Knowledge absorption capacity's efficacy to enhance innovation performance through big data analytics and digital platform capability

Khan, Adnan; Meng, Tao Article Knowledge absorption capacity's efficacy to enhance innovation performance through big data analytics and digital platform capability Journal of Innovation & Knowledge (JIK) Provided in Cooperation with: Elsevier Suggested Citation: Khan, Adnan; Meng, Tao (2022) : Knowledge absorption capacity's efficacy to enhance innovation performance through big data analytics and digital platform capability, Journal of Innovation & Knowledge (JIK), ISSN 2444-569X, Elsevier, Amsterdam, Vol. 7, Iss. 3, pp. 1-13, https://doi.org/10.1016/j.jik.2022.100201 This Version is available at: https://hdl.handle.net/10419/327171 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ Knowledge absorption capacity’sefficacy to enhance innovation performance through big data analytics and digital platform capability Adnan Khan a,b , Meng Tao c, * a School of Management, Jiangsu University, Zhenjiang, 212013, China b School of Business Administration, Dongbei University of Finance and Economics, Dalian, China c Dean at International Business College, Dongbei University of Finance and Economics, Dalian, China ARTICLE INFO Article History: Received 28 December 2021 Accepted 4 May 2022 Available online 17 May 2022 ABSTRACT The 2018 Global Innovation Index ranks Pakistan 118 out of 126 in innovation. One of the main reasons why developing countries, such as Pakistan, fail to innovate is their improvisation of astute and concurrent knowledge. This study explores the contemporary hurdles that lead to manufacturing firms’low agility and innovation performance. Based on the theory of dynamic capability view and the theory of absorptive capacity, we propose that the knowledge absorption capacity of firms can help them organize or utilize dynamic capabilities, such as big data analytics and digital platform capability, to enhance their agility and innovation performance. However, in the presence of a diversified organizational culture (i.e., flexibility orientations and datadriven culture), the desired outcomes may be affected. For this purpose, this study performed a questionnaire survey to collect data for validating the theoretical model. The collected responses from 325 manufacturing firms were analyzed using structural equation modeling, and empirical results reveal a positive relationship between the knowledge absorption capacity, agility, and innovation performance of firms mediated by big data analytics and DP capabilities. Flexibility orientations also showed a significant moderating role, but the role of data-driven culture was not significant. Statistical results reject the hypothesis. This study enriches the scope of the theories mentioned above and comes up with several other interesting theoretical and managerial implications valuable for academicians and policymakers. © 2022 The Authors. Published by Elsevier España, S.L.U. on behalf of Journal of Innovation & Knowledge. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Keywords: Knowledge absorption capacity Big-data analytics capability DP capability Firm’s agility Innovation performance Data-driven culture JEL codes: O3 O320 O310 Introduction The 2018 Global Innovation Index names Pakistan as one of the least innovative countries globally; out of 126 countries, Pakistan only ranked 117th in 2017 and 113th in 2018 (Global Innovation Index, 2018). Similarly, the manufacturing industry only contributes 14% to the total GDP of Pakistan (Economic Survey of Pakistan, 2019). Developing countries have to deal with issues in their technologies, human skills, institutional mechanisms, and infrastructures that hinder their innovation efficiency. Innovation is often explained as a radical and incremental innovation (Varis & Littunen, 2010). Similarly, innovation performance may be defined as upgrading the firm’s products, services, or processes (Flor, Cooper, & Oltra, 2018). The manufacturing sector of Pakistan contributed about 13.5% to 13.8% on average to the country’s GDP over the past decade. However, this sector only witnessed a 13% growth in the latest fiscal year. Both large-scale manufacturing (LSM) and small-scale manufacturing (SSM) contribute to the manufacturing sector and GDP of Pakistan; LSM contributes about 78% and 10.2% to manufacturing and GDP, whereas SSM contributes about 2.0% in both (Economic Survey of Pakistan, 2019). The inconsistent growth of the manufacturing sector of Pakistan may be ascribed to several reasons, but no previous research has explored this problem in-depth. This empirical work aims to solve this problem by boosting the innovation performance of manufacturing firms. Specifically, this study proposes that manufacturing firms organize dynamic capabilities that can enhance their agility and innovation performance, such as big data analytics capability (BDAC) and DP capability (DPC). BDA has changed the traditional dynamics of businesses and significantly improved their performance. According to Dubey et al. (2019), big data and predictive analytics can improve the performance of manufacturing firms and enhance both their organizational performance (Purgat & Mrozek, 2018) and competitive advantage (Shan, Luo, Zhou, & Wei, 2019). Previous studies have highlighted the influential role of BDAC. However, no study has explored the mediating role of BDAC in the relationship between the knowledge absorption capacity (KAC) and FA of an organization. Considering the impact of BDAC on both academia and industry, this * Corresponding author. E-mail addresses: [email protected] (A. Khan), [email protected] (M. Tao). https://doi.org/10.1016/j.jik.2022.100201 2444-569X/© 2022 The Authors. Published by Elsevier España, S.L.U. on behalf of Journal of Innovation & Knowledge. