Antecedents of Relationship Learning in Business-Non-Profit Organization Collaboration Agreements

The literature has highlighted that relationship learning constitutes a value creation pathway in business and non-profit collaborations. However, this process has not been studied in detail. The main objective of this study was to carry out an in-depth analysis of relationship learning by examining the main factors that favor its development. For this, a research model was developed based on the relevant literature in the field of Relationship Marketing and Inter-Organizational Learning. The hypotheses were tested with data collected from 102 Spanish firms in collaboration relationships with non-profit organizations. The results reveal the importance of four relational variables in the direct and indirect explanation of relationship learning—values shared by the partners, trust, commitment, and cooperation. First, shared values form a key antecedent of trust and commitment, these in turn are fundamental precursors of the partners’ cooperation, and commitment and cooperation showed themselves to be the main determinants of relationship learning between firms and non-profit organizations. The proposed model could enrich the relevant literature and provide better understanding of the relational aspects that businesses and non-profit organizations should work on if they wish to improve their level of relationship learning.


Introduction
Recently, there has been growing awareness that important social and environmental challenges should not be left for the public and non-profit sectors to deal with, but should also be seen as one of the duties of the private sector [1]. A good example of this is found in the 2030 Global Agenda that was approved in the United Nations General Assembly in September 2015, which recognizes the business sector as being a fundamental agent for the achievement of the so-called Sustainable Development Goals (SDG) [2].
In this sense, the Cross-Sector Social Partnerships, and in particular, the collaborative relationships between corporations and non-profit organizations (NPOs) in the development of their Corporate Social Responsibility (CSR) policies, have become widely accepted by both types of organizations to work together [3][4][5] in order to confront difficult problems [6,7]. An example of a collaboration agreement created to deal with this type of problem was the social cooperation between the business Unilever and World Wildlife Fund (WWF), aimed at the creation of an ecologically sensitive brand with the objective of helping guarantee the sustainability and integrity of the marine ecosystem. Another marked example of collaboration was the agreement between Starbucks Coffee Company and Conservation International (CI) which had, as an objective, the development of a set of conservation principles for coffee production in order to address more efficiently the sustainability problems existing in developing countries.
The business-non-profit collaborations combine different forms of acting and thinking [8], in order to create value in the relationship [9,10]. Such value is generated through a series of chronological phases: formation, partner selection and implementation [9,11]. In this sense, the literature on collaboration agreements between corporations and NPOs has shown that the implementation phase of the collaborative relationship is the moment when this creation of value is greatest [11]. Among the different factors that determine the generation of value in this phase, several authors highlighted the importance of relationship learning (collective learning) [9,10,12]. This concept, defined as a process in which the two partners work together, exchanging information and know-how as well as analyzing and solving strategic problems [13]), directly improves the success of the collaborative relationships. The reason for this is that the presence of relationship learning favors the creation of value via efficacy (same costs, greater benefits), efficiency (same benefits, lower costs), and/or innovation (creation of new results) [10,12]. However, despite the importance of this process, to the best of our knowledge the literature on Cross-Sector Social Partnerships has not studied it in any depth. We found no previous research that empirically analyzed its main antecedents or determinants, since until now the study of cause-effect relationships in this context appears to have been avoided [14]. This fact may have been influenced by the statistical methods that have been used in this research area, with a large number of qualitative empirical works based mainly on case studies [15,16].
The main objective of the present study was to examine empirically, through quantitative data, the main precursors of relationship learning in collaborative relationships between corporations and NPOs. As branches of this main objective, we set the following specific objectives: first, to examine the role of the values shared by the partners as a key antecedent of the trust and commitment in the relationship; second, to analyze the importance of trust and commitment as fundamental precursors of cooperation between the partners; and, third, to study the role of trust, commitment, and cooperation as the main determinants of relationship learning between corporations and NPOs.
We hope, with this study, to contribute to the literature about Cross-Sector Social Partnerships in two ways. First, based on the relational marketing literature and inter-organizational learning, our research attempts to respond to calls for the development of new knowledge about collaborative relationships between corporations and NPOs [11] by constructing a model of relationship learning that identifies different variables and their inter-relationships. Second, our study tries to address the urgency for generalizable results in the field of corporation-NPO collaborations [15] by applying quantitative data to empirically corroborate the proposed structural model. The rest of this paper is structured as follows. Firstly, we develop the conceptual framework and formulate the research hypotheses. Secondly, we present the method used. Our sample comprised Spanish corporations that have established collaborative relationships with NPOs in recent years. The data collected were analyzed using the Partial Least Squares (PLS) statistical technique. Thirdly, we report the main results of the empirical analysis. Fourthly, we discuss these results, and present the main conclusions, including the theoretical and practical implications. Lastly, we set out the limitations of the study and the main lines for future research.

