How do external openness and R&D activity influence open innovation management and the potential contribution of social media in the tourism and hospitality industry?

This research focuses on how the tourism and hospitality industry is applying the paradigm of open innovation, supported by social media. Open innovation (OI) has been on the topical research agenda, but the previous literature lacks studies in the service sector and specifically for tourism companies. Moreover, the introduction of social media as a tool to implement open innovation is considered the main research gap. Structural equation modelling is applied to data from 181 Portuguese and Spanish companies to study both customer involvement in new product development and the perceptions and results in terms of turnover and competitiveness. The structure of the relationships between research and development, external openness and open innovation management is highlighted with statistical analysis. In addition, the introduction of social media adds value to the proposed model. Currently, there is a lack of available models to give structure to the OI paradigm and to allow us to manage it. The contribution of this research is a comparison of the explanatory power of three models that allow us to test how certain strategic guidelines in tourism companies influence each other and have a mediating or multiplier effect when linked to each other (nested models method). In conclusion, the originality of this research is based on the relationship between open innovation management and social media and the mediating effect of external openness.


Introduction
The concept of open innovation, introduced by Chesbrough (2003), has attracted the attention of scholars and companies as part of an emerging paradigm (Carroll and Helfert 2015). As defined by Chesbrough (2003, p. 43), "Open innovation means that valuable ideas can come from inside places, external ideas and external paths to the market on the same level of importance as that reserved for internal ideas and paths…". Thus, OI means opening the organization and generating spaces of collaboration with stakeholders to introduce external ideas and projects as part of strategy itself and to find a balance between inbound and outbound capabilities to create opportunities and business value.
In the last decade, the production sector has begun to progressively incorporate this ideology into its innovation strategy at the same time that a number of investigations have been undertaken in this field (Hossain and Anees-ur-Rehman 2015). Since then, open innovation has continued to be one of the most discussed issues in the field of management.
Open innovation implies the opening of a company to the exterior world to increase its performance and obtain a differential value in the market. In reviewing the state of the art and future perspectives, two research works are prominent. First is the study by Hossain and Anees-ur-Rehman (2015), in which the most frequently used techniques are identified. Qualitative techniques predominate, followed by regression analysis (OLS, tobit, probit, binomial), and in recent years, despite their limited presence in the literature, structural equation models. Second, Huizingh (2011) concentrates on identifying the nuances of the open Innovation concept to investigate the dependency contexts of this formula of innovation management and to deepen its implementation process.
Most of the research on open innovation has focused on the industrial sector, and case studies have commonly been used to demonstrate the implementation of this phenomenon. This study approach highlights the need for quantitative studies involving companies across sectors and countries (Huizingh 2011), especially in areas of activity, such as tourism, that are under-represented in the literature (Hossain and Anees-ur-Rehman 2015). Thus, this research responds both to the lack of empirical evidence on innovation and tourism (Cagica Carvalho and Sarkar 2014) and to the shortage of available empirical evidence for the study of open innovation as a formula of differentiation in the sector.
The tourism and hospitality industry in southern Portugal and Spain serves as a stage to learn how to implement open innovation with the support of social media as a facilitator. Furthermore, tourism companies from both countries allowed us to investigate the influence of certain strategic orientations in the management models applied. The results show the tourism sector approach to this new vision of innovation and, more important, whether an improvement in competitiveness is perceived as a consequence of it. In a complementary manner, the novel approach of incorporating social media as a technological tool capable of generating co-creation spaces with consumers (Bugshan 2015) should be emphasized. Therefore, the main objective is to understand the relationship between external openness and R&D activity in the implementation of open innovation and to analyse the current influence of social media as a source of innovation management. Hossain and Anees-ur-Rehman (2015) consider that it is convenient to continue developing quantitative studies with complex statistical support, such as structural equation modelling (SEM), to move forward and achieve truly practical conclusions for companies. Setting up alternative scenarios to assess how different variables lead to different results and behaviours provides support for the decision-making process. In this paper, a sample of tourism companies from southern Portugal and Spain is used, and a structural equation model is applied. Comparisons between three different models are carried out according to the SEM nested models method. The study contributes to the literature by both solving the problem of establishing international comparisons and examining differentiated behaviour in a specific sector-in this case, tourism. In comparing the explanatory power of three different models, each model introduces and links significant aspects of strategic guidelines for management. It has been demonstrated that the mediating effect of external openness on R&D activity can help explain the open innovation phenomenon in the tourism sector. In short, this paper provides a broader understanding of this paradigm in the service sector and reinforces the role of social media as a support in implementing it. In contrast to previous studies, the comparison of models better explains how the involved variables behave and the structure of the relationship between them. In summary, the defining model explains that there is a logical link between companies that are aware of R&D and their external openness, which is precisely what introduces open innovation in a natural way. The main insight is that social media is a key tool in implementing this approach.

Open innovation in tourism
Innovation has emerged as a driver of economic growth and prosperity in several countries (OECD 2015), and, at a micro-level, innovation is a source of competitive advantage for companies (Cagica Carvalho and Sarkar 2014). It is also considered a key source for improving the output performance of the service sector (Zhao et al. 2016).
Given the growing importance of the service sector in the economies of developed countries, interest in analysing innovation in this area has increased (Chen et al. 2016). Although different perspectives on the subject coexist, there has been no discussion of the special characteristics of this sector. Innovation in the tourism sector can be considered a continuous challenge due to the need for companies to adapt to the new profile of tourists (Stamboulis and Skayannis 2003) as more demanding, better informed and having achieved a level of empowerment that forces companies to adapt continuously to these changes (Chiang and Hung 2010;Voorberg et al. 2015). Consequently, the tourism sector should offer unique and innovative experiences (Weiermair 2006) that connect them better with the market and maintain companies' competitive standing.
