SCI is defined as the level in which an organization strategically collaborates with its business partners, collaboratively managing the intra and interorganizational processes through a continuous partnership to achieve mutually beneficial strategic goals (Flynn et al., 2010; Kulp et al., 2004; Rai et al., 2006). Therefore, it is highlighted the intrinsic relationship of this construct with the collaboration among agents of a supply chain, conceptualized as an integration component and performing both externally on the organizations’ level and internally among people and departments of the company (Wiengarten et al., 2013). Thus, in this study SCI is considered a broad construct which has the intra and interorganizational collaboration as an integration component.

Although SCI is duly recognized as useful factor for increasing organizational performance, the results of empirical studies on such relationships have been mixed and even controversial (Cao & Zhang, 2011; Flynn et al., 2010). Moreover, despite effective supply chain management has become a potentially valuable way of ensuring better organizational performance, an understanding of why and how this process affects company performance, considering important areas and moderating effects, is still incomplete (Trkman et al., 2010).

In order to deepen this discussion, the Resource Orchestration Theory argues that “managers need to orchestrate their resources to realize any potential advantage” (Chirico et al., 2011, p. 310). This theory was developed from the Resource Based View (RBV), which proposes that the company's performance can be explained by the heterogeneity in having valuable, rare, inimitable and non-replaceable resources (Barney, 1991; Sirmon et al., 2011).

However, researches has shown that the influence on companies’ performance is not only due to resources possession, but involves management actions related to structuring the company’s resources portfolio, and grouping these resources in capabilities in order to obtain competitive advantage (Sirmon et al., 2007). Furthermore, the Resource Orchestration Theory is particularly useful to comprehend the development of resources and capabilities (Liu et al., 2016).

Liu et al. (2016), after reviewing the literature about SCI based on the Resource Orchestration Theory, presented the following constructs, which are complementary and interdependent as described: i) information integration: it reflects the extent in which a company shares information with key partners regarding several activities of the SC, such as information about sales, inventory, production and distribution; ii) Synchronized planning: it refers to the scope in which the company collaborates with its key partners through the preparation of plans, ensuring a greater synchronism regarding future activities of the SC, and the requirements for the continuation of joint efforts; iii) Operational coordination: is related to the extent in which a company simplifies and automates the SC processes with key partners; and iv) Strategic partnership: it refers to the scope in which the company establishes long-term relationships with key partners to achieve strategic goals.

The theoretical model proposed in this study uses these constructs established by Liu et al. (2016), based on the Resource Orchestration Theory to evaluate the SCI.

An increasing number of research in the field of operations management has sought to identify the benefits derived from information sharing among companies of the SC (Kulp et al., 2004). Moreover, there are evidences that the SCI leads to an improvement on companies’ performance (Cao & Zhang, 2011; Devaraj et al., 2007; Lee et al., 2011; Liu et al., 2016; Rosenzweig, 2009; Wiengarten et al., 2015; Wiengarten et al., 2013). Through collaborative business processes, the organizations can create dynamic and flexible integrations in order to synergistically adapt to changing conditions, which allows an improvement on performance and to remain competitive in the global market (Liu et al., 2009).

The SCOR model developed by the Supply Chain Council, in order to contribute to the improvement and dissemination of good practices in supply chains (APICS, 2020), provides a common language directed to processes and to the communication among partners of the SC in the following areas of decision-making: planning, supply, production/operation, distribution, and the processes related to reverse logistics (Lockamy & McCormack, 2004a). The business process associated with reverse logistics was not the focus for the performance analysis due to a series of barriers that limit its usage in Brazilian organizations, such as the governance structure to manage this system (Couto & Lange, 2017).

The SCOR model can be seen as an strategic tool to describe, communicate, implement, control and measure complex processes of the SC to improve the performance (Li et al., 2011). Furthermore, it has been widely used for processes’ performance measurement and SC optimization in recent years (Dissanayake & Cross, 2018; Ivanov et al., 2018; Li et al., 2011; Lockamy & McCormack, 2004a; McCormack et al., 2008; Trkman et al., 2010; Wang et al., 2004).

Thus, for the development of this study, the SCOR model was chosen to measure the performance level of business processes of the companies participating in the SC. This choice was due to the SCOR’s orientation for the process and the broad adoption by the academic and professional communities of the SC (Lockamy & McCormack, 2004b; McCormack et al., 2008).

