INTENTION TO ADOPT BIG DATA IN SUPPLY CHAIN MANAGEMENT: A BRAZILIAN PERSPECTIVE

ABSTRACT Big data applications have been remodeling several business models and provoking strong radical transformations in supply chain management (SCM). Supported by the literature on big data, supply chain management, and the unified theory of acceptance and use of technology (UTAUT), this study aims to evaluate the variables that influence the intention of Brazilian SCM professionals to adopt big data. To this end, we adapted and validated a previously developed UTAUT model. A survey of 152 supply chain respondents revealed that facilitating conditions (e.g., IT infrastructure) have a high influence on their intention to adopt big data. However, social influence and performance expectancy showed no significant effect. This study contributes to the practical field, offering valuable insights for decision-makers considering big data projects. It also contributes to the literature by helping minimize the research gap in big data in the Brazilian context.


INTRODUCTION
The rapid advancement of information and communication technologies (ICTs) has motivated logistics and supply chain management practitioners and scholars (Zinn & Goldsby, 2017b, 2017a to understand the role of these new technologies, and to determine how organizations can capture value through ICT adoption. A highly disruptive and significant technology that has emerged recently is big data (Davenport, 2006;Manyika et al., 2011;Rotella, 2012). The amount of data produced everyday has been increasing drastically (Domo, 2017). This growth has imposed several complexities concerning its management. In this context, big data offers a powerful approach to helping organizations analyze (Croll, 2015) large amounts of data to provide insights into the decision-making process (Abawajy, 2015).
The literature considered big data the "next big thing in innovation" (Gobble, 2013, p. 64) and "the fourth paradigm of science" (Strawn, 2012, p. 34). Big data has impacted practically all business models. For instance, 35% of Amazon.com's revenue is generated through the use of big data (Wills, 2014), along with the remodeling of marketing activities that capture rich data on consumer behavior in real-time (Erevelles, Fukawa, & Swayne, 2016).
Despite the potential benefits of employing big data in supply chain management (Hazen, Boone, Ezell, & Jones-Farmer, 2014;Kache & Seuring, 2017;Schoenherr & Speier-Pero, 2015), awareness of and initiatives on big data in the Brazilian SCM context are rare, and the literature lacks strong empirical results (Queiroz & Telles, 2018). The current initial stage of big data adoption presents an opportunity for scholars and practitioners to fill this gap. For example, to the best of our knowledge, no previous study analyzed the intention of Brazilian SCM professionals to adopt big data. To bridge this gap, this study provides an in-depth understanding of Brazilian supply chain professionals' intention to use big data. We adapt a previously developed and validated unified theory of acceptance and use of technology (UTAUT) model (Venkatesh, Morris, Davis, & Davis, 2003;Queiroz & Wamba, 2019), by including a trust construct. More specifically, this study answers the following research question: How do the variables from the UTAUT model explain Brazilian SCM professionals' intention to adopt big data?
To answer this question, this work draws on the literature on big data (Davenport, 2006;Manyika et al., 2011;Queiroz & Telles, 2018), supply chain management (Carter, Rogers, & Choi, 2015;Mentzer et al., 2001), and UTAUT (Venkatesh et al., 2003;Venkatesh, Thong, & Xu, 2012;Queiroz & Wamba, 2019) to develop the hypotheses and model. The conceptual model was adapted and validated with partial least squares structural equation modeling (PLS-SEM). The main findings offer strong theoretical and managerial implications. From the managerial perspective, we verified that facilitating conditions (e.g., infrastructure) exert high influence on the behavioral intention of big data adoption.
From the theoretical lens, our findings revealed that neither social influence nor performance expectancy are good predictors of the behavioral intention of big data adoption in Brazilian SCM professionals.
This paper is organized as follows: next, we present the emerging theoretical foundations for big data research, SCM, and UTAUT. Then, the hypotheses and the research model are described, followed by the survey methodology and analysis using PLS-SEM. That is succeeded by a discussion on managerial and theoretical implications as well as limitations of the current work and directions for future research. Finally, our conclusions are elucidated.