The literature that focuses on innovation in the tourism sector is insufficient, as innovation is considered a pending field of development in this industry (Hjalager 2010: Gomezelj 2016Marasco et al. 2018;Nordli 2017). The most notable gap is the lack of evidence on the innovation behaviour of tourism enterprises compared to the industrial and services sectors in general (Cagica Carvalho and Sarkar 2014). This lack of evidence is especially clear if we focus on open innovation in this sector. Most of the research has focused on the broader services industry (Hossain and Anees-ur-Rehman 2015;Huizingh 2011), and the implementation of the open innovation paradigm has been investigated mainly in large companies (Xiaobao et al. 2013), whereas research on SMEs (small and medium enterprises) is scarce (Van de Vrande et al. 2009;Wynarczyk 2013). Furthermore, the analysis of open innovation in tourism is practically non-existent (Hossain and Anees-ur-Rehman 2015). Recent studies have begun to detect the influence of customer participation on innovation as well as the need to establish platforms for the exchange of ideas and to create collaborative networks (Binkhorst and Den Dekker 2009;Croft 2016;Marasco et al. 2018).
Open innovation widens the field for in-depth investigation-especially because it fits into the scope of tourism as much from an academic as from a business perspective. Chesbrough (2003) introduced the term to refer to the permeability of a company to stakeholders as a source of differentiation. Open innovation refers to opening up the organization and creating spaces for collaboration with clients, suppliers and other stakeholders, making possible the introduction of ideas and projects from the outside as part of the strategy. However, this approach requires finding a balance between internal (inbound) and external (outbound) capabilities to create opportunities and business value. Currently, open innovation has penetrated the innovation strategy of companies but continues to be considered an emerging paradigm (Carroll and Helfert 2015) and is one of the most debated topics in the field of management. Currently, open innovation continues to cause controversy because over a brief period, several terms with points in common-open innovation, crowdsourcing and co-creation-have emerged, and it is not easy to differentiate between them (Egger et al. 2014). In a simplified form, as already mentioned, open innovation is a new paradigm for understanding innovation, and its implementation takes many forms, including stress crowdsourcing and co-creation. Crowdsourcing consists of the resolution of a problem through undertaking a task of variable complexity and modularity through the voluntary contributions of stakeholders' work, money (in the case of crowdfunding), and knowledge and/or experience (Estelles Arolas and González-Ladrón-De-Guevara 2012). Co-creation, an interesting source of innovation, can be defined as a result of customers contributing their experiences with products and services (Voorberg et al. 2015). Any stakeholder can contribute to innovation and new product/service development.
Tourism companies cannot develop R&D processes in the strict sense, as the industrial sector can. Thus, it is more common to find references to innovation than references to R&D in the tourism literature because of the peculiarities of this economic activity. According to Cagica Carvalho and Sakar (2014, p. 156), "tourism is not just based on the production of goods or services. Several intangible characteristics are embodied in individuals"; thus, the question of how to innovate creates difficulties. In any event, the development of new services (NPS) and the need for continuous adaptation are interesting not only for researchers but also for international organizations that try to measure innovation in this sector (Hall and Williams 2008;Hjalager 2010;Nordli 2017). Furthermore, the strategic value of competitive R&D and innovation in the service sector, and specifically in tourism, is undisputed (Gomezelj 2016). A company's level of innovation-understood as new services and products as well as new technologies and new processes for improving the overall level of servicemay be conditioned by its capacity to open itself to external sources, especially when economic activity develops in what are called dependent contexts. There are numerous studies that consider external openness a natural consequence for companies in this situation (Cheng and Huizingh 2014). For example, some recent studies in this field have highlighted that R&D in tourism implies that customer requirements should be incorporated into companies in order to have an impact on long-run growth (Albaladejo and Martínez-García 2015). Consequently, companies are beginning to engage in this dynamic process as a basis for value and future innovation (Binkhorst and Den Dekker 2009). Therefore, the connections between R&D and new approaches to innovation management should be examined.
External openness is the starting point for OI implementation. Openness "is a search strategy involving external channels of information that are used to innovate" (Wu et al. 2013, p. 705). For its part, external openness is a result of the strong conviction that market orientation is a key factor for promoting innovation in companies (Atuahene-Gima and Ko 2001; Teirlinck and Spithoven 2013).
The possibility that the internal and external elements of an enterprise moderate the effect of open innovation has begun to be treated in a differentiated way so that the orientation of a strategy, the location, the sector and the context of the development of the economic activity acquire some relevance (Huizingh 2011). This leads us to emphasize the value of studying sectors of activity in a disaggregated way and to incorporate empirical studies, such as this one, which does not focus on the overall industry but extends the vision of open innovation to the service field. The relevance of these issues has been emphasized in previous studies, such as that of Laursen and Salter (2006), who found that open innovation is affected in different ways according to the sector and its specific technological context. Indeed, this point highlights the possible contribution of this research work because the analysis of open innovation in the service sector, and specifically in tourism, is under-represented.
Emphasizing the shortage of investigations on open innovation and tourism, companies in the sector have begun to develop a special sensitivity to generating proposals with the collaboration of consumers (Von Hippel 2005). The consequence is the active stimulation of co-creation environments from which new solutions and ideas can be derived (Abbate and Coppolino 2011). In addition to the positive impact of the opening of an enterprise on its level of innovation, other positive effects are improved awareness, connectivity and reputation (Hossain and Kauranen 2016) and even the achievement of competitive advantage (Ernst and Brem 2017).