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Journal of Innovation & Knowledge 7 (2022) 100201 Journal of Innovation &Knowledge https://www.journals.elsevier.com/journal-of-innovation-and-knowledge research sheds light on the mediating role of the BDAC of manufacturing firms in enhancing FA and innovation performance. DPs (DPs) are replacing the traditional ways of interaction between businesses and end-users. For example, IOS and Android platforms provide multiple features and apps with convenience, whereas payment platforms, such as Alipay, WeChat, PayPal, and Apple Pay, offer an unmatchable and valuable contribution to the financial industry. Peer-to-peer DPs, such as Airbnb, Uber, and Task Rabbit, are also gaining popularity. DPs can be used as a dynamic capability for manufacturing firms to enhance their agility and innovation performance. This study distinguishes itself from the extant literature by exploring the mediating role of dynamic capabilities (BDAC and DPC). Furthermore, this study proposes that the attributes of a firm’s dynamic capabilities are established due to this firm’s KAC, which may urge the firm to organize dynamic capabilities. KAC can be defined as a firm’s ability to acquire, assimilate, transform, and exploit knowledge to bolster its performance. Acquisition and assimilation are associated with a firm’s potential absorptive capacity, whereas exploitation and transformation are associated with its realized absorptive capacity. Previous studies have explored the versatile outcomes of AC, such as knowledge utilization (Vasudeva & Anand, 2011), OAC and responsiveness of firms (Liao, Welsch, & Stoica, 2003), KAC and environmental performance (Shahzad et al., 2020), and integration of external knowledge and AC to improve radical innovation (Flor et al., 2018). Integrating firms’dynamic capabilities to enhance their FA and innovation performance should also be considered under the influential role of organizational culture (OC). OC will either support the flow of knowledge or vice versa, given that culture is unavoidable in any organizational outcome (Smircich, 2017). This research expands the idea of Dubey et al. (2019), who argued that OC facilitates the transformation of BDAC to enhance an organization’s performance. This study defines two critical traits of OC, namely, flexibility orientations and data-driven culture (Dubey et al., 2019). An organization’sflexible orientations will influence the effectiveness of KAC in building BDAC (i.e., flexible orientations will positively affect manufacturing firms to equip themselves with BDAC). Meanwhile, control orientations, where firms follow the norms and adopt typical decision-mak- ing mechanisms from the top management, may not drive firms to equip themselves with any ICT-enabled capabilities, such as BDAC. Therefore, this study further broadens the discussion on the role of DDC as a moderator and mediator in the relationship between BDAC and innovation performance. DDC may be influential in either way to transform the outcomes of BDAC and enhance innovation performance. Given the severity of ongoing issues related to the pace of innovation, this study aims to determine how manufacturing firms’ agility and innovation performance can be enhanced. The following research questions are therefore proposed: According to the dynamic capability view (DCV), what are the roles of BDAC and DPC in enhancing the agility of firms? How does FA enhance the innovation performance of manufacturing firms? How does OC (flexibility orientations) moderate the relationship between knowledge absorption capacity and BDAC, and how does DDC moderate the relationship between BDAC and innovation performance? Theory of absorptive capacity In 1990, Cohen and Levinthal introduced the AC theory to explore afirm’s capacity to recognize and value knowledge from external sources, organize and decode such knowledge, and use it effectively to achieve its goals (Tseng, Pai, & Hung, 2011). In the proposed conceptual model, KAC is derived from the theory of AC. Dynamic capability view DCV elaborates on the theme of a resource-based view and posits that “Dynamic capabilities bridge the gap between the firm’s resources and changing business environment”(Barney, 1991b). Unlike RBV, DCV emphasizes building and adopting the necessary capabilities in response to external environmental changes. BDAC and DPC are extracted from DCV to represent the dynamic capabilities in this study. Hypotheses Knowledge absorption capacity and big data analytics capability AC is a vital capability of firms to organize several needed capabilities (Shahzad et al., 2020). BDAC has been used in product or service innovation, production and manufacturing, marketing and management, and business growth (Ritala, Olander, Michailova, & Husted, 2015). A firm’s performance is highly dependent on its effectiveness in processing and interpreting data (Premkumar, Ramamurthy, & Saunders, 2005). A firm needs a set of tangible and intangible resources in technology, culture, technical and managerial skills, and human resources (Chen & Storey, 2012;Tambe, 2014). Janssen, van der Voort, and Wahyudi (2017) argued that the analytic capability complements big data management. Moreover, the performance of a firm is highly dependent on its effectiveness in processing and interpreting data (Premkumar et al., 2005). Big data is becoming an integral component of Industry 4.0, a concept that a German industrialist proposes to represent the fourth industrial revolution (Shamim et al., 2019). Developing economies attempt to accomplish business competency through value creation by using big data (George, Haas, & Pentland, 2014). Moreover, the view of DCV underscores the significance of recreating and renewing the strategic capabilities of a firm to keep abreast with the changing technology-driven business environment (Pisano, 2017). Although big data