Big Data: Fundamentals, concepts, and challenges
Big Data has emerged as a highly disruptive information and communication technology (ICT). A well-articulated and suitable definition of Big Data is "[…] datasets whose size is beyond the ability of typical database software tools to capture, store, manage, and analyze" (Manyika et al., 2011, p. 1). Thus, Big Data can be regarded as providing a robust approach to exploring data in the context of descriptive, prescriptive, and predictive decisions (Phillips-Wren & Hoskisson, 2015). This approach is commonly called Big Data analytics (BDA), and is represented by a 5V approach (volume, velocity, variety, veracity, and value) (Queiroz & Telles, 2018;Wamba et al., 2017). In other words, BDA uses sophisticated statistics, mathematical and computational techniques to explore a large set of data to provide insights to decision-makers. In this study, we use the definition of Big Data proposed by Phillips-Wren and Hoskisson (2015).The authors described Big Data as data that overtake the organization's capabilities, regarding storage, and analysis to support and bring insights to the decision-making process.
The volume of data has increased drastically in recent years, mainly because of the variety of data produced today (Bibri & Krogstie, 2017) (e.g., ERP systems, Twitter, Facebook, Google, Linkedin, GPS, among others) and the velocity of its spread (Munshi & Mohamed, 2017;Srinivasan & Swink, 2018). This complex scenario impels organizations to develop distinctive capabilities for storing, processing, and analyzing data to support the decision-making process. However, creating value is not a trivial task, mainly because of organizations' limited capacity to process and analyze a variety of data. Moreover, data veracity, which indicates data quality and trustworthiness (Munshi & Mohamed, 2017;Nobre & Tavares, 2017), seems to be a huge challenge for organizations.

Supply chain management and the impacts of cutting-edge technologies
Recently, the logistics and SCM fields have been significantly impacted by the exponential growth in ICT usage. Accordingly, scholars and practitioners have strived to understand its potential effects and application opportunities in their business models (Zinn & Goldsby, 2017a, 2017b. In this context, SCM is defined as: The management of a network of relationships within a firm and between interdependent organizations and business units consisting of material suppliers, purchasing, production facilities, logistics, marketing, and related systems that facilitate the forward and reverse flow of materials, services, finances and information from the original producer to final customer with the benefits of adding value, maximizing profitability through efficiencies, and achieving customer satisfaction (Stock & Boyer, 2009, p. 706).
Moreover, SCM can be viewed as a network (Carter et al., 2015) as well as a complex adaptive system (Choi, Dooley, & Rungtusanatham, 2001), and this complexity has impacted the increasing amount of data. Considering the use of Big Data in SCM, it is clear that it assists in the decision-making process by providing powerful insights into SCM dynamics (e.g., customer buying patterns, cost analysis, market trends). With the help of robust prescriptive and descriptive analysis (G. Wang et al., 2016), businesses have witnessed many cases of significant performance enhancement (Akter et al., 2016;Gunasekaran et al., 2017).

Technology acceptance models (TAMs) and Unified theory of acceptance and use of technology (UTAUT)
Scholars have studied the diffusion and proliferation of information technology (IT) (Davis, 1989;Wamba, 2018;Morris & Venkatesh, 2000;Venkatesh & Brown, 2001) in terms of individuals' beliefs and behavior toward their adoption and use (Mamonov & Benbunan-Fich, 2017;Youngberg, Olsen, & Hauser, 2009). The technology acceptance model (TAM) is a seminal and influential contribution in technology adoption (Davis, 1989), with its roots in the theory of reasoned action (TRA) (Azjen & Fishbein, 1980). The core of the TAM resides in two latent variables: perceived usefulness (PU) and perceived ease of use (PEOU). More recently, Venkatesh et al. (2003) proposed the consolidation of the acceptance model theories leading previously into the unified theory of acceptance and use of technology (UTAUT).

