What do we want to know about reshoring? Towards a comprehensive framework based on a meta-synthesis

This study aims to develop a framework for reshoring research, specifying the aspects encompassing a full case description. The article reports an analysis of existing case based research focusing on reshoring and the previous offshoring events, with the aim to advance the knowledge within the field through an accumulation process. The selected methodology to achieve this purpose is meta-synthesis, a type of meta-analysis applied to the analysis of qualitative data. The analyzed data are retrieved from a sample of 14 articles including cases of offshoring and subsequent reshoring processes. The sample represent the total population of articles published in peer-reviewed sources. By focusing on the analysis on the article level and the findings generated by the original authors, the study allowed to build a comprehensive framework to lead future reshoring research, as well as to demonstrate the suitability of the methodology in achieving knowledge accumulation even if the articles have different characteristics in terms of theoretical perspectives, unit of analysis, number of cases and context. In doing so, this study contributes to the field of reshoring and more widely to the operations and supply chain management field. Furthermore, this research generates many practical implications, by condensing all the evidences about reshoring so far collected in one single study, thus providing examples of challenges and best-practices.


Introduction
Reshoring of manufacturing as a means for international firms to optimize their global manufacturing footprints has gained increasing attention in academia. Being quite a recent phenomenon, its magnitude and relevance have been questioned and it has mainly been studied in isolation. However, previous research has recognized a strong path dependency of decisions connected to manufacturing location decisions (Wan et al. 2019), and it is now widely acknowledged that reshoring cannot be pursued unless there has been previous offshoring (Gray et al. 2013). It has even been argued that the reshoring decision can be based on the previous offshoring outcomes, in the sense that less successful offshoring operations may lead to repatriation (Kinkel and Maloca 2009;Di Mauro et al. 2018). Accordingly, researchers have suggested to take a broader perspective and study reshoring in relation to the previous offshoring decision, aiming to understand the complex dynamics behind the interrelated processes forming firms' global manufacturing footprints (Kinkel and Maloca 2009;Barbieri et al. 2018). Reshoring is a relatively new research stream, bringing together scholars from different research fields; with supply chain management, international business and operations management being the most relevant. What unites these fields is that reshoring has primarily been studied conceptually. In fact, a number of frameworks aiming to describe manufacturing relocation have been presented in previous literature. However, these typically only cover relocation in one direction, i.e. offshoring (Mihalache and Mihalache 2016) or reshoring Foerstl et al. 2016;Benstead et al. 2017), or consider reshoring as one of several possible subsequent relocation alternatives to offshoring (Joubioux and Vanpoucke 2016). Given that none of the existing frameworks considers the two location decisions together, this study is moving research forward by adding the connection between offshoring and reshoring.
The emerging empirical evidence on reshoring mainly comes from surveys and case studies (Wiesmann et al. 2017). In particular, several case studies have been conducted to gain a deeper understanding of why and how location decisions are made (e.g. Gylling et al. 2015;Martínez-Mora and Merino 2014;Di Mauro et al. 2018). Indeed, case studies have made substantial contributions in a number of fields, but in general, there have been little accumulation of the understanding gained from primary case studies within Supply Chain and Operations Management (Hoon 2013). This is also true for reshoring, serving as the second major opportunity to move reshoring literature forward, and it has important implications for knowledge development. The rich findings from case studies can reach disparate conclusions, and the knowledge produced tends to stand-alone, while the potential to create cumulative knowledge is being neglected, according to Hoon (2013).
Thus, the meta-synthesis (alternatively called qualitative meta-analysis or iterative triangulation) has become essential for the evolution of knowledge in management studies (Combs et al. 2019). Revisiting the findings of other researchers can accelerate the progress of the field by discovering rich complexities and insightful nuances that cannot be obtained from individual case studies (Goldsby and Autry 2011). Starting from there, developing a comprehensive framework encompassing both offshoring and subsequent reshoring, providing guidance on how to study reshoring, can be helpful in setting a common path and reach a full understanding of the phenomenon. The advantage of such a framework, developed by collecting cases from multiple fields, is its general applicability to study the reshoring phenomenon in a multidisciplinary context.
The purpose of this study is thus twofold. Following the suggestion by Benstead et al. (2017), we first aim to develop a framework for reshoring case study research, specifying the elements characterizing a full case description. Second, we will use the meta-synthesis methodology to analyze existing reshoring case research, advancing the knowledge about this phenomenon by gathering insights coming from multiple fields, as suggested by Goldsby and Autry (2011) and Hoon (2013). The collected evidences will set the basis to refine the framework. In this research, we review 14 articles describing offshoring and subsequent reshoring events. Instead of analyzing the primary data derived from the specific cases, we focus our analysis on the article level and the findings generated by the original authors, in line with the meta-synthesis methodology described by Hoon (2013).
The main contribution of this study is a framework that can lead empirical research on reshoring in terms of how to provide a full case description.
The paper is structured as follows. First, we discuss the relevant literature, with focus on previously developed frameworks. Second, we describe the research design and the methodology. In the main section, we present the final framework and describe it by using illustrations from existing cases, thus synthesizing and discussing the main findings of the original authors. Finally, based on the gaps in previous literature in relation to the framework, we give suggestions on further research and provide implications for managers and researches.