helps policymakers decide based on what they know instead of what they believe (McAfee & Brynjolfsson, 2012), the relevant KAC of a firm facilitates the exploitation of BDAC (Zeng & Glaister, 2018). Based on these arguments, the following hypothesis is proposed: H1: KAC will positively affect the BDAC of manufacturing firms. Knowledge absorption capacity and DPs capability Knowledge absorption is a continuous process; modern firms have developed multiple channels to absorb concurrent knowledge. Afirm’s capacity to use such knowledge greatly depends on exploiting existing knowledge. Apart from big data, machine learning, artificial intelligence, and the Internet of Things, DPC also prioritizes resource allocation. Most companies competing in a digital ecosystem (Subramaniam, Iyer, & Venkatraman, 2019) are based on AC reflecting their competency to acquire, integrate, transform, and utilize external knowledge and affecting platform capability’s adeptness (Ali, Seny Kan, & Sarstedt, 2016;Delmas, Hoffmann, & Kuss, 2011). With emerging technologies, the chances of achieving an optimal advantage depend on establishing a DP for understanding evolving technologies and on the capacity of a firm to undertake the risk of investing in such a platform to improve its business outcomes (Wang, Liang, ZHONG, XUE, & XIAO, 2012). A well-equipped platform supports firms in standardizing, managing, and allocating unprecedented levels of data (Yoo, Henfridsson, & Lyytinen, 2010). The platform capability of digitization not only has changed the means of building a competitive edge over the past two decades (Parker et al., 2016c) but also plays a vital role in defining the value proposition for all sizes of firms by allowing them to seek and handle data and information (Cenamor, R€ onnberg Sj€ odin, & Parida, 2017). Roberts and A. Khan and M. Tao Journal of Innovation & Knowledge 7 (2022) 100201 2 Grover (2012) theorized and discovered that DPC allows firms to sense and answer the demands and needs of customers commendably and thereby depends on the absorptive capacity of these firms. Based on these arguments, the following hypothesis is proposed: H2: KAC will positively affect the DPC of manufacturing firms. Knowledge absorption capacity and firm’s agility KAC utilizes different learning approaches to enhance a firm’s performance, such as exploitative, transformative, and exploratory learning (Lane, Koka, & Pathak, 2006). Assimilation and knowledge acquisition can be linked to potential KAC, whereas a firm’s ability to transform and assimilate this knowledge into its operations can be linked to realizing KAC (Ali et al., 2016;FLATTEN, GREVE, & BRETTEL, 2011). Firms that solely focus on exploitations may face difficulties sustaining their competitive performance (Volberda, Foss, & Lyles, 2010). The indirect relationship between KAC and FA has been measured in previous research. For example, Overby, Bharadwaj, and Sambamurthy (2006) found a connection between knowledge reach and richness of agility. Trantopoulos, Von Krogh, Wallin, and Woerter (2017) studied the relationship between IT and knowledge capabilities on the agility of a firm, and they highlighted some salient features of agile businesses, such as meeting customer requirements quickly, managing new products strategically, and completing organizational tasks on time. Therefore, this study assumes that KAC develops a proactive conception to respond to or organize dynamic capabilities, such as FA. The following hypothesis is then put forward: H3: KAC will positively affect the FA of manufacturing firms. Big data analytics capability and firm’s agility Agility refers to a firm’s ability to ascertain new opportunities, utilize its current knowledge, and adapt to abrupt business changes. Several IT-enabled studies argue that these capabilities positively influence firms’outcomes (Weill, Subramani, & Broadbent, 2002). Apart from the conventional agility concepts, a firm should also have the expertise to sense external changes and promptly respond to them (Seo, Paz, & A, 2008). Zhang and Dhaliwal (2009) investigated how the application of IT can enhance firm performance, whereas Bharadwaj (2000) examined the significance of information technology adoption as one of the primary differentiators among firms with varying performance levels. The firm’s agility resulting from its IT- enabled skills driven by big data interventions is mainly defined as its analytic expertise in information management (Kiron, Prentice, & Ferguson, 2014;Pavlou & Sawy, 2010). Big data analytics involve successfully processing data with large amounts, high velocity, and diverse types (Wamba et al., 2017), which improves FA. The following hypothesis is then proposed: H4: BDAC will positively affect the FA of manufacturing firms. DP capability and firm’s agility DPs play essential roles in various fields, ranging from functional technology to strategic management (Yeow, Soh, & Hansen, 2018). Technology platforms provide digital options for firms that enable them to react effectively to business or economic changes. Firms with DPCs enjoy the competitive edge of creating new networks to access their customers, integrating themselves into their supply chain partners in real-time, improving the efficiency of their domestic operations, and offering their customers modern digital services and products (Wheeler, 2002). Agility can be observed among those firms with superior platform capabilities to readily address their business process digitally (Sambamurthy, Bharadwaj, & Grover, 2003). DPs connect firms to various external information sources, allow them to establish ties in an inter-organizational network, and address their structural shortcomings. With the help of DPs, firms tend to evaluate the external market trends and respond to them rapidly by formulating strategies (Chi, Ravichandran, & Andrevski, 2010). Those