The questionnaire was firstly prepared based on the literature review to identify measurement scales validated in research performed by other authors, which properly represented the constructs involved in the research. Regarding SCI it was used as a reference a scale adopted in the study conducted by Liu et al. (2016) to measure the indicators that compose their four constructs. The questionnaire adopted a five-point interval scale, with options that varied from 1 (“strongly disagree”) to 5 (“strongly agree”).

To measure OPP, it was adopted the scale used by Trkman et al. (2010) based on processes of the SCOR model, considering indicators that focus on the following business processes: planning, supply, production/operation and distribution. It was also used a five-point interval scale, in which the respondents were asked, for each business process, if “in general, it has a very good performance”, with response options that varied from 1 (“strongly disagree”) to 5 (“strongly agree”).

All indicators that compose the conceptual model are reflexive, as described in Table 1. In addition, the questionnaire included items that aimed to characterize the companies’ profile, regarding sector, size and location, as well as the respondent’s profile, considering the job position/function and field of practice.

Table 1 -
Description of measurement scales

Source: elaborated by the authors (2020).

Considering that the measurement scales quoted are originally in the English language, the questions were translated to Portuguese, and it was performed a consistency analysis of the indicators regarding the criteria of clarity, reach, acceptability (Rea & Parker, 2000), reliability, validity and functionality (Marconi & Lakatos, 2010). Subsequently, a pre-test of the questionnaire was conducted, involving professionals from the academy and managers with experience in the theme of business process management. The objective was to ensure a correct understanding regarding the questions and its content validity.

The scientific basis used for collection of secondary data that supported the literature review was mainly the Scopus. Regarding primary data, the collection was operationalized in the period from September to December 2019, along companies of the industry, retail, transportation and services sectors, including different segments.

To determine the sample size, it was used the general rule established by Hair et al. (2014), in which the size should be ten times bigger or equal to the number of indicators of the construct that has the greater amount of formative indicators to the measurement model, or the sample size should be ten times bigger or equal to the number of the greater amount of paths directed to an specific construct of the structural model. Therefore, considering the conceptual model proposed, the minimal sample size should be 40 responses.

The questionnaire was sent for completing online and directed to one manager of each participant organization, with focus on key professionals with knowledge in the areas of business strategies, and/or operations management, and/or administrative and financial management, and/or innovation and information technology. For all cases sent it was performed a new request for participation in the research, within fifteen days after the first submission. Therefore, in the end of the data collection it was obtained 92 total responses.

In two situations, it was observed a suspicious behavior in the responses, considering that the participants answered the same score for all questions. These two cases were eliminated from the database. Thus, 90 observations were considered as valid responses.

First, the descriptive analysis was developed to characterize the profile of respondents and companies. Subsequently, it was developed the multivariate statistical analysis, more specifically the Structural Equation Modeling - SEM, through the use of the software Smart PLS 3.0 student version (Ringle et al., 2015).

It is important to highlight that the conceptual model proposed in this study considers the construct SCI as a second-order construct, in which it was assumed all indicators of first-order constructs in the second-order construct, a method known as repetition of indicators (Hair et al., 2014). Thus, the measurement model was tested to evaluate the reliability and validity, specifically through analyses of unidimensionality and internal reliability of the indicators, convergent validity and discriminant validity.

To analyze the structural model, it was performed collinearity tests among the independent latent variables, tests of significance and relevance of path coefficients established between constructs, as well as the analysis of the coefficient of determination (R²) and effect size f². For all statistical tests performed, it was used the significance level (α) of 0.05.

The measurement model adopted in this research has the following constructs, composed of their respective reflexive indicators, according to Figure 2.

Figure 2 -
Measurement model.
Source: elaboratedby the authors (2020).

Thus, the indicators that compose the first-order constructs were submitted to quality tests proposed by Hair et al. (2014), as described in the methodological procedure.

To analyze internal consistency, it was used as a reference Cronbach’s alpha, the composite reliability and DillonGoldstein’s rho coefficient.

Table 2 -
Reliability of the constructs’ internal consistency

Source: elaborated by the authors (2020).

According to Table 2, all constructs presented adequate internal consistency and unidimensionality for the statistical criteria analyzed.

The analysis of convergent validity was performed considering the outer loadings and the average variance extracted - AVE related to the constructs.

Table 3 -
Convergent validity - outer loadings of the indicators and AVE

Source: elaborated by the authors (2020).