Conceptual Framework and Hypotheses
In this study, the key result of a collaborative relationship is relationship learning. According to Knight [17] (p. 439), inter-organizational learning can be defined as "learning in a dyadic or inter-organizational context, and where the learner can be an individual, a group, an organization, a partner, or a network." Inter-organizational learning is a process that can take place both in short-term relationships, which are limited, and largely oriented towards their own interests, as well as in collaborative relationships, without a definite time limit, and oriented towards a common interest [12]. Relationship learning, on the other hand, is "characterized exclusively by collaborative relationships between partners" [18] (p. 744). In order to provide a definition of this concept that allows it to be differentiated from the more general construct of inter-organizational learning, we present the definition proposed in Barroso-Méndez, Galera-Casquet, and Valero-Amaro [12] (p.174) in a study of collaborative relationships between corporations and NPOs: "Relationship learning involves a learning process by two organizations as a group within the context of their collaborative relationship." Therefore, relationship learning can take place across all learning contexts, with the members of the collaborative relationship being the only learners [12]. Knight and Pye [19] state that relationship learning involves the partners making changes at a relationship level. In other words, if, through their collaborative relationship, two partner organizations change the behaviour of their relationship, these organizations have managed to learn at a collective analysis level. Relationship learning, unlike inter-organizational learning, is notable for the existence of a close relationship between the partner organizations [6,20]. Relationship learning comprises three main sub-processes based on collaborative relationships between the partners [21]: • Exchange of information. The exchange of information between partners is a starting point and a necessary element of relationship learning [21]. Both organizations must be willing to share data and relevant information in order to plan their actions. This results in notable improvements in their levels of organizational performance [22]. • Shared interpretation. The interpretation of the information exchanged between partners works as a mechanism for sharing the link existing between information and meaning [18]. The dialogue between the two partner organizations plays a key role in this sub-process of the relationship learning [12]. Indeed, given that the partners may not understand the shared information in the same way, during this step it is necessary for both partners to use mechanisms to form learning fields between their organizations [23], for example, establishing frequent management meetings between the different work teams [21]. • Integration of knowledge. According to Cheung, Myers and Mentzer [24], the integration of knowledge takes place when the partner organizations elaborate specific memories of the collaborative relationship in which the knowledge exchanged is collected. Among other aspects, the memory of the relationship includes the values or performance routines [25]. Additionally, it should be noted that the memory has a shared character. That is, it is external to each participating organization, but internal to their collaborative relationship. This implies that the generation of knowledge is not owned by either of the partners, but rather by the interaction relationship, and therefore both can access it from any location [25,26]. In this sense, the memory of the relationship interacts with the unique knowledge of each participating organization, which gives rise to more innovative ideas [18].
The business-NPO partnership literature highlights that relationship learning can create numerous benefits for NPOs and businesses. On the one hand, relationship learning can improve processes [6], foster unique capabilities [27] and create opportunities for innovation [9,11,28] for NPOs, and, on the other hand, it can favor new levels of innovation [9,11,29] and competitiveness [27] for businesses. Moreover, relationship learning can contribute to improving the impact of such collaborations by providing better solutions that are able to address complex social issues more effectively [10], as shown in Table 1.

Subject
Type of Outcome Outcomes NPO New Outcomes Opportunities for innovation [9,11,28].

Better Outcomes
Opportunities for improvement of processes [6]. Development of unique capabilities and knowledge creation [27].

New Outcomes
Opportunities for innovation [29].