firms connected to the digital network help other firms receive up-to-date information. DPC allows firms to rapidly develop or improve their products or services in a globally challenging market (Kayworth, Chatterjee, & Sambamurthy, 2001). The rapid development of DPs is evident in almost every industry. DPs have opened new corridors of thinking beyond the traditional business approaches. These platforms help firms connect to their customers and other businesses simultaneously and improve their products and services (Xiao, Tian, & Mao, 2020). Based on these arguments, the following hypothesis is proposed: H5: DPC will positively affect the FA of manufacturing firms. Firm’s agility and innovation performance Sambamurthy et al. (2003) defined agility as the capacity of a firm to understand and respond to its customer demands, operational agility as a firm’s expertise in structuring operation procedures, and partnering agility as the competency of a firm in forming business relationships. The competitive environment constantly challenges businesses. The literature on FA reveals that agility affects firm performance. Specifically, agility can help firms gain dexterity and speed (Singh et al., 2013), which are vital, especially in a rapidly changing global environment (Heckler, Illinois, & Powell, 2016). A firm’s agility also reflects the excellence of a firm in detecting and entering niche markets to redefine its business opportunities. Therefore, agility adds to a firm’s innovation performance by addressing and finding solutions to problems and responding to the challenges in the market (Song, 2015). FA also has an imperative impact on a firm (Dove & Palmer, 2004), especially on its performance outcomes than its structural or operational excellence (Yauch, 2011). C^ orte-Real, Oliveira, and Ruivo (2017),Wagner, Beimborn, and Weitzel (2014), and Yusuf et al. (2014) explored the influence of agility on business and innovation performance. The following hypothesis is then proposed: H6: FA will positively affect the innovation performance of manufacturing firms. BDAC and innovation performance Over the last few years, big data has come to light as an emerging frontier of efficiency and opportunity to transform businesses. The ways of doing business have markedly changed due to BDAC (Barton & Court, 2012). Previous studies show that BDAC can transfigure management and practice (George et al., 2014), which are substantial for innovation and considered the “fourth archetype”in science. According to the theoretical foundation of DCV, BDAC refers to an organization’s peculiar capabilities for superlative price setting and improving the quality and contributing to the innovative performance of firms. By using the information technology ecosystem, organizations can transform data into a resource that they can analyze during decision making (Rivera & Shanks, 2015). Data analytics serves as a competitive discriminator (Jeble et al., 2018) that positively affects the firm’s innovative performance (Ramakrishnan, Jones, & Sidorova, 2012). Previous research shows that BDAC innovates the entire business system from product to process and from the infrastructural system to the segmental one (Caputo, Marzi, & Pellegrini, 2016). Therefore, in fostering structural innovation, the foundation of data based on BDA plays an influential role (Tempini, 2017), whereas the personalization paradigm facilitates the innovation of services (Ng & Wakenshaw, 2017). Big data extends a company’s capabilities and leverages innovation in business models (Vecchio et al., 2018). BDAC shows potential in disrupting the innovative performance. The following hypothesis is then proposed: H7:BDAC will positively impact the innovation performance of manufacturing firms. DP capability and innovation performance DCV discusses the higher-order practices of operational capabilities to enlarge the scope and adapt and adjust the existing operational A. Khan and M. Tao Journal of Innovation & Knowledge 7 (2022) 100201 3 skills of a firm for value creation and value addition (Pavlou & Sawy, 2010). However, as the evolution and amalgamation of technologies in every field complement the performance of firms, the implementation of information and communication technology firmly positions itself on a further higher order of dynamic capability view (Parida & Ortqvist, 2015). Digital transformation changes the procedures, routines, or processes based on a technological foundation and is driven by information technology. DPs have emerged in response to the technical expansions and development triggered by the rapid spread of multiplexed technologies (Parker et al., 2016c). Therefore, designing and embracing platform capability can help firms witness a radical innovation fueled by digitization. This innovation drive emphasizes the importance of focusing on and exploring opportunities for DPs. The following hypothesis is then proposed: H8:DPC will enhance the innovation performance of manufacturing firms. Mediation of BDAC between KAC and FA Previous studies suggest that information technology empowers the agility of firms by accelerating their decision-making process, simplifying their communication, and allowing them to respond to changes swiftly. A unified platform can be established using big data and facilitate the standardization and fusion of these data, which is essential in dexterity. Integrating big data enables firms to gather and distribute information quickly. This capability also allows firms to access real-time, persistent, and comprehensive data, which can help them make quick, efficient, and appropriate decisions (Gupta & George, 2016). The BDAC guarantees extensive data handling and integrates diversified data coming at various speeds, pushing firms to be more agile in responding to this filtered stream of data (Wamba et al., 2017). KAC helps firms