Related literature
Manufacturing location decisions have a long history of publications in the Operations and Supply Chain Management fields (Jain et al. 2016; Barbieri et al. 2018). In particular, starting from the 1980's, many researchers have focused on the offshoring phenomenon that has been the dominant trend for decades (Mihalache and Mihalache 2016). In their literature review, Mihalache and Mihalache (2016) define offshoring as "the assignment of business activities to locations outside a firm's national borders in order to support existing business operations" (Mihalache and Mihalache 2016;p. 1105), thus highlighting that the offshoring decision is considered to be a rational choice, based on the assumption that a host country provides a comparative advantage with respect to the home country.
As such, the offshoring decision is far from being irreversible (Kotabe et al. 2008;Antelo and Bru 2010). As a consequence, a new phenomenon, which entails the reverse movement from the offshore location to the home country, has started to spread in the last decades. Interestingly, both academia and practice have struggled to find the right terminology, as well as to set the boundaries for this phenomenon Stentoft et al. 2016a;Wiesmann et al. 2017;Barbieri et al. 2018). In this paper, we will use the term "reshoring", given that it is the most widespread among recent studies. We adopt the definition of back-reshoring proposed by Fratocchi et al. (2014), defining reshoring as "a voluntary corporate strategy regarding the home country's partial or total re-location of (in-sourced or out-sourced) production to serve the local, regional or global demands" (Fratocchi et al. 2014;p. 56). Moreover, we will specifically focus on the relocation of manufacturing activities, given that service companies have proven to follow a different pattern, being more flexible towards relocations (Albertoni et al. 2017).
According to Barbieri et al. (2018), previous research in this field strongly focused on the "Why?" (44 articles over the 57 analyzed in their literature review), "What?" (35 articles) and "Who?" (28 articles) questions. The "Why?" question deals with the motivations (or drivers) that induce a company to bring production back to its home country (Barbieri et al. 2018). It has been so widely studied, that there are systematic literature reviews specifically focused on the categorization of drivers (Stentoft et al. 2016b;Wiesmann et al. 2017). The "What?" question concerns the definition and characterization of the phenomenon. As stated before, this has been a big concern within academia, even if consensus about what reshoring is has been reached nowadays (Barbieri et al. 2018). The "Who?" question deals with the characteristics of firms embarking in the reshoring journey. Here, contingency theory could play an important role, as pointed out by Benstead et al. (2017). Interestingly, the "Where?" -connected to the geographical aspect -"When?" -that entails the duration of the offshoring experienceand "How?" -about the decision-making and implementationquestions appeared to be much more underresearched (Barbieri et al. 2018). Despite the fact that offshoring research is being more developed than the relatively recent reshoring-focused literature, scholars have acknowledged a shallow understanding of the phenomenon (Mol et al. 2005;Bunyaratavej et al. 2007). Mihalache and Mihalache (2016) recognize that the underlying reason is "the fragmentation of existing research due to limited cross-fertilization between the many research fields studying the phenomenon" (Mihalache and Mihalache 2016, p. 1104) that, in turn, has prevented the accumulation of knowledge. Particularly, it has been claimed that the insights from case study research are difficult to accumulate in the field of Operations Management (Barratt et al. 2011). We believe that research on reshoring can still avoid this trap, given its early stage.
As a starting point for this study, we reviewed frameworks from existing literature in order to build on, and consolidate, the knowledge already produced in previous studies.
In particular, we searched for frameworks of the reshoring process, since this is the main focus of our study. Thus, frameworks only describing the offshoring process were discarded. We noticed that the frameworks developed in the reshoring field are relatively recent, given that they started to appear in 2016. Interestingly, they all focused on different but concurrent aspects of reshoring. Bals et al. (2016) and Foerstl et al. (2016) firstly provided a characterization of the phenomenon in terms of location and ownership, based on the matrix by Jahns et al. (2006), aiming to provide a holistic picture of possible relocation alternatives. Then, Bals et al. (2016) developed a framework focusing on the reshoring decision-making and implementation process, while Foerstl et al. (2016) presented a framework pinpointing the relationship between drivers and outcomes and the moderating role of contingencies. However, none of them provided empirical support for their frameworks. Fratocchi et al. (2016) developed a theory-based framework of reshoring drivers dividing them into four quadrants depending on two elements, namely if they are either internal or external and oriented either towards cost efficiency or customer perceived value. This framework has later been applied to four Italian cases in the textile-clothing-leather-footwear industry to study both offshoring and reshoring drivers (Di Mauro et al. 2018). Joubioux and Vanpoucke (2016) developed and empirically refined the only framework encompassing both offshoring and reshoring, thus supporting the relevance of studying the two phenomena together. Although, they consider reshoring only as one of the possible decisions consequent to reshoring. Finally, Benstead et al. (2017) highlight the relevance of contingencies in their framework, by assuming that both drivers and implementation considerations change according to multiple contingent factors (i.e. related to company and industry, product, and behavioral or individual aspects). They developed and empirically refined a framework that start to show comprehensiveness towards the main aspects that need to be considered by a company when reshoring. The same framework has been adapted and used as basis for quantitative analyses of secondary data (Moore et al. 2018). Table 1 summarizes the frameworks retrieved from previous literature that will serve as the starting point for the meta-synthesis.