SOCIETY New Outcomes
Better innovations to address the needs of the beneficiaries [10].
At this point, it is worthy to mention that, despite the numerous outcomes that relationship learning could generate in collaborative relationships between businesses and NPOs, this literature has not adequately addressed the study of their main antecedents.
Previous studies who have examined relationship learning in collaboration agreements between corporations have mentioned that the capacity of a relationship to learn is intimately linked with the relational context of this relationship [13,21,[30][31][32]. In other words, achieving relationship learning is linked to the presence of multiple behavioral and psychological aspects that influence the relationship over time, and that find expression in different relational factors. Within this set of relational factors, the literature on relationship marketing and inter-organizational learning between corporations [21,31,[33][34][35][36] notes that the learning capacity of a relationship is related to the level of trust, commitment, and cooperation existing between the partners, aspects which, in turn, gain in strength in relationships where both partners share a set of similar values or beliefs. The literature on collaboration agreements between corporations and NPOs has not proposed and empirically corroborated the direct or indirect influence these antecedents might have on relationship learning, an issue that we intend to address in this study.

The Influence of Shared Values on Trust and Commitment
The construct shared values have been conceptualized by different authors. However, the definition most often cited in the Relationship Marketing literature is that provided by Morgan and Hunt [37] (p. 25) "the degree to which partners have common beliefs about whether behaviour, objectives and policies are important or insignificant, appropriate or inappropriate, correct or wrong." Following this conceptualization, in our research we can define the construct of shared values as "the level at which a corporation and an NPO share, within their collaboration relationship, a set of analogous values on different aspects." (For example, in terms of beliefs about the importance of different social problems or issues, necessary ethical behaviour, etc.). According to different authors [33,36,37], the presence of similar values is closely related to the existence of behaviour norms accepted by both partners, since this type of norm prescribes behaviours of actions to maintain the relationship, making behaviours directed exclusively towards the interests of an individual partner difficult, and constitutes a guarantee against abuse in the relationship [33]. According to the relationship marketing literature, when partners perceive that their counterpart has common beliefs about aspects that are mutually important for both, they will be willing to increase their trust [33,[36][37][38] and their commitment to the relationship [35,37,39], since the presence of values that the partners share favors their advancing in the relationship. Therefore, we formulate the following two research hypotheses: Hypothesis 1 (H1). Shared values directly and positively affect trust in collaboration agreements between corporations and NPOs.

Hypothesis 2 (H2).
Shared values directly and positively affect commitment in collaboration agreements between corporations and NPOs.

The Influence of Trust on Commitment
Trust and commitment are two fundamental constructs in any relationship because they both favor collaboration between partners. According to Moorman, Deshpandé and Zaltman [40] (p. 82), trust can be conceptualized as "a willingness to rely on an exchange partner in whom one has confidence." This definition is similar to that proposed by Morgan and Hunt [37] (p. 23), according to which, "trust exists when one part is confident in the reliability and integrity of its counterpart." According to Dwyer, Schurr and Oh [41] (p. 19), commitment refers to "the implicit or explicit sign of the continuity of the relationship between the parties in the exchange." Anderson and Weitz [42] (p. 191) argue that the essence of commitment is stability and sacrifice, and on this basis they define this construct as "the desire to develop a stable relationship, a willingness to make short-term sacrifices to maintain the relationship, and a confidence in the stability of the relationship." Most of the relationship marketing literature in a business-to-business context has shown that, in relation to the link between the two constructs, trust is the main precursor of commitment [43][44][45][46], because, without trust, either of the partners could consider it to be risky committing to the relationship [37,47]. A similar argument can be found in the business-non-profit collaboration literature [48]. Indeed, qualitative research in this context has shown that "when individuals develop a personal relationship of trust within the collaboration then the level of embeddedness of the relationship becomes more evident" [49] (p. 422). Therefore, in this study, we expected trust between corporations and NPOs to encourage a commitment to the relationship, so that we formulated the following hypothesis:

Hypothesis 3 (H3).
Trust directly and positively affects the commitment in collaboration agreements between corporations and NPOs.