Relationship between R&D and external openness
The R&D achievements of a company are not always derived from the adoption of open innovation (Schroll and Mild 2012). However, it seems logical that companies with a high level of innovation are more predisposed to opening to the outside and therefore rely on the open innovation model. The literature on the subject presents contradictory scenarios. According to Keupp and Gassmann (2009), when internal R&D is intense and well organized, open innovation is not as relevant. Studies such as Segarra-Ciprés et al. (2014) have demonstrated that the incorporation of R&D into a company's innovation strategy is moderated if the company already has sufficient capabilities (Grimpe and Kaiser 2010). Following this line of argument, openness takes place precisely to alleviate deficiencies of resources, skills and knowledge; i.e., permeability provides a company with innovations that it would not be able to develop and implement alone (Grimpe and Kaiser 2010;Keupp and Gassmann 2009;Spithoven et al. 2013). The direct relationship between increased R&D activity and the adoption of open innovation is the basis for a second group of studies (Cheng and Huizingh 2014). Evidence of the positive effect on the degree of innovation of companies adopting a market and entrepreneurship orientation has been collected in studies such as Atuahene- Gima and Ko (2001) and Teirlinck and Spithoven (2013). In short, openness is more a strategic issue for enterprises than a matter of industry trends (Keupp and Gassmann 2009, p. 338). These ideas are emphasized in this paper, and the first hypothesis is based on them.
H1. Tourism enterprises with greater investment in R&D and/or innovation are characterized by a higher degree of external openness (EO).

Implementation of open innovation
Further refining the focus requires links to be made between external openness and the implementation of open innovation in a company. The concept of openness is doubtless the basis of the emerging innovation paradigm (Von Hippel 2005). Cheng and Huizingh (2014) show that if a company is able to empower open innovation as part of its strategy, it obtains better results in terms of product and service innovation as well as financial and consumer performance. In this line, studies demonstrating the positive effect of open innovation on innovation performance have multiplied in the last decade. Most of the investigations concentrate on demonstrating how the combination of internal and external sources has a positive impact on enterprise performance (Wang et al. 2015;Zhao et al. 2016).
Nonetheless, opening to the market through open innovation involves difficulties. Each channel of collaboration requires the company to develop effective processes and practices (Laursen 2011). In this sense, open innovation can become a competitive weapon if the company establishes protocols for collecting inputs from the outside and, through a series of management decisions, enables these inputs to produce significant outputs (innovation in services). Thus, Laursen and Salter (2006, p. 6) assert that "firms require the capability to absorb new ideas from external sources and then integrate them into their internal processes to achieve an innovation". The need for systematization leads us to consider the implementation of open innovation in tourism from the perspective of management, as open innovation is widely recognized in the literature as being key for strategic management and as a preliminary step towards improving performance (Lichtenthaler and Lichtenthaler 2009). This decision makes it possible to examine not only whether companies consider the open innovation philosophy positive but also how they apply it; in addition, it allows us to expand a less developed area in the literature on the subject. In this sense, a hypothesis that links open innovation with management fills one of the gaps often indicated by management scholars by investigating how tourism enterprises implement open innovation in practice.
H2. The degree of external openness (EO) of tourism companies has a positive effect on the implementation of open innovation (OI) from the point of view of management.

Particularities of R&D activity in tourism and open innovation
As discussed above, tourism has some particularities in terms of R&D activity. So far, few studies have examined the relationship between this strategic area and the effective implementation of an open innovation management model, which this study seeks to investigate. Weiermair (2006) specifies the positive effects of innovation on reducing production costs, enhancing marketing and providing product value. The tourism sector has a number of factors, such as high competitiveness and changing scenarios (Najda-Janoszka and Kopera 2014) as well as the continuous demand for new experiences and tourist empowerment (Sigala 2012), that encourage innovative activity.
Therefore, the level of innovation demanded by the tourism industry favours the adoption of open innovation, as the following hypothesis proposes: H3. There is a direct relationship between the R&D activity developed by a tourism company and its level of implementation of open innovation (OI). Vanhaverbeke et al. (2008) affirm that a prerequisite for a company to implement open innovation is to progressively increase the capacity to absorb and integrate the ideas and proposals of different key agents into the internal intelligence system. However, there is some debate about the motivations that lead an enterprise to implement this paradigm of innovation. While some consider that adopting this paradigm is an offensive strategy to maintain competitiveness levels and generate a differential advantage ( Van de Vrande et al. 2009), others argue that it is a consequence of the weaknesses of the company in developing its own innovation processes (Keupp and Gassmann 2009;Laursen and Salter 2006) or simply an effective formula for responding to the costs and financial risks of internal R&D (Chesbrough 2012). Whatever the motivation, it necessarily implies opening up to the outside environment and being receptive to the suggestions and proposals of different stakeholders, that is, applying open innovation to achieve a higher level of innovation.