The meta-synthesis methodology
This study follows the meta-synthesis methodology developed by Hoon (2013), which is an exploratory, inductive methodology aimed to make contributions beyond those presented in original primary qualitative case studies (Hoon 2013). Supply Chain accumulate the evidence from different studies "such that findings can be more confidently extolled to both academic and practitioner constituencies" (Goldsby and Autry 2011, p. 324). But meta-analysis has rarely been used by OM and SCM scholars, and thus the few exceptions provide valuable contributions and have been highly cited (White 1996;Lewis 1998;Nair 2006;MacKelprang and Nair 2010;Cao and Lumineau 2015;Habersang et al. 2019). The meta-synthesis has been described as an "analysis of the analyses", as it focuses on the insights and interpretations of the original authors, rather than on the firsthand data about the specific cases as obtained from interviews or observations. More specifically, it aims at "extracting, analyzing, and synthesizing qualitative evidence to build theory" (Hoon 2013;p. 523). In this study, we aim to refine theory by developing a comprehensive framework based on existing reshoring case studies. We will use the aggregation synthesis, which refers to the generation of interpretative explanations and accumulation of knowledge by identifying categories and patterns from the studies in focus (Hoon, 2013). In this research we relied on the seven step research design proposed by Hoon (2013). The most relevant steps are described in the following sections.

Conceptual framework development
In order to guide the research process, we developed an initial conceptual framework by integrating the elements considered by the previous frameworks found in reshoring literature. Previous frameworks include a number of different elements of the reshoring process (see Table 1), primarily focusing on the decision-making and drivers. In addition, the implementation process Benstead et al. 2017), contingencies Benstead et al. 2017) and outcomes ) have also been acknowledged as important aspects. Thus, we incorporate all the identified aspects to provide a complete framework. Joubioux and Vanpoucke (2016) provide the only framework considering both offshoring and reshoring. However, they mainly focus on the offshoring activities while reshoring is given minor attention. Even if our purpose is to build a framework for reshoring, we model both offshoring and reshoring with similar processes, in order to understand their relationship and thereby fully understand the reshoring decision. The framework is depicted in Figure 1 and is comprised of the following key elements: decision making, implementation and outcome of the offshoring and reshoring processes, and contingency factors. As a starting point, we used the framework developed by Benstead et al. (2017), which we consider to be the most complete and intuitively accessible framework for the reshoring process in existing literature. However, in our proposed framework we also include the offshoring process, structured identically to reshoring. Reshoring is consequent to offshoring, in line with previous scholars stating that reshoring can only occur if offshoring has occurred previously (Gray et al. 2013). Moreover, in contrast to Following Foerstl et al. (2016), we also add an element related to the outcomes of the operations, in terms of benefits and/or problems (Johansson and Olhager 2018a). In terms of contingencies, we consider all the elements that may have influenced the offshoring and reshoring processes, but without being considered specifically, either in the decision making or in the implementation phase. In addition, we consider time to be an important element. Time has been identified by a number of studies as a relevant factor, since the conditions that made the offshoring decision rational might change over time in favor of domestic production and thus making reshoring a rational decision (Ellram et al. 2013; between the offshoring and reshoring decisions may shed some light on the question whether reshoring is a result of an offshoring failure or of a strategy change (Kinkel 2014;Ancarani et al. 2015;Di Mauro et al. 2018). Ancarani et al (2015) proposed a number of testable propositions regarding factors that affect the duration of the offshore operations, such as firm size and industry, reshoring motive and firm strategy. In Germany for example, the majority of firms reshored production within two to five years after the offshoring operations (Kinkel and Maloca 2009;Kinkel 2014). This implies that reshoring could be a correction of a prior offshoring miscalculation rather than a reaction to slowly emerging developments. In addition, time is relevant to consider in the implementation phase, as ramp up times were significantly longer than expected for the surveyed firms (Kinkel and Maloca 2009). Time is intrinsically part of the conceptual framework, with offshoring preceding reshoring and with the different elements of the two processes being represented as sequential.