The Influence of Trust and Commitment on Cooperation
Cooperation has been conceptualized by many authors [50][51][52]. However, the definition which has the most citations in the literature is the one proposed by Anderson and Narus [53] (p. 45), according to which cooperation can be defined as "a set of coordinated actions performed by two corporations in a relationship of interdependence to achieve mutual results." Cooperation suggests, therefore, a relationship where both partners work together to achieve a series of joint objectives, exceeding the limits established in the association's contracts.
Most of the research that has focused on the study of the cooperation construct, mainly in a business-business context, has proposed that the existence of trust and/or commitment between partners favors the development of cooperative behaviour between them. Numerous authors [43,51,54,55] consider that trust is a key precursor of cooperation, because a partner will only risk carrying out joint activities when having full confidence and assurance in the other [37,43]. Other authors [33,37,45,46,56] indicate that commitment is also a clear facilitator of cooperation, because only when a partner is committed to the relationship will they carry out joint activities with the other in order for the relationship to work and for the proposed objectives to be achieved. Consequently, in this study, we expected trust and commitment between corporations and NPOs to encourage cooperation in the relationship. Thus we formulated the following hypotheses: Hypothesis 4 (H4). Trust directly and positively affects cooperation in collaboration agreements between corporations and NPOs.

Hypothesis 5 (H5).
Commitment directly and positively affects cooperation in collaboration agreements between corporations and NPOs.

The Influence of Trust, Commitment, and Cooperation on Relationship Learning
Numerous authors have highlighted the importance of trust as a mechanism of governance which promotes the creation of a relationship learning culture, both in business-business collaboration agreements, and in those between corporations and other social actors [48,57,58]. Morgan and Hunt [37] mention that when both partners believe that they will not be used by the other one, they tend to share more information and know-how. Other authors [59,60] also point out that the presence of trust in a relationship makes both partners more likely to exchange ideas, as well as to solve problems or conflicts in a friendly way. Muthusamy and White [57] indicate that trust between partners can favor their degree of openness and availability, as well as improve their level of transparency and mutual exchange of knowledge. Finally, several authors [13,21] have proposed and empirically corroborated that trust improves relationship learning in business-business collaboration agreements, since its presence favors a situation in which the partners exchange useful and valuable information, establish a dialogue aimed at interpreting the said information, and elaborate shared memories of the relationship in which the knowledge generated is stored [21]. Similar arguments can be found in the business-non-profit collaboration literature. Thus, for example, Barroso-Méndez, Galera-Casquet, and Valero-Amaro [12], who presented a case study of the multi-year collaboration between Starbucks Coffee Company and Conservation International (CI), highlighted the importance of an atmosphere of mutual trust between the two partners with the objective of improving their conjoint relationship learning. Accordingly, we expected that trust fosters relationship learning between corporations and NPOs. Thus, we formulated the following hypothesis: Hypothesis 6 (H6). Trust directly and positively affects the relationship learning in collaboration agreements between corporations and NPOs.
Some authors highlight the role of commitment in the emergence of relationship learning [21,31,48]. Selnes and Sallis [21] note that the partners' commitment, linked to the objectives set out at the beginning of the collaboration, improves their exchange of information and know-how. Other authors [61,62] emphasize that commitment promotes the exchange and creation of new knowledge. In this line, various authors [21,31] have postulated and empirically corroborated that commitment directly and positively influences relationship learning in business-business agreements since its existence makes both partners willing to share information that is of value for both, to engage in dialogue to give meaning to the shared information, and to elaborate shared memories of the relationship [21]. Consequently, we expected commitment between corporations and NPOs to have a positive influence on relationship learning, and formulated the following research hypothesis:

Hypothesis 7 (H7). Commitment directly and positively affects relationship learning in collaboration agreements between corporations and NPOs.
Several authors [13,21,31,57] highlight the importance of cooperation for the development of relationship learning. Muthusamy and White [57] and others argue that the absence of cooperative behaviour between partners hinders the achievement of relationship learning, since certain information, especially the most sensitive, or the know-how of each participating organization cannot be easily exchanged through simple interaction relationships. In this sense, we expected the cooperation between corporations and NPOs to have a positive influence on relationship learning, and formulated the following hypothesis:

Hypothesis 8 (H8). Cooperation directly and positively affects relationship learning in collaboration agreements between corporations and NPOs.
Therefore, the structural model that we propose for the study of learning in collaboration agreements between corporations and NPOs is the following (Figure 1