The potential of social media for customer involvement
In recent years, social media has indisputably penetrated the Internet and increased connectivity between people. The power of the customer has increased as a result of the multiplication of online purchase options, available information, etc. (Hays et al. 2013). The influence of social media on the tourism field is especially remarkable because travel-related decisions can be influenced by others. People search for information and organize their trips, as well as commenting on and sharing experiences with others, on their personal networks, in online communities, or by using other collaborative tools (Leung et al. 2013;Wozniak et al. 2017). This is precisely where customer involvement comes in through a company's interaction with customers to achieve better outcomes at different stages of the new service development process (Carbonell et al. 2009;Füller et al. 2009). Despite the increase in research on this topic, a review of the literature suggests that there is little empirical evidence about the effectiveness and outcomes of interacting with customers, even in studies with less focus on social media as a co-creation environment to achieve such interaction (Sigala and Chalkiti 2014;Carbonell et al. 2009;Wozniak et al. 2017). Not only do such opportunities benefit consumers, but a tourism enterprise is open to a wide range of options if it participates in and manages web 2.0 platforms (Binkhorst and Den Dekker 2009). Social media can support both the marketing department (Hvass and Munar 2012) and the R&D department of a company. Social media tools make it possible to effectively apply a market orientation strategy (Jiménez-Zarco et al. 2011; Sigala 2012) because they act as an ICT element and provide a solid basis for the incremental innovation of any service. Ultimately, the tourism sector could take advantage of co-creation via social media (Hays et al. 2013;Kietzmann et al. 2011;Sigala and Chalkiti 2014). However, only very limited empirical work has attempted to examine the return on investment (ROI) of social media use in tourism organizations (Wozniak et al. 2017). Greco et al. (2015) highlight that open innovation is often analysed from a theoretical point of view and is much less frequently viewed from the perspective of management which is-understood as the guidance and control of the actions required to put innovation in the company into practice. This lack of research invites researchers and practitioners to delve deeper. Moreover, particular attention has been paid to social media as a support for innovation management due to its potential as an interactive tool (Kaplan and Haenlein 2010). Laursen and Salter (2006) find that the implementation of open innovation changes in terms of how inspiration is obtained to develop new products, and technological support facilitates meeting this challenge. In addition, the emergence of platforms such as TripAdvisor and Lonely Planet have required companies to be aware of the active and intensive use of social media by tourists. According to Chesbrough (2012), it is important to create platforms, architecture and systems that allow for openness to the outside. The nature of social media makes it an effective tool for developing effective co-creation environments (Abbate and Coppolino 2011;Hays et al. 2013). Customer involvement is recognized as a strategic practice for innovating, especially in its contributions to new product or service development (Sigala 2012). Although further research is needed, attention has become more focused on these topics in recent years. Despite controversies about the limitations of customer contributions to innovation, web 2.0 seems to be a key tool for enabling collaboration (Pitta and Fowler 2005)-so much so that the virtual environment, specifically social media, has made it possible for the customer to play an active role in new product development (NPD)/NSD (Füller et al. 2009).
It is worth pausing to examine the concept of user-generated content (UGC). Blogs, web forums, wikis, bookmarking sites, and photo and video sharing communities, as well as social media platforms for tourists and companies, should be considered in order to achieve the effective implementation of Open Innovation. Currently, any consumer, but especially those in the tourism field, can create, modify, share and discuss Internet content, and this content is often more trusted than the official websites of destinations or tourism companies (Kaplan and Haenlein 2010;Kietzmann et al. 2011). This could be considered a chance to create co-creation channels that allow tourism agents to put the OI paradigm into practice (Binkhorst andDen Dekker 2009, Jabreel et al. 2017). The use of social media in the tourism sector has been widespread, as is evident in the low residual percentage of companies that have not engaged with such platform (2%) (WTTC 2019).
However, in the open innovation literature, social media is not perceived as a dynamic and interactive openness channel (Bugshan 2015). Although the research focusing on tourism and social media has multiplied in the last decade (Mkono and Tribe 2017), it has focused mainly on social media and user-generated content from the perspective of marketing (Hvass and Munar 2012;Jabreel et al. 2017) and has not addressed how social media connects with open innovation. Social media in tourism has started to attract the attention of scholars, but authors such as Leung et al. (2013) in their review of the literature have concluded that the research on this subject is still far from maturity. It appears that this field continues to lack empirical evidence that allows a better understanding of social media platforms and usage related to tourism (Hays et al. 2013;Minazzi 2015;Wozniak et al. 2017). On this theoretical basis, this study analyses the influence of social media, especially in relation to how companies establish relationships with customers through a particular channel. The degree of openness is evidenced by established collaboration with the outside, for example, the number of external knowledge sources incorporated, as well as the intensity of collaboration with each of these sources (Laursen and Salter 2006).
The literature shows that the use of different Internet tools improves market intelligence, so to some extent, these platforms allow the establishment of continuous communication between consumers and enterprises. Social media allows the implementation of a market orientation strategy (Jiménez-Zarco et al. 2011;Sigala 2012;Pitta and Fowler 2005) and is an excellent way to take advantage of collective intelligence that could have a strong impact on the innovative activity of a company (Bugshan 2015;Ernst and Brem 2017). The theoretical framework is finalized with the proposal of a hypothesis regarding social media as a key element for the openness of tourism companies: H4. The number of social media (SM) platforms from which the company manages to incorporate ideas from the outside influences the level of external openness (EO) and increases the implementation of open innovation. Figure 1 shows the different models as well as the relationships between the R&D, external openness (EO), open innovation management (OIM) and social media (SM) variables, reflecting the hypotheses and providing a basis for the proposed SEM analysis to validate the variables. It should be stressed that the comparison between models is adopted because the models seem complementary at the empirical level, and it allows us to pursue a competing models strategy. According to Hair et al. (2014, p. 542), it is necessary to use "a modeling strategy that compares the proposed model with a number of alternative models in an attempt to demonstrate that no better-fitting model exists. This approach is particularly relevant in structural equation modeling, because a model can be shown only to have acceptable fit, but acceptable fit alone does not guarantee that another model will not fit better or equally well". For this reason, the models are compared following the SEM nested models method because "a powerful test of alternative models is to compare models of similar complexity, yet representing varying theoretical relationships. A common approach is through nested models, where a model is nested within another model if it contains the same number of variables and can be formed from the other model by altering the relationships, such as either adding or deleting paths" (Hair et al. 2014, p. 587) Model 1 recognizes the influence of R&D on EO (H1) and of EO on open innovation management (H2) as well as the direct effect between R&D and OIM (H3). In Model 2, the only variation is that the direct influence of R&D on OIM (H3) is not considered, but an influence modulated by the degree of openness is suggested. Finally, Model 3 incorporates social media (H4) as a support tool for openness. How do external openness and R&D activity influence open…

Variables and measuring instrument
The objective of this research is to understand how companies from southern Portugal and Spain face innovation management by testing a number of specific variables, such as permeability to external openness and level of assimilation of open innovation. The survey was modelled on three main studies. Atuahene-Gima and Ko's (2001) study was used to measure the orientation of enterprises to the market and to entrepreneurship (innovation), while to conceptualize open innovation, the main reference was the study of Laursen and Salter (2006). The items related to the company's R&D activity were derived from the proposal of the OECD (2015).