Locating relevant research
In this study we aimed to analyze the two connected events, i.e. reshoring and previous offshoring, with offshoring being considered as an antecedent to reshoring. Thus, in order to identify all relevant literature for our study, we developed a search strategy aimed at finding case study research on the reshoring phenomenon, including descriptions of the preceding offshoring process. We retrieved the sample of potentially pertinent literature from Scopus and Web of Science as we focused on peer-reviewed sources with strong academic contributions that are relevant to synthesize, and not merely providing illustrative examples of manufacturing relocation. The chosen databases had also been used in previous literature reviews on reshoring (Stentoft et al. 2016b;Wiesmann et al. 2017; Barbieri et al. 2018). We discarded unpublished works in order to increase the scientific rigor, generated through a peer-reviewed publication process (Hoon 2013), and we avoided publication bias (Kepes et al. 2012) with a comprehensive and transparent search process (Hoon 2013). We used the search terms "reshoring" and "case study", as well as related terms (i.e. "backshoring", "nearshoring", "onshoring", "backreshoring", "backsourcing", "homeshoring", "inshoring", "reinsourcing"), as identified in previous literature. These were applied in the fields title/abstract/keywords in Scopus and topic in Web of Science. The initial sample consisted of 363 articles. Here, we applied a set of clearly specified selection criteria (complete list in Table 2) in order to ensure a high validity of the synthesis, which strongly depends on the quality of the primary studies on which it is based (Dalton and Dalton 2008). Particularly, we only included articles from peer-reviewed sources that used case-based methodologies, including both offshoring and subsequent reshoring decisions of manufacturing activities in specific. Thus, conference papers, articles focusing on other activities than manufacturing (e.g. services or IT) and studies not using case-based methodologies were excluded. After initial screening of titles and abstracts and applying the selection criteria, the sample consisted of 18 articles. This result was triangulated with available systematic literature reviews on the topic (Stentoft et al. 2016b;Wiesmann et al. 2017;Barbieri et al. 2018), and we found an additional 2 articles. Further, we added 3 articles that we were knowledgeable about from previous studies, and we reached a baseline sample consisting of 23 articles. In order to make sure that we had identified all the available reshoring cases in peer reviewed literature, we also run the same search without the keyword "case study" but no further articles were identified, proving the robustness of the search strategy.
Individually, both researchers read all the articles and made a preliminary coding based on the initial framework. This process resulted in another round of exclusions, as nine articles did not report enough information to be analyzed. Thus, the final synthesis sample includes 14 articles. All the articles were checked in terms of quality: all of them are based on the methodological standards widely accepted in OM and SCM fields (e.g. the ones provided by Eisenhardt (1989), Voss et al. (2002) and Yin (2009)).