Sample
In line with the growing importance of the role the private sector has in solving current social and environmental challenges, we decided to restrict our study to the business environment and define our sample as "those corporations based in Spain that have maintained some collaboration agreement in the field of Corporate Social Responsibility-philanthropic donations, patronage of events, programs of cooperative volunteers, etc., with one or more NPOs in recent years." In order to achieve an adequate sample on which to test our research hypotheses, we decided to look for a database containing this information. However, despite making a broad search, none of the databases available offered secondary data about our population. As a solution, we decided to create our own database, using as a source of information the NPOs that belonged to the Spanish Fundraising Association. The decision was made to select the Spanish Fundraising Association because, among other reasons, of its ample national coverage, incorporating as associates a wide range of NPOs based in different parts of Spain. Once we knew the associated NPOs, we reviewed their last three annual reports for the collaborating firms mentioned in their Acknowledgements sections. This gave us a representative sample of 657 firms based in Spain that had carried out some type of collaboration with the NPOs during those years.
To collect information about each of these firms, a very elaborate search process was carried out using both their own web pages, as well as other sources on the Internet. For each business, we tried to collect information about: 1) name of the business; 2) name of the NPO(s) the business had had any collaboration agreement with during the last years; 3) contact person responsible for managing the said collaboration agreements, normally, the head of the Department of Social Cooperative Responsibility, of the Department of Communication, or of the company's Foundation; 4) e-mail address; 5) telephone number; and 6) complementary information.
In our research, the fieldwork was carried out via e-mail over a period of 6 months in 2015. At the end of this period, we decided to close the time window for questionnaire reception, having obtained a total of 102 responses, which represents a response rate of 15.53% with respect to the total number of corporations in the survey population. Table 2 shows the characteristics of the sample:

Sample
In line with the growing importance of the role the private sector has in solving current social and environmental challenges, we decided to restrict our study to the business environment and define our sample as "those corporations based in Spain that have maintained some collaboration agreement in the field of Corporate Social Responsibility-philanthropic donations, patronage of events, programs of cooperative volunteers, etc., with one or more NPOs in recent years." In order to achieve an adequate sample on which to test our research hypotheses, we decided to look for a database containing this information. However, despite making a broad search, none of the databases available offered secondary data about our population. As a solution, we decided to create our own database, using as a source of information the NPOs that belonged to the Spanish Fundraising Association. The decision was made to select the Spanish Fundraising Association because, among other reasons, of its ample national coverage, incorporating as associates a wide range of NPOs based in different parts of Spain. Once we knew the associated NPOs, we reviewed their last three annual reports for the collaborating firms mentioned in their Acknowledgements sections. This gave us a representative sample of 657 firms based in Spain that had carried out some type of collaboration with the NPOs during those years.
To collect information about each of these firms, a very elaborate search process was carried out using both their own web pages, as well as other sources on the Internet. For each business, we tried to collect information about: (1) name of the business; (2) name of the NPO(s) the business had had any collaboration agreement with during the last years; (3) contact person responsible for managing the said collaboration agreements, normally, the head of the Department of Social Cooperative Responsibility, of the Department of Communication, or of the company's Foundation; (4) e-mail address; (5) telephone number; and (6) complementary information.
In our research, the fieldwork was carried out via e-mail over a period of 6 months in 2015. At the end of this period, we decided to close the time window for questionnaire reception, having obtained a total of 102 responses, which represents a response rate of 15.53% with respect to the total number of corporations in the survey population. Table 2 shows the characteristics of the sample: Regarding the collaboration processes of the participating firms with their respective NPO partners, we collected several types of agreements, mainly philanthropic (based on donations of money and produce) (see Table 3), and with different durations, ranging between 3 months and 42 years, although most have remained in force for about 6 years (73 months) (see Table 4): Table 3. Types of collaboration agreements.

Type Frequency
Donations in money or produce 92

Patronage of events 19
Marketing with cause (donation of a percentage of the sales of their product to an NPO) 22

Cooperation volunteers 32
Strategic collaboration (wide range of activities, multiple and distinctive interchangeable resources, and mutual value-creation 35 Other 5 Note: In the question referring to the type of agreement the business had with their partner (NPO), multiple responses were permitted so as to collect the different formats of collaboration that one business may have had with their NPO partner during their partnership. Hence, the total frequency of the number of responses in the Table is greater than the sample size.