UTAUT
The UTAUT model (Venkatesh et al., 2003) is a robust and influential approach to understanding technology adoption and use at the individual behavior level. The model has four constructs directly focused on technology's intended use: performance expectancy, effort expectancy, social influence, and facilitating conditions.
Performance expectancy refers to "the degree to which an individual believes that using the system will help him or her to attain gains in job performance" (Venkatesh et al., 2003, p. 447). Effort expectancy is "the degree of ease associated with the use of the system" (Venkatesh et al., 2003, p. 450). Social influence denotes "the degree to which an individual perceives that important others believe he or she should use the new system" (Venkatesh et al., 2003, p. 451). Finally, facilitating conditions indicates "the degree to which an individual believes that an organizational and technical infrastructure exists to support use of the system" (Venkatesh et al., 2003, p. 453). The UTAUT model also has four moderators: gender, age, experience, and voluntariness of use. However, following a previous study (Weerakkody, El-Haddadeh, Al-Sobhi, Shareef, & Dwivedi, 2013), we do not use these moderators in our adapted model (explained in the next section) because this is a preliminary study of BDA adoption in the Brazilian SCM context.

Hypotheses and research model
Supported by the emerging literature on Big Data, SCM, and UTAUT, we adapted a recent model reported in Queiroz and Wamba (2019) to comprehend the Big Data adoption behavior of Brazilian supply chain professionals. We adopted some of the constructs and hypotheses proposed in Queiroz and Wamba´s (2019) model ( Figure 1) as these have been adopted and validated by previous studies (Exhibit 1). To these previous constructs reported in Queiroz & Wamba (2019) we added a trust construct, previously validated in the literature (Alalwan, Dwivedi, & Rana, 2017;Gefen, Karahanna, & Straub, 2003). Moreover, the constructs in our model have different relationships than the ones reported in the literature (Queiroz & Wamba, 2019).

Facilitating conditions
Facilitating conditions play a fundamental role in predicting user acceptance and usage behavior (Venkatesh et al., 2003(Venkatesh et al., , 2012. In this study, facilitating conditions denotes SCM professionals' knowledge of their organization's capabilities and infrastructure available to support the use of Big Data. Previous studies have reported that facilitating conditions are a good predictor of the behavioral intention of Big Data adoption (Huang, Liu, & Chang, 2012;Sabi, Uzoka, Langmia, & Njeh, 2016). In this study, we theorize that facilitating conditions, besides influencing behavioral intention directly, are critical in professionals' effort expectancy  and influence their performance expectancy (C. Wang, Jeng, & Huang, 2017). Therefore, we propose the following hypotheses:

Trust
The trust construct has been studied extensively in the business management and management information systems (MIS) fields (Colquitt & Rodell, 2011;K. Wu, Zhao, Zhu, Tan, & Zheng, 2011).
Trust is defined as "the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party" (Mayer, Davis, & Schoorman, 1995, p. 712). This definition implies that trust is a willingness to depend on the partner based on integrity, benevolence, and credibility. In this context, Big Data is trustworthy for users. In line with prior works (K. Wu et al., 2011), we hypothesize that: H2a: Trust positively affects performance expectancy.
H2b: Trust positively affects behavioral intention to adopt Big Data.

Social influence
As reported previously, social influence is a good predictor of technology behavioral intention and usage (Venkatesh et al., 2003). In this work, social influence denotes the extent to which SCM professionals believe their colleagues should use Big Data. Previous studies highlight social influence as a predictor of behavioral intention (Batara, Nurmandi, Warsito, & Pribadi, 2017;Venkatesh et al., 2012). Our study argues that in the SCM context, social influence relationships exert significant influence on trust (A. Chin, Wafa, & Ooi, 2009) and, in turn, on the behavioral intention (Alalwan et al., 2017). Thus, we propose the following hypotheses: H3a: Social influence positively affects trust.
H3b: Social influence positively affects behavioral intention to adopt Big Data.