Coding and analysis
We followed the suggestions by Durach et al. (2017) and developed a coding scheme in two levels; (i) first according to the articles' characteristics, and then (ii) following the main elements in our framework. In particular, for the first level coding the information was retrieved mainly from the introduction and methodology sections of the original articles. For the second level we specifically focused on the findings, discussion and conclusion sections (Hoon 2013). The coding and analysis were done iteratively in two phases, following the recommendations by Hoon (2013). First, we made an initial coding scheme based on the initial framework, aiming to retrieve information about contingencies, decision-making, implementation, outcomes and time. After the first round of coding, we refined the framework based on the evidences in the articles by adding details and new elements that we assessed as relevant. Then, we conducted a second round of coding using an updated coding scheme following the final framework.
To keep track of the coding process and to support the data analysis we used the software NVivo Plus 12. In order to avoid any potential bias along the process, two researchers were involved in all the phases and results were discussed after each phase with experienced researchers knowledgeable about the field of study. The two researchers coded the primary studies independently, reaching an interrater agreement of 83%. Then, the two databases were merged and contrasting coding results were discussed with a consensus resolution approach (Larsson 1993). When necessary, experienced researchers were involved as devil's advocates. Involving two coders has been proved to be beneficial for the reduction of mistakes and to avoid the omission of relevant information, as well as to contribute to achieve a better final coding scheme (i.e. the final framework) (Hoon 2013;Miles and Huberman 1994).
Concerning the data analysis, for the first level coding (i.e. characteristics of the articles) we analyzed the characteristics in terms of frequencies. For example, we counted the number of articles adopting different theoretical perspectives (e.g. Transaction Cost Economics, Resource Based View). Analysis of the second level coding was made first through the development of mind-maps for each article (Seuring and Gold 2012), that favored discussion among the researchers, as well as identification of patterns. The mindmaps were used to recall examples and illustrations described in section 5. Secondly, we evaluated the single article coverage of each code, as well as the presence of cross-coding (namely codes appearing together), with the software NVivo 12 Plus. Finally, as a result of the coding process, we were able to develop the final framework.

Meta-synthesis of case-based reshoring research
The studied articles are presented in Table 3

Characteristics of the articles
According to Durach et al. (2017), studies within SCM normally reflect different characteristics (e.g. theoretical perspectives, unit of analysis, research methods). This is true also for studies on reshoring, as demonstrated by Barbieri et al. (2018). This challenges the possibilities to compare studies and thus synthesize the empirical evidence within the field, without a structured method. Accordingly, we analyzed the sample articles to understand if we needed to apply a specific contextual lens when comparing results. Table 4 details the characteristics of the articles investigated in this study. Quite surprisingly, four out of fourteen studies did not mention the time span between offshoring and reshoring at all.

Analysis of the analyses and framework refinement
The second level coding scheme was applied to the analyses of the original authors, i.e.
the results, discussion and conclusions sections of the studied articles. The elements included in the final framework are displayed in Table 5 Additionally, within offshoring and reshoring implementation, we highlighted the preparation activities that were reported in some of the articles as an initial phase of the implementation process. Finally, we acknowledged, during the coding process, that contingencies, drivers and barriers for offshoring and reshoring included similar items.
Therefore, we recognized as useful to identify a set of factors that during time can shift between being contingencies versus being drivers or barriers for the offshoring and reshoring decisions. Each time we identified a contingency, driver or barrier, it was also coded with any of the factor categories, thus specified in more detail. The categorization of factors is inspired by Wiesmann et al. (2017), but the terms are adapted to fit our framework. As an example, we added a category related to firm specific factors at the offshore site, named offshore internal factors. In fact, we used the differentiation between internal and external factors, as suggested by Fratocchi et al. (2016). In the end, the factors categorization reflects the country (i.e., domestic or offshore), the level (i.e., global, supply chain or firm) and the orientation (i.e. internal or external).   Table 3. While no article discusses all the coded elements, the results show that all the elements in the framework have been discussed by at least two articles in our sample. Time and contingencies were widely discussed in all articles. The drivers for reshoring were also discussed by almost all the articles, which is not surprising since this was the focus in most articles. Interestingly, both the operations at the domestic and at the offshore locations were almost equally discussed. Preparation, implementation and outcome were only moderately discussed for both offshoring and reshoring. The least discussed element was the offshoring tipping point, while a reshoring tipping point was identified in half of the articles. The factors, reflecting either the contingencies or the relocation drivers and barriers were widely discussed. The most commonly cited factor was domestic internal, mentioned in almost all the articles. During the coding we did not find anything that could not be categorized with the coding scheme. Thus, we concluded that the coding scheme is complete, and the framework sufficiently covers everything that is reported in our studied articles.

Contribution per article (number of codes in text)
We also assessed the cross-coding (i.e. how often two codes appeared together), which was possible since we allowed to code the same sentence in a specific article with multiple codes. Table 6 show the percentage of cross-coding for each couple of codes, assessed as the ratio between the number of shared codes and the total number of codes for the least coded element of the two (i.e. the maximum number of shared codes possible to achieve).