According to Table 3, the indicators Q1 and Q9 presented values for outer loadings lower than 0.708, which indicates a problem of lack of commonality of these indicators in their respective constructs (Hair et al., 2014). When disregarding those indicators of the measurement model, the AVE values of the respective constructs that contain the indicators presented an increase from 0.597 to 0.684 for the construct “Information integration”, and from 0.600 to 0.644 for the construct “Synchronized planning”.

Hair et al. (2014) argue that when there is a considerable variation in AVE, it is recommended the removal of the indicators. Thus, indicators Q1 and Q9 were excluded from the measurement model for the benefit of the criterion associated with convergent validity. Therefore, the results of the following statistical tests were obtained considering the exclusion of those indicators.

The discriminant validity was analyzed based on indicators’ cross loadings, and the comparison between square roots of the constructs’ AVE values. Table 4 shows that discriminant validity was reached for all indicators.

Table 4 -
Discriminant validity - cross loadings.

Source: elaborated by the authors (2020).

In addition, the discriminant validity was analyzed according to the criterion established by Fornell-Larcker, a second and more conservative approach (Hair et al., 2014). This criterion was fully achieved (Table 5).

Table 5 -
Discriminant validity - Fornell-Larcker criterion

Source: elaborated by the authors (2020).

Henseler, Ringle e Sarstedt (2015) recommend for the discriminant validity to be tested also using the HTMT method (Heterotrait-Monotrait), since it is more rigorous.

Table 6 -
Discriminant validity - HTMT criterion

Source: elaborated by the authors (2020).

The coefficient obtained between the constructs “Information integration” and “Strategic partnership” was 0.981 (Table 6), above the limit established (0.9). However, in situations where the loading values that compose the constructs are high and homogeneous, which was a condition that has been met by both constructs, it cannot be affirmed the lack of discriminant validity between them (Henseler et al., 2015).

Therefore, after performing the tests to evaluate the measurement model of this research and considering the changes made in the instrument to better fulfill the statistical criteria established in the literature, the reviewed measurement model properly enables the analysis of the structural model proposed.

The results shown on Table 7 do not demonstrate the non-observance of collinearity among the constructs, because the variance inflation factors - VIF of constructs were less than five, considering the evaluation criterion established by Hair et al. (2014).

Table 7 -
VIF values.

Source: elaborated by the authors (2020).

Therefore, the specific tests to evaluate the structural model can be conducted.

The path coefficients observed in the relations among constructs that compose the structural model were all significant, according to Table 8. The relation between the construct SCI and OPP was significant, with a path coefficient of 0.720, which indicates a strong and positive relation between these constructs. Thus, the hypothesis H1 was confirmed.

Table 8 -
Path coefficients among constructs and significance analysis (α = 0,05).

Source: elaborated by the authors (2020).

Furthermore, the path coefficients of the first-order constructs - “Information integration”, “Synchronized planning”, “Operational coordination” and “Strategic partnership” - were positive and significant (Table 8), which corroborates the conceptual premise that they compose the construct SCI. The construct “Synchronized planning” (β=0.367; p<0.05) presented the higher path coefficient, followed by “Strategic partnership” (β=0.359; p<0.05), “Information integration” (β=0.206; p<0.05) and “Operational coordination” (β=0.203; p<0.05).

Regarding the construct OPP, the adjusted R² value was equal to 0.514, which indicates that 51.4% of the variation observed in this construct is explained by the construct SCI. This result corroborates with previous studies, and in general it had a higher predictive accuracy compared to different R² obtained by the analysis of the relationship between SCI and organizational performance, as shown further in the discussion section. After the evaluation of the structural model, Figure 3 presents the schematic representation of the model, as well as the statistical results reached.

Figure 3 -
Structural model results.
Source: elaboratedby the authors (2020).

Regarding the evaluation of the effect size of first-order constructs on the OPP, the results point out that the constructs “Information integration”, “Synchronized planning”, “Operational coordination” and “Strategic partnership” presented sizes of 0.01, 0.07, 0.02 e 0.02, respectively. Thus, the effect sizes of first-order constructs that compose SCI can be considered small (Hair et al., 2014).