Measurements
All the variables of our research model were measured by adapting pre-validated scales in the existing literature, mainly in the context of collaboration agreements between corporations. The indicators of each variable were measured on a Likert scale of 7 points (from total disagreement to total agreement).
The construct shared values were measured by adapting the scale of MacMillan, Money, Money and Downing [63]. In this way, this construct was modeled as a one-dimensional construct composed of three reflective items, which represent manifestations of the construct being considered [64]. Trust was measured by adapting the scale of Vázquez, Iglesias and Álvarez-González [65]. It was thus modeled as a second-order construct comprising two dimensions: credibility and benevolence. Both of these dimensions were modeled as first-order reflective constructs, with six items corresponding to each. First-order reflective constructs are theoretical concepts inferred from their indicators. The changes that occur in the constructs are manifested by their indicators [66]. The commitment and cooperation constructs were measured by adapting the scales used by Wittmann, Hunt, and Arnett [46]. In this way, both constructs were modeled as one-dimensional constructs, each composed of five reflective items. Finally, relationship learning was measured by adapting the scale of Selnes and Sallis [21]. This was thus modeled as a second-order construct comprising three dimensions: information exchange, shared interpretation, and knowledge integration. All of these dimensions were modeled as first-order reflective constructs, including seven, four, and six items, respectively.

PLS Analysis
In our study, we used the Partial Least Squares (PLS) statistical technique to test the proposed research model. The choice of PLS was for the following reasons [67]: (i) the objective of our study, oriented towards prediction; (ii) the complexity of the proposed structural model, both in the number of variables and in their dimensionality; and (iii) the use of the latent variable scores in subsequent analyses.

PLS Results
A PLS analysis involves a two-stage process [68]-first, evaluating the measurement model, and second, evaluating the structural model.

Measurement Model
The literature notes that, in the case of constructs with reflective items, as is our case, their reliability and validity must be evaluated [68]. Firstly, both the dimensions and the reflective indicators met the reliability requirement, obtaining loadings greater than 0.7 [69] (see Table 5). To achieve this result, we carried out an item trimming process, removing some weak items from the scales of trust (C5 and C7), commitment (CP3), and cooperation (CC4 and CC5) (see Table A2). Secondly, all the constructs (first-and second-order) satisfied the requirements of construct reliability and convergent validity, obtaining composite reliabilities greater than 0.7 [70], and values of average variance extracted (AVE) greater than 0.5 [71] (see Table 5). Thirdly, all the constructs achieved discriminant validity in accordance with the criteria of Fornell and Larcker [71]. In Table 6, one observes that the square root of the AVE of each construct (values in boldface along the diagonal) is greater than the correlation between that construct and the rest of the constructs of the model.

Structural Model
The literature notes that, in order to evaluate the structural model, it is necessary to (i) analyze its predictive power, (ii) study the predictive relevance of the endogenous constructs, (iii) examine the significance of the path coefficients, and (iv) analyze the model's goodness-of-fit [67].
With regard to the evaluation of the predictive power, it should be emphasized that this is performed by studying the R 2 values of the endogenous constructs [72]. In this respect, Chin [72] notes that R 2 values of 0.67, 0.33, and 0.19 correspond to substantial, moderate, and weak predictive powers, respectively. As can be seen in Table 7, in our model, trust, commitment, and relationship learning reached a moderate level of predictive power (0.36, 0.54, and 0.53, respectively), and cooperation a substantial predictive power (0.68).
With regard to the predictive relevance of the endogenous constructs with a reflective measurement model, it should be noted that this study was performed using the Stone-Geisser test (Q 2 ). As can be seen in Table 7, all the endogenous constructs of our model have predictive relevance, with Q 2 values greater than 0 [73].
With regard to the significance of the path coefficients, we should mention that this study was performed using a bootstrapping procedure, which provides the standard error and t-values of the parameters. According to Hair, Ringle and Sarstedt [68], to perform this analysis, it is necessary to bootstrap with 5000 subsamples, and use Student's one-tailed t-distribution with (n-1) degrees of freedom, where n is the number of subsamples. With the exception of one hypothesis, H6, positing a relation between trust and relationship learning, we found empirical support for the rest in our research, since the relationships that they posited between the different constructs considered were statistically significant (see Table 8).   Lastly, with regard to the goodness-of-fit of the proposed research model, it should be noted that this study was performed by analyzing the value of the SRMR (Standardized Root Mean Square Residual) measure, which represents the Euclidean distance between the empirical correlation matrix and the correlation matrix implicit in the model [74]. Henseler et al. [75] suggest that the value of this measure should not exceed a threshold of 0.08, above which the general fit of a model would be unsatisfactory. As can be seen in Table 8, the value of SRMR (0.05) was less than the recommended threshold, and therefore, the conclusion would be that the general fit of the proposed model was very satisfactory.