Comprehensive framework with illustrations
Based on the results of the coding, we refined the framework as depicted in Figure 2. The framework is structured in a graphical way that makes it intuitively easy to follow, so to   Baraldi et al. 2018). In general, we find evidence that firms find it hard to make correct calculations to base the offshoring or reshoring decision on (Gray et al. 2017;Engström et al. 2018). However, experience from previous offshoring improves the assessment of cost and performance implications, thus implying that there is a positive learning effect. As part of the drivers, we specifically distinguish the tipping point, as it emerged as relevant in several of the studied articles.
For offshoring, the tipping points mentioned were requirement from a key customer (Baraldi et al. 2018) and bad profitability at the domestic site (Gylling et al. 2015). For reshoring, negative events at the offshore site (e.g. quality issues or IP violation) (Benstead et al. 2017;Gray et al. 2017), diminished cost differentials (Gylling et al. 2015;Benstead et al. 2017), free capacity at the domestic site (Nujen et al. 2018b), and a new business model (Robinson and Hsieh 2016) triggered the decision. In all of these cases the firms had more than one reason to relocate manufacturing. But each of these reasons were not important enough to drive the decision. Instead, the issues were building up and the firms had to reach a tipping point when one driver made the current situation unsustainable, leading to the decision to repatriate. Our study shows that there is an interplay by several factors, determining the optimal location of manufacturing activities.
Things or events can happen in several locations, i.e. at the domestic manufacturing plant, at the offshore plant, in the external environments in both locations (for example in the supplier or partner network, on national level, etc.), in the supply chain and/or in the global environment that influence the macro economic conditions. These events could happen one at a time or simultaneously. They are not static but change over time, and the longer the offshoring period, the less likely it is that these conditions have remained stable (Nujen and Halse 2017;Baraldi et al. 2018). Thus, drivers and barriers are dynamic and need to be reevaluated on a regular basis, as already pointed out by for example Ellram et al. (2013) and Tate et al. (2014). As part of the decision-making, we also include the product as well as the specific activities to be relocated. Interestingly, Di Mauro et al.
(2018) find that the product seems to be interrelated with the location for the offshore activities. The reason is that firms are searching for specific skills and competences . For reshoring, activities are fine sliced, which means that firms are reshoring very specific activities that fit within the current activities at home. In fact, reshoring was in many cases made partially, by keeping some production activities offshore. Baraldi et al. (2018) is referring to "selective reshoring", indicating that the firm specifically select the activities to repatriate, depending on how well they fit within the local operations.

Implementation
The implementation process of reshoring is highly unexplored, according to Bals et al. (2016). Among our reviewed articles, only Benstead et al. (2017) explicitly studied how the reshoring decision can be operationalized. We have followed their aspects of the implementation phase, with the only difference that we believe tipping point is part of the decision-making. Thus, we included aspects such as governance mode, degree of relocation, incremental or instantaneous process, maintaining production at the domestic site, information sharing, and preparation activities. Even though they did not explicitly study the implementation process, several studies briefly reported on related aspects. For example, in the studied papers offshoring is implemented gradually, starting with outsourcing, purchasing small batches, and then increasing the magnitude of offshore operations, maybe leading to a partnership or captive operations (Gylling et al. 2015;Di Mauro et al. 2018). It could even be hard to distinguish the boundaries between the decisions (Benstead et al. 2017). On the other hand, some firms are being flexible in terms of "degree of reshoring" (Gylling et al. 2015;Joubioux and Vanpoucke 2016;Benstead et al. 2017), thus still producing the same products they offshored at home, but in smaller volumes. This strategy provides flexibility and the possibility to shift volumes between locations. In terms of governance mode, offshoring is made in all forms (from outsourcing to joint ventures and fully owned factories). However, Joubioux and Vanpoucke (2016) found that the preferred governance mode for offshoring was outsourcing. Even though it requires extensive control mechanisms, outsourcing reduces the risks and provides flexibility as well as facilitates repatriation since it is considered relatively easy to end the relationship with a supplier. Ending a supplier relationship could be sensitive, though.
Re-outsourcing activities to the same supplier in the future could be problematic because of the damaged trust between partners (Engström et al. 2018;Nujen et al. 2018b).
Interestingly, three papers out of fourteen were specifically studying the organizational readiness for reshoring, concerning for example how to deal with the shortage of skills and competences. Thus, as part of the implementation process, a preparation phase has been identified. This is a new aspect compared to existing frameworks. As part of the preparation, the assessment of the organizational readiness for reshoring is of a crucial importance (Nujen et al. 2018b). In fact, over time, the previous knowledge base in the home country might be diminished, and neglecting to evaluate the access to skills and knowledge can turn out to be a fatal error for reshoring implementation (Nujen and Halse 2017). In such a context, the management role in identifying existing competences and developing dynamic capabilities becomes fundamental (Nujen et al. 2018a). Surprisingly, also articles with a focus far away from the preparation phase mention details that help in structuring its characteristics. Among others, some of the mentioned elements are the development of in-house training programs (Gylling et al. 2015), improving efficiency and freeing space (Engström et al. 2018), and of course organizational readiness (Nujen and Halse 2017;Nujen et al. 2018a, b). Naturally, after having found evidence of the reshoring preparation phase, we expected to find information also about the offshoring preparation. Even if this aspect was not widely discussed in the sample articles some details were mentioned, as the transfer of knowledge, competences and technology from the home to the host country (Gylling et al. 2015;Joubioux and Vanpoucke 2016;Di Mauro et al. 2018), the assessment of knowledge availability in the host country (Di Mauro et al. 2018), the preparation of managers to face offshoring challenges (Gray et al. 2017), and the management of the relationship setup with suppliers in case of outsourcing (Martínez-Mora and Merino 2014;Joubioux and Vanpoucke 2016). The offshoring preparation phase is an absolute novelty of our framework with respect to previous reshoring frameworks.