organize their smart capabilities, which can improve their performance. BDAC urges firms to make decisions based on factual and accurate information, paving the way for knowledge to come through their KAC. Based on these arguments, the following hypothesis is proposed: H9: BDAC will mediate the relationship between KAC and FA. Mediation of DPC between KAC and FA The introduction of warehouse management platforms has scaled the flexible capacity of repositories, yet the internet has channelized the market to a higher order. Although information technology competency leverages firms to be more agile, the degree to which these options can be employed depends on these firms’knowledge absorption and utilization capacity. With the emergence of new technologies, the opportunities for businesses to establish an edge have increased; these firms must possess knowledge management capacity and prevision to understand the significance of emerging technologies. Sambamurthy et al. (2003) explored digitized knowledge capital using platforms to produce knowledge warehouses and share this knowledge throughout an organization to increase its agility. IT- enabled capability refers to developing a DP that reflects the flexibility of technology infrastructures and applications in addressing external business requirements. The risk of long-term stiffness can be overcome using DPs; accordingly, using these platforms has become an important strategic priority for several organizations. Firms discover their agility by utilizing the data and information they collect from DPs (Cenamor, Parida, & Wincent, 2019). Upgrading from legacy systems to internet-based DPs provides these firms with enough flexibility to digitize their processes. Previous studies have proposed diverse definitions of an ecosystem networked by DPs. A DP can be defined as a collection of digital resources, including content and services, that help promote valuable interactions between customers and suppliers (Parker, Van Alstyne, & Choudary, 2016a). DPs do not maintain physical resources, such as infrastructure. They help gain market insights in real-time, support the development of products and services, and allow firms to restructure their processes quickly. DPs are connected directly to consumers, providing firms with a gateway to develop their potential absorptive capacity to acquire, assimilate, and identify knowledge from external sources (Zahra & George, 2002). The following hypothesis is then proposed: H10: DPC will mediate the relationship between the KAC and FA of manufacturing firms. The moderating role of flexibility orientations An organization’s culture is fundamental in determining its business performance and long-term competitive strength. Meanwhile, its performance is substantially dependent on the philosophy and beliefs of work established by enterprise managers. The efficacy of maintaining strong communication and improving performance outcomes is contingent upon integrating a thriving organizational culture (Idris, Wahab, & Jaapar, 2015). The management and decisionmakers typically face many challenges in establishing a flourishing organizational culture, which is integral to improving productivity and performance (Kenny, 2011). However, only a few studies have explored the effects of organizational culture on knowledge absorption and facilitating the adoption of information technology. Although, the previously examined constructs, i.e., knowledge absorption management as the dexterity of valuable information recognition, apprehension and its application to the commercial purpose with the corporate culture, which is a paradigm of ideas and values that frame the performance of an organization, potentially affect the affluent knowledge application. Developing BDAC requires a combination of tangible and intangible resources in line with the decisionmaking capacity that has fostered a flexible and swift culture that supports factual-based judgments. The following hypothesis is then proposed: H11: Flexibility orientations (control orientations) will negatively moderate the relationship between KAC and BDAC, whereas flexible orientations will positively moderate the relationship between KAC and BDAC. The moderating role of DDC Deshpande, Farley, and Webster (1993) argued that organizational culture is vital in deciding how a firm responds to external events and strategies. Organizational culture determines the strategy and the steps taken by a firm in response to technological and business competitions. Technology-oriented firms typically rely on the information and knowledge coming from new resources by engaging in BD analytics. BDAC improves innovation performance based on the decisions made after analyzing massive datasets. Given its value, BD has attracted much attention from service-providing and productmanufacturing firms (Constantiou & Kallinikos, 2015). Nevertheless, extracting real value from BD depends on the DDC of a firm. Several investments in BD projects have failed to draw the desired output due to the lack of an adequate data-driven culture (Lehrer, Wieneke, vom Brocke, Jung, & Seidel, 2018). Following the logic of DCV, BDAC gives firms a competitive advantage in a high-order construct that is greatly influenced by their strategic resources and data-driven decision-making capability to achieve an excellent performance. A detailed review of the literature on environmental and social sustainability, BDAC, and predictive analytics reveals that core insights are driven by data-encompassing interdepartmental cooperation in the modern economy. Manufacturing and technology-oriented industries depend on consumer data, competitive market orientation, and financial and economic information to identify the traits and hallmarks they can add to their future products. The following hypothesis is then proposed: H12: DDC will positively moderate the