The survey was structured in five blocks. The business community generally recognizes the benefits derived from customer involvement, but specific mechanisms are not always put in place to make such involvement possible. The reason is based on the focus on open innovation from a management perspective. Thus, specific questions were included that show the degree to which there are mechanisms, measurement indicators, incentives for participation, feedback on the best proposals, etc. • Social media (SM). According to the proposal of Laursen and Salter (2006), which uses the concepts "breadth" and "depth", the survey asked about the social media used by the companies of the sample. On this theoretical basis, more sources of openness to the outside (breadth), in our case restricted to social media, indicate a greater level of implementation of open innovation. For the second concept, depth, given that it measures the intensity of use of social media, we used the previous block of questions related to open innovation management. In any event, the "depth" variable was chosen based only on Laursen and Salter (2006), Keupp and Gassmann (2009) and Chiang and Hung (2010). Although the variable "breadth" is often referred to, it was not suitable for this study because the focus was on how open innovation is perceived and developed by the sample companies, and individual perceptions in companies could introduce a bias in the results.
For the survey responses, a Likert scale was used with values between 1 and 7, with 1 representing the lowest degree and 7 representing the highest degree of agreement, importance or implementation. To clarify the specific meaning of the values linked to the questions, a clarification of their statement was included for questions 4-16. The respondents selected a number on a scale from 1 to 7 for all survey questions, as shown in Table 1. The questionnaire also provided an explanation of open innovation to ensure the companies' understanding. The definition was based on Chesbrough (2003), in which open innovation is understood as the use of purposive inflows and outflows of knowledge to accelerate the level of innovation in the company and to engage more with stakeholders, especially tourists.
Additionally, the validity of the questionnaire items was checked by collaboration with the most representative associations in the tourism sector in southern Portugal and Spain. 1 This pilot test was useful for enhancing the respondents' understanding of the questions and adapting some terms to be more suitable for tourism.

Processing methods and data collection
The analysis of open innovation in the tourism and hospitality industry in the case of Spain and Portugal is justified by the representativeness of this sector in global economic activity. According to the OECD (2015), both are among the European countries that are most dependent on tourism; in both cases, the contribution of tourism and hospitality to GDP exceeds 15%, and this industry represents 15.2% and 18.4% of national employment for Spain and Portugal, respectively (World Travel Tourism Council-WTTC 2019). Notably, 80.6% of the sample companies invest in tourism innovation, and approximately 6% of their turnover is dedicated to R&D. In disaggregated terms, the Algarve in Portugal and the Costa del Sol in Spain are areas of great affluence and overall tourist impact on these countries. The Algarve is the second most popular destination in Portugal by number of tourists and economic activity generated (Instituto Nacional de Estatística 2016). Additionally, southern Spain occupies second place in the national ranking, and the Costa del Sol, a region of the Andalusian autonomous community, contributes the most to tourism in terms of number of travellers and turnover (Instituto Nacional de Estadística 2016). It is worth mentioning the two main aspects of the contribution of this study: the multicountry approach and the choice of the tourism sector. The study of the phenomenon of open innovation in two countries not only allows a comparison that adds value to the research but also follows the suggestions for future lines of research proposed in previous studies. Thus far, the percentage of multi-country investigations among the countries chosen for analysis remains comparatively low. However, above all else, the study tests how open innovation is implemented in the tourism sector and seeks links between the awareness of R&D and an openness strategy in this approach. Additionally, social media is investigated to gain a better understanding of the new scenarios of stakeholder relationship management in tourism. There are no precedents of a multi-country perspective in the field of open innovation, specifically in tourism, so this perspective should be highlighted as a main insight.
Open innovation has previously been examined in studies that limit their investigation to only one country-estimated to account for about 26% of the total research work, according to the comprehensive literature review conducted by Hossain and Anees-ur-Rehman (2015). Furthermore, according to the results of the review by Hossain and Anees-ur-Rehman (2015), the Web of Science database contains no studies on open innovation focused on tourism. As a result, the contribution of this research can is an overview of the implementation of open innovation in a sector that is under-represented in the literature and that, because of the impact of social media, is especially interesting in terms of the adoption of this paradigm (Huizingh 2011).
The questionnaire described above was distributed via e-mail, with a link to an online survey included in the body of the message. We requested that a general manager or R&D manager respond to the questionnaire. To identify the sample group, we listed companies operating in the sector in each objective region on the basis of official statistics from Spain and Portugal; the public boards of tourism companies, especially on official tourism websites; and information provided by many (sectoral) business associations.
Data were collected between 2016 and the beginning of 2017. Data were collected in 2016, and the information was analysed over a period of 4 months. In view of the response rate, in February 2017, we decided that a second wave of questionnaires was necessary to improve the results and to enhance the statistical analysis.
In the first phase, 135 responses were obtained, and in the second phase, 46 were received. The second phase was conducted to address possible non-response bias. This methodological decision was made under the assumptions of simple random sampling, which assumes an error of 7.3% with a confidence level of 95% and assumes maximum indeterminacy (p = q = 0.50). In summary, the sample a priori consisted of 347 companies; since 181 companies answered, and the non-response percentage was 47.85%.