Effort expectancy
Effort expectancy is related to the system's complexity of operation (Venkatesh et al., 2003). In this study, effort expectancy refers to the ease of use of Big Data systems for an SCM professional.
Previous studies discussed the direct effect of effort expectancy in the behavioral intention and usage of a new technology (Batara et al., 2017;Venkatesh et al., 2012;Y. Zhao, Ni, & Zhou, 2018) as well as in the adoption of blockchain in the SCM field (Francisco & Swanson, 2018). Accordingly, this study hypothesizes that:

Performance expectancy
In this work, performance expectancy denotes the level to which an SCM professional perceives that Big Data will improve his productivity and performance. With Big Data application, organizations can analyze different types of data employing powerful statistics and machine learning techniques (Kune, Konugurthi, Agarwal, Chillarige, & Buyya, 2016). This implies considerable time savings and productivity improvement for organizations, therefore helping enhance its performance Wamba et al., 2017). Thus, we propose that:

METHODOLOGY Sample and data collection
A survey instrument based on Queiroz and Wamba (2019) was used to test our proposed hypotheses. The web-based questionnaire was grounded on constructs and scales that have been validated by previous studies (Venkatesh et al., 2003(Venkatesh et al., , 2012Gefen et al., 2003

PE3
Using big data helps me accomplish tasks more quickly.

PE4
Using big data increases my productivity.

TR4
I feel assured that legal and technological structures adequately protect me from problems on big data.

TR5
Big data has the ability to fulfil its task.

RESULTS AND ANALYSIS
Partial least squares structural equation modeling (PLS-SEM) (Ringle, Wende, & Becker, 2015;Shim, Lee, & Kim, 2018;Sun & Teng, 2017) was applied to analyze the research model. PLS-SEM is a powerful approach for analyzing simple and robust models in business management (Hair, Sarstedt, Hopkins, & Kuppelwieser, 2014;Hair, Hult, Ringle, & Sarstedt, 2017), and has gained the attention of SCM scholars (Autry, Williams, & Golicic, 2014;Grawe, Daugherty, & Ralston, 2015;Han, Wang, & Naim, 2017;Yadlapalli, Rahman, & Gunasekaran, 2018). Its main advantages are its flexibility in working with small samples (e.g., 100 respondents) and its formative and reflective constructs (Hair et al., 2017). Table 1 reports the characteristics of the respondents. Male respondents comprised almost 90% of the total. Regarding age distribution, most respondents (52.63%) were aged 34-41 years. A total of 55.26% respondents had a postgraduate/ MBA-the highest education level in our sample-followed by 39.47% holding bachelor degrees and 5.26% holding a master of science degree. Considering the experience at their respective organizations, 50% respondents had worked there for 2-5 years, followed by 21.05% having worked for 6-10 years and 18.42% working for less than one year. Finally, 46.05% of the sample comprised logistics analysts, followed by 26.32% transportation managers, 18.42% operations managers, and 9.21% supply chain managers.
We analyzed the research model with SmartPLS 3.0 (Hair et al., 2017;Ringle et al., 2015). First, the model was assessed by its loadings, Cronbach's alpha, composite reliability, average variance extracted, and discriminant validity.

Measurement model
All outer loadings highlighted in Table 2 exceeded the 0.70 threshold recommended in the literature (Hair et al., 2017). Table 3 shows the main measures for construct reliability and internal consistency of items. In this study, both Cronbach's alpha value and composite reliability were above the 0.70 threshold, and all average variance extracted values were above the 0.50 threshold (Hair et al., 2017;Nunnally, 1978;Riffai, Grant, & Edgar, 2012). Therefore, all constructs in the model have their utilization validated. Table 4 presents the discriminant validity results. In this case, the square root of the average variance extracted for each construct should be greater than the correlations between the constructs (Fornell & Larcker, 1981;Henseler, Ringle, & Sinkovics, 2009). Our results are higher than the 0.70 threshold (Fornell & Larcker, 1981), confirming that all constructs show discrimination (Ahmad & Khalid, 2017;Martins, Oliveira, & Popovič, 2014 Table 5 and 6 present the results of our structural model. Table 5 highlights the path coefficients statistics.The findings indicated that FC has a significant positive effect on EE (β = 0.578, p < 0.001). Thus, H1a is supported. H1b hypothesized that FC has a significant positive effect on PE. The results (β = 0.380, p < 0.001) support H1b. H1c theorized that FC has a significant positive effect on BI. This hypothesis was also supported (β = 0.490, p < 0.001). Next, H2a argued that TR has a significant positive effect on PE. Our results (β = 0.413, p < 0.001) support this hypothesis. Then, H2b argued that TR has a significant positive effect on BI. The results supported H2b (β = 0.327, p < 0.05). H3a theorized that SI has a significant positive effect on TR.