relationship between BDAC and innovation performance (Figure 1). A. Khan and M. Tao Journal of Innovation & Knowledge 7 (2022) 100201 4 Methods This research is conducted based on a simple random sampling technique. In the first stage, the manufacturing firms and their relevant statistics were collected from the official websites of the Government of Pakistan, such as the Economic Survey of Pakistan and the Statistics Bureau of Pakistan. An ISO-certified data collection firm was also recruited to collect data from the senior, middle, or frontline managers of manufacturing firms, given their excellent knowledge about their operations. This firm used a self-administered questionnaire designed by the research team to collect data from the target respondents. The recruited managers were contacted by email and other social media platforms. Three hundred forty-seven responses were received, yielding a low response rate of 14%, which was understandable given the COVID-19 pandemic. Among these 347 responses, only 325 were deemed appropriate for the final analysis. Table 1 presents details on Pakistan’s manufacturing industry, the research population, and the profiles of the respondents. Measurement items All constructs used in this study were adapted from the literature and measured on a five-point Likert scale ranging from “strongly disagree”to “strongly agree.”KAC was adapted from Jansen, Van Den Bosch, and Volberda (2005) and Shahzad et al. (2020). Sample items included “We have effective routines to identify, value, and import new information and knowledge.”BDAC was adapted from C^ orte- Real, Ruivo, Oliveira, and Popovi c (2019) and Chen, Preston, and Swink (2015). Sample items included “Our enterprise uses BDA purchasing analytics for purchasing.”DPC was adapted from Cenamor et al. (2019). Sample items included “We have developed DPs for consumers to share prior experiences, knowledge, and expertise.”FA was adapted from Tallon and Pinsonneault (2011) and Ashrafi, Zare Ravasan, Trkman, and Afshari (2019). Sample questions included “Adopt new technologies to produce better, faster, and cheaper products and services.”FO was adapted from Dubey, Gunasekaran, and Childe (2019). Sample items included “Our firm follows formal rules and policies which involve less risk.”DDC was adapted from Gupta and George (2016) and Dubey et al. (2019), with sample questions including “We base most of the decisions on data rather than instinct.”IP was measured with the sample item “In terms of novelty, our firm is always the first one to come up with new ideas about the product'', adapted from Maurer, Bartsch, and Ebers (2011) and Prajogo and Ahmed (2006). Table 1 Demographics. Details of Demographics (n= 325) Attributes Distribution N Percentage Job Title Senior Manager 27 8% Production Manager 40 12% Supervisor 56 17% Middle Manager 117 36% Frontline Manager 85 26% Education Technical 41 13% Graduation 104 32% Master 162 50% Above Master (MS/MPhil) 18 6% Gender Male 232 71% Female 93 29% Industry Textile 37 11% Coal and Petroleum 20 6% Automobiles 36 11% Fertilizers 30 9% Wood and Papers 31 10% Food and Beverages 32 10% Pharmaceutical 35 11% Surgical Instruments 23 7% Engineering Products 21 6% Chemical Products 17 5% Sports Good 30 9% Misc. Manufacturing 13 4% Ownership Public Firms 97 30% Private Firms 228 70% Fig. 1. Research model. A. Khan and M. Tao Journal of Innovation & Knowledge 7 (2022) 100201 5 Data analysis and results Measurement model The reliability and validity of data and instruments were assessed in the measurement model (Barclay, Higgins, & Thompson, 1995). Internal consistency evaluates data reliability based on two measures, namely, Cronbach’s alpha and composite reliability. Meanwhile, data validity can be measured via content validity, face validity, convergent validity, and discriminant validity (Chin, 1998;Hair, Ringle, & Sarstedt, 2011). Cronbach’s alpha measures the psychometric reliability of data, the inter-item correlation of each construct, and the average correlation of the actual items. Cronbach’s alpha has a minimum threshold of 0.60 (Hair et al., 2011). As shown in Table 2, all Cronbach’s alpha values in this study exceed this threshold, thereby suggesting that the average and actual correlations between the items are exact and that the data are reliable and can be used for further analysis. Composite reliability reveals all indicators of a particular construct (Henseler, Ringle, & Sarstedt, 2015) and can be measured using PLSSEM. This measure has a minimum threshold of 0.60 (Fornell & Larcker, 1981;Hair et al., 2011). Measures of composite reliability work better when the items are reflective. If these items are formative, then the VIF value is used instead to test the reliability of indicators (Hair et al., 2011;Kutner, Nachtsheim, Neter, & Li, 2005). Composite reliability was employed at the first stage of this study, given that the constructs have reflective items. Table 2 shows that the composite reliability of all items exceeds the 0.60 thresholds. Convergent validity illustrates the theoretical relationship among the constructs of a model and indicates the degree of correlation between the study variables in the context of the same model. If the variables are not correlated, they do not need to be combined into a single model. Convergent validity is measured based on the average variance extracted, with a minimum acceptable value of 0.50 (Fornell & Larcker, 1981a;Hair et al., 2011). Table 2 shows that all AVE values exceed this threshold, thereby confirming that the constructs are interlinked in the context of the model. Common method bias (CMB) or variance is related to the adopted measurement method instead of the constructs. CMB arises when the data for the dependent and independent variables are collected from the same set of respondents (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). CMB