To evaluate the non-response bias, the possible differences between the respondents in the first phase and the second phase were investigated based on the assumption that the respondents in the second phase had more in common with the nonrespondents than with the respondents in the first phase (Armstrong and Overton 1977). In particular, the variables that characterized the companies were investigated to determine whether the enterprises that did not answer were different from those that did. The differences between the two phases were investigated with respect to the qualitative variables, nationality and activity of the company, and the quantitative variables, turnover and number of social media platforms. The Chi squared test of homogeneity was used for the former variables and confirmed that both samples showed homogeneous behaviours (p-values 0.337 for activity and 0.087 for nationality). For the latter variables, the ANOVA test showed that the characteristics of the companies that answered in the first phase and the second phase did not vary in turnover (p-value 0.586) or in the number of social media platforms used (p-value 0.723). In short, the lack of evidence of non-response bias was demonstrated in the way that these tests are usually performed in the literature (Kidwell and Fish 2007).
Regarding the structure of the sample with respect to activity, the majority of the sample was tourist lodges (37.8%), followed by enterprises organizing cultural or nautical tourism activities at the destination as active tourism companies (14.8%). Travel agencies and companies dedicated to thematic tourism (cultural tourism, senior tourism, health tourism and ecotourism) represented nearly 12%, closely followed by tourism facilities such as spas, health and beauty centres, golf resorts and marinas, which made up 11.1%. Regarding nationality, 65.7% of the companies managed establishments in southern Spain, which was proportional to the larger business dimension of this area. Regarding the size of the enterprises, small companies predominated, and there was a significant representation of SMEs with limited economic results; specifically, 36% of the companies had a turnover of up to € 250,000, and only 6% exceeded € 7,500,000.

Results
The empirical comparison of the formulated hypotheses of this investigation was carried out using structural equation modelling (SEM) to analyse the data collected from the survey. The technique most often used for the study of open innovation is regression analysis of different types (OLS, tobit, probit, binomial, etc.), followed by structural equation modelling, which, due to its advantages for this type of analysis, has displaced other statistical techniques in recent years (Hossain and Aneesur-Rehman 2015). The regression models in a structural equation model are less restrictive, allowing us to include measurement errors in both the dependent variables and the independent variables, particularly, because of the ability to measure the direct and indirect effects among factors. Moreover, it should be emphasized that the nested models method allows comparison of the fit among three different models in a competing strategy. According to Hair et al. (2014), nested modelling is a powerful option in SEM.
Before applying this multivariate technique, it is important to analyse the correlations between the different items of the survey at a univariate level. Table 2 presents the averages and standard deviation per item. The relative dispersion is small in all items, so the averages are representative, and no treatment of atypical data is needed. In addition, Table 2 shows that the averages generally oscillate around 5 or 7 but are somewhat lower for the variables corresponding to OIM4. This block of variables presents differentiated behaviour, with a decline in the scores based on the type of items that make up the block. As in the study of Hossain and Kauranen (2016), the enterprises, especially SMEs, were receptive to open innovation but were currently in the phase of the implementation and co-creation of specific mechanisms for putting it into practice. Thus, since these questions characterized how the open innovation paradigm materializes, it seems logical that there was a greater absolute and relative dispersion of the associated values.

Descriptive analyses
Finally, the exogenous social media (SM) variable, prior to its introduction in the proposed model, is related to external openness (EO). As in Laursen and Salter Table 2 Descriptive statistics and correlation matrix (Source: Authors) Unilateral p-value, *p < 0.05, **p < 0.01, ***p < 0.001 Mean SD RD1

Structural equation modelling
Structural equation modelling (SEM) analysis of the survey was undertaken using the AMOS program (version 24). SEM was selected as the statistical methodology because of its several advantages over regression modelling, including more flexible assumptions; the use of confirmatory factor analysis to reduce measurement error by having multiple indicators per latent variable or construct (survey); the desirability of testing overall models rather than individual coefficients; the ability to test models with multiple dependent variables (in this case, three dependent variables: EO, number of social media platforms and OIM); the ability to model mediating variables rather than being restricted to an additive model, as in regression (EO includes the influence of RD on OIM); the ability to model error terms; and the desirability of the strategy of comparing alternative models to assess relative model fit. Therefore, three alternative models are presented.
Before applying SEM, the data were checked to ensure that they met the assumptions required by the technique and the estimation method, maximum likelihood (ML), that we intended to use. First, "it is generally accepted that the minimum sample size is 100 to 150" (Hair et al. 2010, p. 632). The data set had responses from 181 companies, more than meeting the requirements of SEM.
Another important assumption is normality, which the AMOS program allows to be checked through the coefficients of skewness and kurtosis and which remained within the acceptable range of ± 2 (Schumacker and Lomax 2010). Reliability and discriminant validity are initial tests that are relevant to SEM. The reliability analysis of the final constructs indicated that all measurement scales exceeded the 0.7 threshold for Cronbach's α, therefore demonstrating satisfactory internal validity except in a case that is somewhat minor. A list of all the latent variable items, their standardized factor loadings and the α for each scale is presented in Table 3 (some problematic items have been removed from the analysis). Finally, a Harman's one-factor test was conducted to examine the discriminant validity. The results indicated that no single method factor exists, as the first factor accounts for less than 50% of the variance, as it retains 35.213% of the information. Thus, "common method bias does not appear to be a significant problem" (Koropp et al. 2014, p. 8).
Having verified the validity of the assumptions, they were applied, and the technique and data analysis followed a two-step approach using SEM (Anderson and Gerbing 1988). First, the model accuracy was assessed using a measurement model. Second, the path relationships were analysed. Table 3 shows the measurement model; the constructs are grouped appropriately; the indicators are always positive; and although their standardized coefficients are not always as high as those desired, they are always highly significant. In addition, the fit is correct, as shown in Table 4, in which the values of the adjustment of the confirmatory factor analysis or measurement model (CFA row) are shown (along with other values).