Discussion and Conclusions
Although the current literature highlights the importance of relationship learning as a way of creating value in Cross-Sector Social Partnerships, little attention has been given to its study. Thus, we believe that the present research constitutes an important step in this direction, proposing and empirically validating an explanatory causal model of relationship learning in corporation-NPO collaborative relationships.
It should be noted that the proposed model achieves a generally good fit on the basis of the data collected, and supports seven of the eight research hypotheses formulated.
Hypotheses H1 and H2, which represent the possible influence of shared values on trust and commitment, respectively, found empirical support, which is consistent with previous studies [33,[36][37][38].
Although shared values act as a good predictor for both of these constructs, we found its impact to be greater on trust, explaining 36.3% of its variance. This result suggests that when partners perceive that their counterpart has common beliefs about mutually important aspects for both of them, the relationship will advance, and there will be notable improvement in the level of trust existing between them. Therefore, both partners try always to keep their word, to comply with the obligations agreed on at the beginning of the association, and not to carry out unexpected actions that might have a negative impact on the other partner.
The relationship between trust and commitment, hypothesis H3, yielded statistically significant results, in line with other research studies [43][44][45][46]48]. This reflects the importance of this relationship, independently of the research context being analyzed, and therefore the need to establish an environment of trust between corporations and NPOs that can serve as a basis for both partners to make every effort to maintain their relationship in the future.
Hypotheses H4 and H5, which represent the possible influence of trust and commitment on cooperation, respectively, found empirical support, again consistent with previous research [33,34,46,47,52]. Although trust and commitment are good predictors of cooperation, explaining 68.7% of its variance, commitment has the greater impact (48.1%). Therefore, the mere existence of a climate of trust in the relationship may not lead to a significant increase in the level of cooperation between the partners. In contrast, when both partners are committed to the relationship, with this collaborative relationship being mutually important to both, they are more motivated to work together, trying to adapt to each other when they have to make decisions that affect the achievement of their mutual goals.
Hypotheses H7 and H8, which concern the relations between commitment and learning and between cooperation and learning, produced statistically significant results. These results were consistent with suggestions made by various authors [21,31] according to which both of these relational factors are fundamental determinants to improving relationship learning in relationships between businesses. In the context of our paper, these results agree with the conclusions of different qualitative studies [3,12,76,77]. Thus, for example, Jamali, Yianni, and Abdallah [76], in a comparative analysis of six case studies of collaborative relationships between businesses and NPOs in the UK, observed that the existence of strong ties between the partners in commitment and cooperation was an important aspect favoring their conjoint learning. However, when going into greater depth in the study of the variance explained, one can observe that, while commitment explains 28.8% of the variance of relationship learning (53%), cooperation predicts 15.4% of the variance of that construct. Therefore, the partners' commitment to the relationship turns out to be fundamental for the emergence of relationship learning, since its presence means that both partners will be willing to share information, to maintain a dialogue aimed at giving meaning to that shared information, and to elaborate shared memories of the relationship. Contrary to our expectations, we found no empirical support for hypothesis H6, which represents the possible influence of trust on relationship learning. High levels of trust might embed "hidden costs" that limit its effectiveness in the collaboration, and may induce a lack of critical information exchange [30]. A high level of trust is usually accompanied by positive emotions regarding the partner. In this context, there is the risk that critical information is not shared for fear of endangering the relationship's "good atmosphere" [21,30]. Our study, therefore, confirms the lack of any direct connection between trust and relationship learning in the case of collaborative partnerships between businesses and NPOs. This weak direct effect of trust had already been implicitly observed in this context by Sanzo, Álvarez, Rey, and García [48], who found a lack of any direct relation between trust and the development of innovation in social partnerships. However, this non-significant effect of trust on relationship learning does not mean that this variable is not important since it is a driver of commitment and cooperation, and constructs with a direct impact on relationship learning. On the one hand, trust contributes to increasing mutual commitment, since when both partners believe that the other one keeps their promises and is sincere and honest in the partnership, they try to look after this relationship, putting maximum effort (financial and non-financial) into maintaining it for the future, thus increasing the possibility of relationship learning. On the other hand, greater trust between the partners significantly improves cooperation, since when both partners believe that the other does not lie, and is concerned about their well-being and interests, they are willing to work together to both be successful, with the result being a cooperation between the partners that facilitates a process of deeper mutual learning.
We believe that the results of our study are relevant for both academics and professionals. On the one hand, our study makes several contributions to existing knowledge. First, it responds to calls for the development of a new theory on Cross-Sector Social Partnerships [11], by constructing a model of learning that seeks to identify the main relational factors that directly or indirectly favor the appearance of relationship learning in collaboration agreements between corporations and NPOs. Second, our study contributes to mitigating the scarcity of generalizable results in the field of corporation-NPO collaborations [15] by corroborating empirically, through a large sample of companies, the proposed learning model. On the other hand, our study generates various practical implications for business managers who are responsible for the development of collaboration agreements with NPOs. We recommend that managers who wish to enjoy a quality relationship in terms of trust and commitment should pay especial attention to the phase of selecting their partners, and only establish collaboration agreements with those NPOs that have similar values or beliefs as them in aspects that are important for both-social causes, ethical behaviour, etc. We suggest that managers who wish to improve the level of cooperation with their NPO partners should strengthen their trust and commitment, be credible and benevolent in their everyday actions, and devote a maximum of resources to the relationship. Finally, managers who wish to improve relationship learning must commit to the relationship, dedicating time and effort to its maintenance, as well as be willing to cooperate with their NPO partner, working together, and increasing their common understanding and empathy.