The results have confirmed the conceptual premise, in the sense that the constructs “Information integration”, “Synchronized planning”, “Operational coordination” and “Strategic partnership” compose SCI, corroborating the results obtained on previous studies (Cao & Zhang, 2011; Liu et al., 2016). The constructs “Synchronized planning” and “Strategic partnership” present, in this order, the higher path coefficients, indicating a positive and relevant influence in SCI. Thus, this result point out that in order to have an integrated and collaborative supply chain, the organizations that compose it should have aligned strategic goals, that favor the integration of business processes and the practice of synchrony of decisions regarding joint planning of actions in the level of market and products (Cao & Zhang, 2011; Liao & Kuo, 2014).

Furthermore, in line with this findings, Harland et al. (2007) empirically emphasize that the strategic alignment among business partners is a decisive factor for the integration among large-sized companies in relation to small and medium-sized. The non-compliance with this alignment is characterized as a barrier to SCI.

Similarly, in the study conducted by Cao e Zhang (2011), the congruence of goals among business partners positively influences the integration of the SC. Moreover, it facilitates the understanding and anticipating needs of the industry, in order to better meet the operational requirements (Flynn et al., 2010).

The construct “Synchronized planning” is important in the SC management, since its absence makes it difficult to have integration among organizations and, consequently, to the operation of business processes. Therefore, “Synchronized planning” is essential, it favors the optimization of companies’ resources in order to achieve an specific set of goals, through the practice of mutual planning and decision making among collaborative business partners (Cao & Zhang, 2011).

Lockamy and Mccormack (2004a) concluded that the planning processes are important in all areas of decision-making involved in the business processes of the SCOR model, from the strategical level to the operationalization. Thus, they are necessary to determine the most effective way to use the organization’s resources, in order to obtain better results for the companies.

The constructs “Operational coordination” and “Information integration” presented lower path coefficients, however, they also present a positive and significant effect in SCI. Considering the increasing amount of transacted data among organizations, boosted by the technological systems, it is becoming more and more relevant for organizations to expand their analytical capability with the purpose of strengthen the integrated management of the SC. Thus, resource sharing, including equipments, technologies and information, is an important activity for the leverage process of capabilities and assets with partners of the SC, with a positive impact on companies’ performance.

The path coefficient that measures the impact of SCI in OPP proved to be significant and relevant (β=0.720 p<0.05). This result corroborates with previous researches that analyzed the effect of SCI in the company’s performance (Devaraj et al., 2007; Flynn et al., 2010; Lee et al., 2011; Liu et al., 2016; Panahifar et al., 2018; Rosenzweig, 2009; Sundram et al., 2016; Wiengarten et al., 2013).

Regarding the coefficient of determination, it was found that 51.4% of the variation observed in the construct OPP is explained by the construct SCI. In studies that evaluated the impact of SCI in the performance of companies using operational indicators, it was verified R² values that ranged of 0.269 (Devaraj et al., 2007), 0.255 (Flynn et al., 2010), 0.456 (Rosenzweig, 2009) and 0.41 (Liu et al., 2016). Concerning the impact on business performance, measured through indicators associated to an increase in sales, and/or profit, and/or market share, and/or return on investments, Flynn et al. (2010) obtained a coefficient of determination of 0.131. This value is lower than the coefficient of determination reached when the performance was measured through operational indicators.

On the other hand, in the research conducted by Panahifar et al. (2018) that assessed the impact of interorganizational collaboration on the business performance jointly indicated through financial and operational indicators, it was obtained an R² value of 0.63. Thus, when observed the R² value of this research (0.514), measured in the level of business processes, it was superior to most coefficients of determination found in previous studies, except for the study by Panahifar et al. (2018).

Thus, while recognizing that there are differences between the indicators used in these studies to measure organizational performance, the analysis of the coefficient of determination corroborates the assumption that the effect of SCI in companies’ performance should be examined in the level of business processes. In this level the resources are allocated and explored to implement strategies, and the first order results are frequently observed (Jeffers et al., 2008).

Regarding the result of effect size (f²), it was observed no substantial impact of the first-order constructs on OPP, since the effect sizes were all small. This result corroborates the complementary and interdependent character of SCI, aligned with previous studies that emphasize the importance of conceptualize the SCI as a multidimensional construct in order to evaluate the effects on companies’ performance (Cao & Zhang, 2011; Flynn et al., 2010; Liu et al., 2016). In other words, the full SCI probably will not be achieved without the complete development of all constructs that compose it.

gonzalezinayara@gmail.comrinaldocastrooliveira@gmail.com