Studies using SEM rely on a variety of fit indices to evaluate model goodness of fit. It is common to use GFI (goodness of fit index) and RMSEA (root mean square error of approximation) and several other indices, such as Aloini et al. (2015). To give greater rigour to the analysis, Garson (2015) recommends "reporting Chi square (CMIN), RMSEA, and one of the baseline fit measures (NFI, RFI, IFI, TLI, CFI); and if there is model comparison, also report one of the parsimony measures (PNFI, PCFI) and one of the information theory measures (AIC, BIC, CAIC, BCC, ECVI, MECVI)" and also advises adding a note to the table to identify the conventional cut-off used.
The good fit of the measurement model suggests that the survey respondents were able to distinguish between the latent variables. Therefore, we advanced to testing the hypothesized model depicted in Fig. 1. First, only the relationships between the constructs based on the survey (RD, EO and OIM; Model 1 and Model 2) were observed, and the exogenous variable number of social media platforms was added  Table 5. The coefficients of the relationship between the variables are always positive, indicating that the relationship is significant, as was expected according to the hypotheses developed in this study. Table 4 shows the level of fit of the models, and all of them have a sufficient goodness of fit. However, in Model 1, the relationship between RD and OIM is shown to be nonsignificant (p-value = 0.512), and the adjustment level is lower than that in Model 2, which indicates that the influence of innovation (RD) on the open innovation management (OIM) is possible only through external openness (EO) since if EO does not exist, the circuit is broken, and the influence of innovation (RD) cannot pass through to open innovation Management (OIM).
The use of the structural equations allows us to understand that although the two alternative models presented are both correct in their adjustment levels, Model 2 is better since it always yields significant coefficients with the correct sign and higher adjustment levels (see Table 5). In this model, the standardized coefficients are 0.626 to quantify the influence of RD on EO and 0.332 for the influence of EO on IMO4, both positive and significant at any level. Model 3 enriches Model 2 with the addition of the variable number of social media platforms used by the company. The hypothesis that was formulated for this aspect (Hypothesis 4) established that the greater external openness of the enterprise (EO) influenced the number of social media platforms used by the company since in the tourism sector, openness is manifested through the use of the Internet. The standardized coefficient for this relationship is 0.217 (p-value = 0.005), and although it cannot be said with complete confidence that adding the social media variable increases the explanatory capacity of the model (Model 3), with respect to the model that does not introduce it (Model 2), it cannot be said that there is no effect, and there is no doubt that the EO relationship with social media is significant. Figure 2 presents the model that shows the relationship between the RD variable and OIM through the mediating effect of EO (Model 2) and the model that enriches this scheme with the introduction of social media (Model 3).

Discussion and conclusions
The results of this study confirm that open innovation is a strategy that requires the establishment of effective communication channels to stakeholders. The involvement of stakeholders in innovation processes means that their suggestions are implemented. Furthermore, openness and permeability result in open innovation even when it is not a corporate philosophy, and open innovation becomes a business practice according to Hypothesis 1 and in line with the studies of Laursen and Salter (2006) or Teirlinck and Spithoven (2013). The adoption of this paradigm can benefit internal R&D activity. However, as shown in this research, there is no direct relationship between a higher level of R&D and the implementation of the management model of this type of innovation (H3) since the relationship is through the mediating effect of exterior openness, as shown by the evidence supporting Hypotheses 1 and 2, in line with studies such as Vanhaverbeke et al. (2008). This investigation demonstrates that the most receptive companies abroad have developed an open innovation management model that supports this new perspective, confirming Hypothesis 2 (H2). This approach to management is one of the topics that continue to be most debated in the present literature and, by extension, is one of the contributions of this study. The conviction of the positive effects of open innovation seems to have been amply internalized, but the channelling of efforts, processes and indicators that organize and structure this external knowledge to improve innovation performance is, in practice, a pending issue (Chen et al. 2016). Previous studies have already pointed to the relationship between external openness and level of innovation as a point of departure for implementing a new vision of Another key result is the usefulness of social media in making the market orientation strategy effective, according to Hypothesis 4. The introduction of the social media variable is relevant in the model. In this sense, these web 2.0 tools facilitate contact with the outside to collect ideas that can be transferred to the development of new products (NPD) or services (NSD); these conclusions are in line with those obtained by the study of Bugshan (2015). At the moment, the majority of the use of social media in the literature is more related to the marketing and communication objectives of the target markets (Hvass and Munar 2012;Hays et al. 2013), while in this study, it is related to innovation management. Following this line of argument, it is worth noting that the tourism sector uses social media but is not yet fully certain of its benefits, either for return on investment, for marketing purposes (Hvass and Munar 2012;Wozniak et al. 2017) or from the point of view of innovation (Wang et al. 2015;Zhao et al. 2016). The level of openness through the social media use of the enterprises in the sample may be a consequence of the positive perception of innovation. In any case, it is possible that, as in the study of Ernst and Brem (2017), the direct impacts have not always been measured, and the results are not desirable, but the degree of penetration of social media predicts a scenario of collaboration with consumers that has increasingly appeared in the recent literature (Jabreel et al. 2017;Kaplan and Haenlein 2010;Lei et al. 2017).
The research approach fills some of the gaps detected in the previous literature by increasing the number of papers that take a multi-country perspective and delving into the open innovation phenomenon with a quantitative study of which the sample units are companies (Hossain and Anees-ur-Rehman 2015). However, there is still a need to develop studies that do not focus on the industrial sector, especially of high-tech companies, and to perform analyses that disaggregate sectors (Huizingh 2011) to determine whether there are differences in implementation derived from the nature of the activity (Wang et al. 2015). Focusing on tourism, Abbate and Coppolino (2011) emphasize that the particularities of the sector imply a different approach to the management of innovation, an issue that is verified with the research carried out. In addition, the tourism and hospitality industry is one of the least developed fields in the open innovation literature (Hossain and Anees-ur-Rehman 2015;Buonincontri et al. 2017).