Outcomes
The offshoring and reshoring outcomes were included as new aspects compared to previous frameworks. The outcomes, categorized as problems or benefits, are highly interesting from a managerial perspective, as they could be used to derive the success factors of manufacturing relocation. But they are also relevant from a decision-making perspective, since the offshoring outcome in many cases act as a driver of reshoring. For example, insufficient quality at the offshore site is an offshoring outcome, but it is also a main driver of reshoring (Kinkel and Maloca 2009;Johansson and Olhager 2018b). This is true in survey studies as well as in several of the studied papers (Joubioux and Vanpoucke 2016;Gray et al. 2017;Baraldi et al. 2018;Di Mauro et al. 2018;Engström et al. 2018;Sayem et al. 2018). One question that has received quite a lot of attention is whether reshoring is reflecting a strategy change or if it is a correction of a previous offshoring mistake, i.e. if reshoring is a reaction to an offshoring failure (Fratocchi et al. 2016). From the studied papers, it is evident that reshoring could be both. For example, Gray et al. (2017) concluded that the reshoring decisions could not be completely explained by changes in relative costs between locations. Instead, they say that offshoring was made without completely evaluating risks and performance challenges, and that reshoring thus was a correction of a prior offshoring mistake. Similarly, Joubioux and Vanpoucke (2016) found that reshoring was implemented after continuous problems with quality at the host country site. On the other hand, in some studies the firms had explicitly stated that offshoring was not a failure (Martínez-Mora and Merino 2014; Di Mauro et al. 2018). Di Mauro et al. (2018) argue that the offshore duration (over ten years) in combination with the strategy change driving the reshoring decision, indicate that offshoring was not a failure, but that reshoring is done as a consequence of changing conditions. Similarly, in most of the studied papers the firms have had offshore operations for over ten years before the reshoring decision was taken. Thus, it is hard to argue that reshoring was made as a reaction to an offshoring mistake. However, it should be noted that the offshore duration also clearly affects the possibilities to repatriate manufacturing since the availability of skills and knowledge might have diminished over time (Nujen and Halse 2017;Nujen et al. 2018a). Interestingly, the outcomes of reshoring were not discussed widely in the studied articles. However, similarly to offshoring, reshoring success or failure could possibly influence the location strategy of a firm, thus influencing the trajectory of its global operations.

Contingency factors
Concerning the contingency factors, from the analyzed articles it was clear that many contingency factors play a role in influencing the offshoring and reshoring processes. the financial crisis in 2008 was a turbulent period for many firms, with low demand, free capacity and high uncertainty, leading many firms to consider reshoring. Home country and host country contingencies are included in domestic external and offshore external factors respectively (e.g. the presence of an industrial district, the industrial landscape, labor market or governmental incentives/regulations either in the home or in the offshore country). Finally, supply chain factors concern for example the type of relationships with suppliers and customers, the length of the supply chain and the position in the supply chain. Therefore, contrarily to what was expected from the initial framework, we found that contingencies were spread all over our final framework. Moreover, we propose that every part of the framework can become contingent to the others. A clear example is represented by the debate on the relationship between offshoring and reshoring. Many studies claims that reshoring is strongly influenced by the previous offshoring decision, so much that it is worthwhile to study them together (Joubioux and Vanpoucke 2016; considered a contingency factor influencing how the reshoring decision is managed (Benstead et al. 2017). As an example, Di Mauro et al. (2018) argue that a captive offshoring mode makes firms less inclined to repatriate operations. In conclusion, contingencies result to be spread all over the final framework and to play a changing role over time, depending on the specific decision. This reveals the challenge to control for all the contingencies that might influence reshoring.