is a severe problem that can jeopardize the results of any study. Accordingly, researchers have used several methods to address CMB, including Harman’s single factor test (Maxwell & Harman, 1968), Liang’s method (Liang, Wang, Xue, & Ge, 2017), Bagozzi’s approach (Bagozzi, Yi, & Phillips, 1991), and Kock’s inner VIF method (Kock, 2015). Bagozzi’s method was employed in this study to test CMB. According to this method, if the correlation among the variables is less than 0.90, the data are free from CMB and can be further analyzed. The inner VIF proposed by Kock (2015) was also employed to test CMB by performing a full collinearity test. The inner VIF was calculated while considering each variable dependent once. As shown in Table 2, all inner VIF values are less than the 5 thresholds (Kock, 2015), proving that CMB is not a severe concern in this study. Discriminant validity (Fornell − Larcker criterion) One way to measure the discriminant validity is using the Fornell −Larcker criterion, which compares the square root of AVE with the inter-construct correlation. Specifically, the square root of AVE should be greater than the inter-variable correlation to confirm discriminant validity (Fornell & Larcker, 1981a). The shared variance of the model was less than the square root of AVE. Table 3 reports that the square roots of AVE are greater than the inter-construct correlations reported in the same column. Heterotrait-Monotrait (HTMT) ratio In a contemporary research context, a higher factor loading can contaminate the results of the Fornell−Larcker criterion and subsequently affect the discriminant validity of the constructs. The HTMT ratio can be used as an alternate measure of discriminant validity (Henseler et al., 2015). HTMT ratio is a breakthrough in the context of PLS-SEM. Results of Monte Carlo simulations even show that the HTMT ratio outperforms the other measures of discriminant validity in terms of accuracy. Table 4 reports that all HTMT ratios are below the minimum threshold of 0.90 (Henseler et al., 2015), confirming discriminant validity. Assessment of the structural equation model Coefficient of determination (R2) R 2 assesses the variance in the independent variable due to the independent variables. R 2 has no minimum threshold; depending on their study context and discipline, researchers should decide whether their obtained R 2 can explain enough variance (Henseler, Ringle, & Sinkovics, 2009;Hulland, 1999). R 2 measures the overall predictive efficiency of the model that illustrates the combined variance of all independent variables. BDAC obtained an R 2 value of 0.578, suggesting that KAC explains 57.8% of BDAC variance. Similarly, KAC explains 29.7% of the variance in DPC, KAC, DPC, and BDAC collectively explain 64.5% of the variance in FA, and FA, DPC, and BDAC collectively explain 56% of the variance in IP. Determining effect size (f2) The effect size measures the influence of individual variables by omitting the independent variable from the model and subsequently observing the change in this variable. PLS-SEM uses the parameter f 2 to capture the effect of individual variables on the dependent variables. In previous studies, f 2 values of 0.02, 0.13, and 0.26 are categorized as a low, medium, and high, respectively (Cohen, 1988). According to the standard, all the values are small, medium, and high, except one with no DPC effect on IP. Predictive relevance (Q2) To analyze the predictive relevance of the model, Q 2 is calculated in Smart-PLS via a blindfold procedure. In previous studies, Q 2 values of 0.02, 0.13, and 0.26 are categorized as low, medium, and high. All variables in this study have a sufficiently high Q 2 , thereby indicating the high predictive relevance of the model. Direct path analysis Bootstrapping uses a replacement procedure to enhance the sample size. Each observation was selected from the population each time and replaced with other elements; in this way, all elements have an equal chance of being chosen as samples. An observation may be selected more than once or may not be included in the sample. The minimum sample size for bootstrapping should equal the actual sample size (Wetzels, Odekerken-Schr€ oder, & Van Oppen, 2009). However, a subsample of 5000 observations has been recommended in the literature (Hair, Risher, Sarstedt, & Ringle, 2019). Following this suggestion, this study applied bootstrapping with 5000 subsamples to obtain more accurate estimates. Bootstrapping returns all the relationships specified in the model and their significance and strength. Table 5 presents the path coefficients of the direct, indirect, and moderating effects as specified in the model. The direct relationships proposed in H1 to H8 were all supported by the results in Table 5. H1 proposed a direct and positive effect of KAC on BDAC (b=0.360, T-value= 5.226 p<0.001), H2 proposed a positive and direct impact of KAC on DPC (b=0.545, T-value= 6.389 p<0.001), H3 proposed a positive effect of KAC on FA (b=0.298, T-value= 3.808 A. Khan and M. Tao Journal of Innovation & Knowledge 7 (2022) 100201 6 p<0.001), H4 proposed a positive effect of BDAC on FA (b=0.422, T- value= 5.213 p<0.001), H5 proposed a positive effect of DPC on FA (b=0.213, T-value= 3.681 p<0.001), H6 proposed a positive effect of FA on IP (b=0.264, T-value= 3.115 p<0.001), H7 proposed a positive impact of BDAC on IP (b=0.367, T-value= 4.699 p<0.001), and H8 proposed a positive impact of DPC (b=0.193, T-value= 2.273 p<0.05). Mediation and moderation analysis Using PLS-SEM to test mediation has been debated by researchers for several decades. The mediation analysis procedure proposed by Baron and Kenny (1986) has been widely adopted in recent studies. However, contemporary research on methodologies (e.g., Hayes & Table 2 Convergent validity. Constructs Items-loading Cronbach’s Alpha Composite Reliability Average