In summary, this research shows that the relationship between R&D activity and open innovation management is produced through external openness. The first connection is shown by Cheng and Huizingh (2014), while the effect of external openness as a strategy for increasing the innovation level is highlighted in studies such as those by Atuahene-Gima and Ko (2001) and Teirlinck and Spithoven (2013). However, the novel aspect of this study is the identification of the relationship structure between the variables, including the mediating effect of external openness. Therefore, this study responds to the suggestion by Huizingh (2011, p. 1) "to build path models to understand chains of effects". In short, tourism companies show a higher rate of implementation of open innovation if the corporate culture is permeable to the outside environment and has systems and procedures for customer involvement. Finally, the introduction of the social media variable helps explain the orientation of the tourism sector towards the outside, which is in line with Lei et al. (2017). These web 2.0 communication platforms have become supporting mechanisms for the implementation of this philosophy and, therefore, the basis for the implementation of this new paradigm from the point of view of management. Furthermore, using social media with this goal can contribute to efficiently putting open innovation into practice through crowdsourcing and co-creation in the tourism sector. This idea agrees with Binkhorst andDen Dekker (2009) andJabreel et al. (2017), who identify open innovation as a challenge for destination and tourism enterprise management.
This study has many practical implications for the tourism sector and, by extension, for the management of tourism destinations. One of the most important practical implications for companies is that open innovation requires that companies preview the assimilation of strategic guidelines for external openness and the mediating effect of R&D activity on external openness. In brief, open innovation should not be a conceptual point of view in companies but should be reinforced with other strategic decisions. Above all, the efforts must be aligned with the appropriate management perspective to enable an effective change in the company. On the other hand, there is a need to appreciate the potential of social media for furthering tourist relationship management and branding, in particular in terms of contributions to the innovation level. Consequently, the encouragement and management of conversations generated on social media could be a tool for competitive improvement in innovation management in the tourism sector. According to the results, although companies have come a long way, much more remains to be done to achieve the desired results. Tourism companies interested in underscoring their innovation should establish channels of stakeholder involvement. First, social media is a strategic tool, and UGC is a valuable source of ideas for improving, innovating and differentiating a company in the sector. The qualitative leap can be made when open innovation is not only a philosophy but is implemented by defining channels, systems of evaluation, and integration and measuring its performance. At the present time, open innovation is appreciated but is not always implemented, and open innovation strategies often do not take social media into account. Currently, these tools are used more for engaging and branding than as a systematic way to innovate. Open innovation is especially interesting for SMEs because with fewer resources, they can show a competitive level of innovation owing to customer involvement. This idea coincides with the findings of Odriozola-Fernández et al. (2019), but it requires ongoing effort and the participation of the whole company.
In a strategic approach, open innovation offers a new dimension to the tourism sector that increasingly reflects the perspective of stakeholder involvement. It is highly recommended that companies adopt an integrated and strategic approach to innovate with the stakeholder support in order to become more competitive. Moreover, open innovation enhances the channels and co-creation spaces that are needed to provide information for management decision-making in order to achieve desirable innovation outcomes.
This research is not free of limitations, and these issues are related to the challenge for future research directions. First, the contributions of this study are eminently empirical, so it is advisable to make comparisons with other sectors of activity and deepen the focus on the tourism sector. Including companies from other countries is desirable, as is extending the sample to other tourist regions in Spain and Portugal. In addition future research might include tourist subsectors to ascertain whether different activities might benefit from different innovation and openness strategies and approaches. In applying SEM, it is possible to identify paths of behaviour based on such variable categorization. Regardless, in this initial approach, tourism is considered an economic activity consisting of a set of several varying types of ventures: accommodation, transport, hospitality services, leisure activities, etc. Therefore, in this study, the sector should be understood as an integrated whole, and sub-sectors were not used as a criterion for disaggregation of the sample. In addition, the response rate to this survey was only 52.16%, and although it is clearly superior to the response rate of other studies in the tourism sector that have rates of 12-26% (Kidwell and Fish 2007;Haemoon 2003;Weaver 2012;Duman and Mattila 2005), it remains low. Although the responses were tested on the basis of four key variables (nationality, activity of the company, turnover and number of social media platforms), and no apparent evidence of non-response bias was found, the non-existence of such bias cannot be completely guaranteed. Therefore, researchers should evaluate the results of the study cautiously.
Future research should explore the evolution of the use of social media as an element of support for the management of innovation as a key aspect to strengthen market orientation and permeability to the environment and to verify whether the results are obtained in a way similar to that of this study. Additionally, future research could add information to the analysis to contemplate not only the number of social media platforms employed but also the intensity of use and types of use. This proposal is in line with the research of Laursen and Salter (2006). However, a thorough and qualitative method of analysis would be necessary to collect data properly for this variable (Chiang and Hung 2010). Likewise, this study shows that more open companies are aware of tourists' involvement and consequently establish channels to put open innovation into practice, but there is no evidence regarding whether that practice means better performance in innovation. This is precisely a future line for further development of the research.
Another issue to consider in the future is the performance analysis derived from open innovation (Wang et al. 2015;Zhao et al. 2016). Companies are aware of the need to improve the implementation of open innovation from the point of view of management and to increase their social media use to collaborate with tourists. Future research should aim to demonstrate the maturity of open innovation in the tourism sector. In this sense, it is necessary to continue the analysis of the open innovation phenomenon and its effects on competitiveness both to advance the investigation and to benefit tourism companies.