Limitations and Future Lines of Research
There are always certain limitations to any research that may condition the generalization of its results. The main limitations of our study are: On the one hand, our research took, as a sample population, a Spanish corporation which had maintained some collaboration agreement with one or more NPOs in recent years. This circumstance restricts the generalization of the results beyond this context. On the other hand, our study corroborated empirically the proposed causal model contemplating only the perception of one of the parties involved in the collaborative relationshipthe business. However, it stands out that these limitations indicate paths for future research. First, future studies can expand the sample population to corporations in other countries. Second, other studies can test the proposed research model using the perceptions of NPOs. Third, future research can expand the causal model that was developed by considering other relational factors as antecedents of relationship learning, such as the partners' opportunistic behaviour or the functional conflict existing in the relationship, the presence of which may reduce or favor, respectively, the development of relationship learning in associations between corporations and NPOs. Funding: This research was co-financed by the Extremadura Regional Government and the European Regional Development fund (FEDER) (Code GR18027).

Conflicts of Interest:
The authors declare no conflict of interest.  Table A2. Initial loadings of the first-order construct indicators.

Relationship learning (SMC)
Information exchange (reflective dimension) A1: As partners, we both exchange information on successful and unsuccessful experiences in implementing different social programs 0.837 A2: As partners, we both exchange information regarding changes in the needs of the target population of the programs that we develop together 0.887 A3: As partners, we both exchange information regarding changes in the specific environment of the programs that we develop together 0.857 A4: As partners, we both exchange information regarding new techniques for implementing our programs, new methods, or identification and intervention tools 0.809 A5: As partners, we both exchange information as soon as possible on any unexpected problem 0.868 A6: As partners, we both exchange information regarding changes in the strategies and policies of our organizations 0.837 A7: As partners, we both exchange information that is sensitive for us, such as financial information, know-how, and new developments 0.738

Shared interpretation (reflective dimension)
A8: In our relationship, it is common to set up joint teams to solve operational problems 0.829 A9: In our relationship, it is common to set up joint teams to analyse and address strategic issues 0.821