Structural model
The results supported H3a (β = 0.710, p < 0.001). The rest of the hypotheses had unexpected results. H3b theorized that SI has a significant positive effect on BI. Surprisingly, the relationship was found to be negative and non-significant. Therefore, H3b was not supported (β = -0.073, p = 0.519). H4 argued that EE has a significant positive effect on BI. This hypothesis was not supported either (β = 0.166, p < 0.1). Next, H5 theorized that PE has a significant positive effect on BI. Surprisingly, the results (β = -0.214, p < 0.05) showed a negative significant effect on BI.
Thus, H5 was not supported.

DISCUSSION AND IMPLICATIONS
The main objective of this study was to gain an in-depth understanding of the intention of Big Data adoption in the Brazilian supply chain context. In light of the lack of Brazil-based studies on cutting-edge technologies (Queiroz and Telles, 2018), this work contributes to advancing the literature on BDA, SCM, and TAMs. The results offer significant managerial and theoretical implications as well as valuable directions to adapt and extend the adopted model.

Managerial implications
We believe the main findings of this study provide important implications for managers and practitioners interested in gaining deeper insights about BDA in SCM and their adoption enablers. In line with the literature that regards Big Data as an essential tool to improve supply chain performance Hazen, Skipper, Ezell, & Boone, 2016;G. Wang et al., 2016) (Sabi et al., 2016;Venkatesh et al., 2003). Surprisingly, despite the literature reporting performance expectancy as a good predictor of behavioral intention towards technology adoption Farooq et al., 2017;Venkatesh et al., 2003;Weerakkody et al., 2013) This study suffers from some limitations. We believe that, first, a moderator variable could be incorporated into the model (Venkatesh et al., 2003(Venkatesh et al., , 2012 to capture the nuances and differences in the sample, such as industry, gender, and experience. Second, because of the scarcity of Brazilian studies on Big Data adoption, our findings cannot be compared with other similar works in this context. However, it opens up opportunities for scholars and practitioners to apply the validated model and to adapt it to other contexts. Third, the adopted model was tested in an emerging economy; because of this, the results cannot be generalized globally. Consequently, obtaining more empirical evidence by applying the adopted model in other countries could be an exciting stream for future research. Finally, this study was one of the first attempt to understand the behavioral intention to adopt Big Data by Brazilian SCM professionals. There is an urgent need and opportunities for additional investigations on this and other cutting-edge technologies (e.g., blockchain, internet of things, 3D printing, etc.), regarding the relationship, as also compare the hypotheses of this model into other contexts.

CONCLUSION
The purpose of this study was to gain an in-depth understanding of Big Data behavior intention among Brazilian SCM professionals and to adjust and apply a model that captures the constructs of adoption behavior. In this regard, our study contributes to a more thorough understanding of the intention to adopt BDA in the Brazilian SCM field.
The contributions of this study are threefold. First, supported by a strong theoretical literature (Akter et al., 2016;Alalwan et al., 2017;Davis, 1989;Venkatesh et al., 2003Venkatesh et al., , 2012Queiroz & Wamba, 2019) we adapted and applied a model to understand behavioral intention concerning Brazilian SCM professionals. Second, our findings provide strong implications for theory and practice. For instance, one implication is that only facilitating conditions, and trust were good predictors of behavioral intention. Contrary to findings of previous studies (Venkatesh et al., 2003(Venkatesh et al., , 2012, social influence was not a predictor of behavioral intention, but this result is in line with the recent findings reported by Alalwan et al. (2017). Third, both performance expectancy and effort expectancy were not good predictors of behavioral intention. This interesting finding opens up opportunities to further exploration of this behavior. Finally, our study contributes to fill a gap in the Brazilian empirical literature on Big Data in SCM, while simultaenously motivates logistics and SCM scholars to advance this stream of research.