What do we want to know about reshoring?
The meta-synthesis of previous reshoring cases provides an accumulation of the knowledge collectively produced within this field, thus giving evidence of "what do we already know about reshoring?". Moreover, the analysis of the articles' coverage (Table   5), showed many areas that were not deeply studied and thus emerged as potential areas for future research. In this section, we suggest some directions for future research and answer the question "what do we want to know about reshoring?".
1. Decision making process: Even though many previous studies focus specifically on the reshoring decision-making drivers (Wiesmann et al. 2017;Barbieri et al. 2018), the actual decision-making process is not well documented. It would be highly interesting, especially in light of the discussion about offshoring possibly being a managerial mistake, to investigate how the decision-making is carried out within an organization and who is involved in the process. When the relocation decision is considered a managerial mistake, what was overlooked in the decisionmaking and how could the process be improved?
2. Tipping point: the tipping point or trigger of the reshoring decision has just recently been acknowledged in academia (Benstead et al. 2017;Hartman et al. 2017 In addition, the cross-coding (Table 6)

Conclusions
In this study the meta-synthesis methodology was used to accumulate the knowledge produced within the field of reshoring, by studying the empirical evidence in prevailing case studies. Based on the meta-synthesis, a comprehensive framework for describing and studying reshoring was developed. The framework is based on previous reshoring frameworks found in literature, including all aspects of the reshoring process that has previously been identified. Thus, it is argued to be more comprehensive than previous reshoring frameworks as it encompasses all aspects from decision-making, through implementation and outcomes, as well as contingency factors and the time aspect making location decisions dynamic over time. In addition to the reshoring process, the framework also includes the previous offshoring process, since previous studies have indicated that reshoring can only be fully understood in light of the preceding offshoring. This represents one of the first efforts in guiding future reshoring research to a full understanding of the reshoring phenomenon, and to make more comparable studies, thus overcoming the "contingencies control" challenge. This paper clearly demonstrates that even if articles are grounded on different characteristics in terms of theoretical perspectives, units of analysis etc., the information can be rich enough to allow a metasynthesis to be successful. In fact, we succeeded in retrieving useful pieces of information from each article. This allowed to accumulate the existing knowledge within the field and to create a comprehensive framework for future studies, as suggested by Goldsby and Autry (2011) and Hoon (2013).

Implication for research and practice
The main contribution of this study is a comprehensive framework for research on reshoring, providing guidance on how to present a complete case description in order to enable advancement of the accumulated knowledge within the field. In doing so, the article addressed the main limitations of previous frameworks, that give only a partial picture of reshoring. The study demonstrated the benefit of applying the meta-synthesis methodology to advance and complement the results achieved from the single articles in the field of reshoring. The implications for managers are connected to the possibility to gain a wider picture over the reshoring process as well as the preceding offshoring process, and to gain useful insights about the dynamic nature of such processes and of the variety of challenges that previous cases had to face in multiple contexts. This study thus provides an important contribution for companies that are relocating manufacturing.

Limitations and suggestions for further research
The paper does not come without limitations. First, the meta-synthesis method relies on secondary data, namely the information provided in the original articles. Of course, we are aware of the fact that the authors of such articles might have limited space to present the relevant information and discuss it; therefore, authors might have favored those evidences that allowed them to answer to their specific research questions, as well as they may have discussed them by adopting specific theoretical lens. Second, the choice to limit the articles to peer reviewed sources allowed to control for the quality of the studies, but at the same time it might have caused the exclusion of some interesting cases. In this case, we evaluated to have a fair number of articles and we tried to avoid the publication bias by including a wide array of journals, without choosing to limit our analysis to top journals articles. As the quality of the meta-synthesis depends on the quality of the primary studies (Dalton and Dalton 2008), similarly the limitations of the meta-synthesis might be constrained by the limitations of the primary studies. Fortunately, we are confident in having addressed the main concerns from the original studies, namely context bias (either concerning country, industry or size), by choosing meta-synthesis as method, that allowed to convey results from multiple articles characterized by multiple contexts. Of course, we are aware that most of the studies were conducted from the European perspective, given that most of the authors were from Europe. Interestingly, we also realized that the articles in our sample were complementary in a way, since some of the limitations raised by certain articles were overcome by others, often becoming their main contributions. Instead, one limitation that certainly persist in our study is the lack of statistical generalizability. Nevertheless, the theoretical generalizability is made stronger by the lack of researcher or study biases. Future researches, besides trying to overcome the hereby presented limitations, could focus on the application and testing of the proposed framework in real cases, as well as on quantitative data collection.