energy partial correlation energy-related capabilities found in literature that correlate with this skill Impact assessment found Environmental science not found Science not found

Table F3
Overview of capabilities proposed in the CE literature with partial correlation with skills proposed in the taxonomy of this study Type Capability Mapping against skills in taxonomy General Technological innovation (incl. ICT based) out-of-the-box thinking skills broadly correlate with this capability, together with technical skills and digital (ICT) skills (but tech-based innovation not identified explicitly as a skill) Empowerment for bottom-up innovation out-of-the-box skills thinking broadly correlate with this capability, together with leadership (but bottom-up empowerment not identified explicitly as a skill) Ecosystem/stakeholder engagement and collaboration (vertical and horizontal) on sensing and seizing value chain collaboration skills broadly correlate with this capability, together with ecosystem building, customer service, product design (UX), out-of-the-box thinking, SCM, business propositions/strategy (but collaboration on sensing/ seizing not identified explicitly as a skill) R&D/innovation collaboration value chain collaboration skills broadly correlate with this capability, together with out-of-the-box thinking (innovation) and research (but innovation collaboration not identified explicitly as a skill) Value chain (and social) collaboration (including vertical/horizontal, engagement, information sharing/traceability, governance/trust, shared culture, training, etc.) value chain collaboration skills broadly correlate with this capability, together with skills related to different aspects of value chain collaboration capability, including governance, training, and information systems (but collaboration including all elements proposed by literature not identified explicitly as a skill) Access to stakeholder information value chain collaboration skills broadly correlate with this capability (but stakeholder information access not identified explicitly as a skill) Supply chain ambidexterity SCM skills broadly correlate with this capability, together with out-of-the-box thinking (innovation) and quality control and continuous improvement (but focus on supply chain improvements/innovations not identified explicitly as a skill) Supply chain big data predictive analytics SCM skills broadly correlate with this capability, together with data analytics/ science (but data analytics with focus on supply chain not identified explicitly as a skill) Customer engagement in product design customer service skills broadly correlate with this capability, together with product design (UX) skills and value chain collaboration skills (but customer engagement for product design not identified explicitly as a skill) (continued on next page) Environmental (management) accounting environmental management skills broadly correlate with this capability (but environmental accounting not identified explicitly as a skill) Energy management and auditing energy efficiency and sustainable energy skills broadly correlate with this capability, together with environmental management (yet also n/a as these skills do not indicate specifically an intra-organizational energy management and auditing capability) Circular Continuous systematic learning from product returns (incl. Identification of valuable information, Knowledge infrastructure, Integrated return processes (customer 360 and forward/reverse logistics integration), Governance, Incentives) quality control and continuous improvement skills broadly correlate with this capability, also information systems, SCM and governance (but learning from product returns not identified explicitly as a skill), additionally skills found without circular notion in startups Comprehension of (environmental/circular) regulatory landscape Reverse logistics Circular legal Circular (waste-efficient) procurement Circular/Green/Waste-related patenting Circular (waste) contractor assessment Reverse omnichannel CE value chain collaboration Circular storytelling Use of information systems for CE Circular SCM/purchasing (incl. supplier material/parts certification, integrated SCM system) Circular/efficient material use in design Reverse logistics Design for recovery and multiple use cycles Circular/green IT management Circular manufacturing (including reverse and re-manufacturing, dematerialization, novel manufacturing solutions and use of IT tool/analytics) Circular product (eco-) design (incl. openness to recycled products, use of recycled materials, flexibility, reconfiguration, maintenance, user experience, etc.) Circular impact assessment Circular (production) process design/planning (incl. dematerialization, cleaner production, modular assembly, remanufacturing, recycling, maintenance, etc.) Circular manufacturing (incl. Remanufacturing)

Table F5
Overview of sustainable skills and capabilities proposed in the CE literature with partial correlation with general skills proposed in taxonomy of this study Skill Capability Sustainable material Sustainable business model design and reconfiguration Environmental/ecological economics Focus on sustainable impact commitment/strategy and innovation/ideation culture Green talent management Eco-innovation HR Green marketing Green warehousing