Variance Extracted (AVE) VIF Big Data Analytics Capability BDAC1 0.708 0.849 0.885 0.525 2.743 BDAC2 0.721 3.608 BDAC3 0.741 3.558 BDAC4 0.699 2.173 BDAC5 0.756 2.421 BDAC6 0.761 2.617 BDAC7 0.682 1.454 Data-Driven Culture DDC1 0.854 0.920 0.927 0.720 2.199 DDC2 0.796 3.046 DDC3 0.91 3.132 DDC4 0.795 3.3 DDC5 0.88 2.943 Digital Platform Capability Connect to Businesses CTB1 0.884 0.907 0.942 0.844 0.907 CTB2 0.943 4.959 CTB3 0.927 4.419 Connect to Customers CTC1 0.88 0.770 0.868 0.688 2.343 CTC2 0.846 2.217 CTC3 0.757 1.269 Firms Agility FA1 0.679 0.883 0.907 0.550 1.832 FA2 0.776 2.486 FA3 0.725 2.277 FA4 0.768 2.186 FA5 0.721 1.993 FA6 0.769 2.173 FA7 0.748 1.984 FA8 0.741 1.865 Flexibility Orientation FO1 0.858 0.827 0.887 0.666 2.19 FO2 0.893 2.675 FO3 0.871 2.363 FO4 0.612 1.256 Innovation Performance Process Innovation PIN1 0.762 0.713 0.836 0.629 1.143 PIN2 0.815 2.068 PIN3 0.802 2.036 Product Innovation PRIN1 0.749 0.823 0.883 0.654 1.633 PRIN2 0.798 1.765 PRIN3 0.838 2.122 PRIN4 0.845 2.150 Knowledge Absorptive Capacity Acquisition AQC1 0.88 0.825 0.896 0.741 2.011 AQC2 0.859 1.959 AQC3 0.842 1.713 Assimilation ASM1 0.864 0.821 0.894 0.737 1.826 ASM2 0.856 1.831 ASM3 0.855 1.864 Transformation TRNS1 0.888 0.864 0.917 0.786 2.399 TRNS2 0.886 2.132 TRNS3 0.885 2.178 Exploitation EXP1 0.905 0.876 0.924 0.802 2.447 EXP2 0.885 2.304 EXP3 0.897 2.408 A. Khan and M. Tao Journal of Innovation & Knowledge 7 (2022) 100201 7 Scharkow, 2013) reported some theoretical and methodological deficiencies in this procedure. Results of this procedure pointed toward both direct and indirect effects. The direct effects were significant and positive, whereas the indirect or mediating effects were significant and pointed in the same direction as the direct effects. Therefore, a partial mediation was observed between the KAC and FA. These results support H9, which proposed that BDAC mediates the relationship between KAC and FA (b=0.152, T-value= 4.511 p<0.001), and H10, which proposed that DPC mediates the relationship between KAC and FA (b=0.116, T- value= 3.119 p<0.001). Two moderating effects were also tested. First, the moderating effect of FO on the direct relationship between KAC and BDAC was significant, thereby supporting H11 (b=0.069, T-value= 1.651, p<0.10. The interaction graph in Fig. 2 shows that negative values of FO weaken the relationship between KAC and BDAC, whereas its positive values strengthen such a relationship. Therefore, flexible orientation increases the positive effect of KAC on BDAC and vice versa. Second, control orientation weakened the positive effect of KAC on BDAC. Specifically, control orientation was insignificant with a negative coefficient, suggesting that a weak data-driven culture weakens the relationship between BDAC and IP, thereby rejecting H12 (b=0.090, T-value= 1.147 p>0.10). Data-driven culture also produced an insignificant moderating effect. Fig. 3 shows the moderating effect of DDC on the relationship between BDAC and IP. Fig. 4 presents the path coefficients along with their values. Overall model fit The outer model calculates the reliability and validity, whereas the inner model evaluates the predictive efficiency. The standardized root means square residual (SRMR) has a minimum acceptable value of 0.08 (Henseler et al., 2015). The other value, normed fit index (NFI), is associated with the chi-square index and preferably has higher values. An NFI value of near 1 is considered acceptable. In this study, the SRMR and NFI values were calculated to assess the model fit. Under the saturated model, the obtained SRMR value was 0.07, below the 0.08 threshold. Meanwhile, the NFI value was 0.70, near 1, suggesting a good model fit. The goodness-of-fit (GOF) index considers the performance of both the measurement and structural models. This index also provides operational solutions to the problems faced by previously developed models in measuring the GOF of PLS path models. Accordingly, the GOF index has been widely employed (Chin, 2010). Studies using PLS-SEM also adopt this index for global validation of models (e.g., Duarte & Raposo, 2010;Rigdon, Ringle, & Sarstedt, 2010). In this study, the GOF index was calculated as follows; Table 3 Fornell Larker criterion. Fornell-Larker Criterion Constructs AQC ASM TRAN EXP BDAC CTB CTC DDC FA FO PIN PRIN AQC 0.86 ASM 0.61 0.86 TRAN 0.41 0.5 0.89 EXP 0.59 0.75 0.41 0.9 BDAC 0.51 0.49 0.55 0.43 0.73 CTB 0.31 0.33 0.41 0.35 0.48 0.92 CTC 0.45 0.44 0.49 0.46 0.63 0.75 0.83 DDC 0.01 0.03 0.01 0.09 0.01 0.03 0.04 0.85 FA 0.51 0.53 0.59 0.55 0.74 0.49 0.67 0.004 0.74 FO 0.39 0.28 0.33 0.33 0.64 0.42 0.49 0.04 0.51 0.82 PIN 0.38 0.36 0.34 0.42 0.47 0.35 0.44 0.07 0.5 0.39 0.79 PRIN 0.46 0.49 0.57 0.46 0.67 0.45 0.58 0.02 0.62 0.5 0.43 0.81 Table 4 HTMT ratio. HTMT Ratio Constructs AQC ASM TRAN EXP BDAC CTB CTC DDC FA FO PIN PRIN AQC ASM 0.74 TRAN 0.48 0.60 EXP 0.70 0.88 0.47 BDAC 0.61 0.57 0.63 0.50 CTB 0.36 0.38 0.46 0.39 0.55 CTC 0.55 0.55 0.60 0.55 0.77 0.90 DDC 0.05 0.05 0.05 0.08 0.04 0.05 0.06 FA 0.60 0.62 0.67 0.61 0.84 0.55 0.80 0.05 FO 0.47 0.35 0.39 0.40 0.77 0.48 0.60 0.04 0.59 PIN 0.49 0.45 0.40 0.50 0.58 0.42 0.58 0.08 0.60 0.52 PRIN 0.56 0.60 0.68 0.54 0.78 0.52 0.72 0.05 0.72 0.61 0.54 Table 5 Path coefficients and significance. Path Coefficients and Significance Hypothesis Coefficient Standard Deviation T Statistics P-Values Direct Relationships H1: KAC ->BDAC 0.360 0.069 5.226 0.000 *** H2: KAC ->DPC 0.545 0.085 6.389 0.000 *** H3: KAC ->FA 0.298 0.078 3.808 0.000 *** H4: BDAC ->FA 0.422 0.081 5.213 0.000 *** H5: DPC ->FA 0.213 0.058 3.681 0.000 *** H6: FA ->IP 0.264 0.085 3.115 0.002 *** H7: BDAC ->IP 0.367 0.078 4.699 0.000 *** H8: DPC ->IP 0.193 0.085 2.273 0.023 ** Mediating Relationships H9: KAC ->BDAC ->FA 0.152 0.034 4.511 0.000 *** H10: KAC ->DPC ->FA 0.116 0.037 3.119 0.002 *** Moderating Effects H11: KAC*FO ->BDAC 0.069 0.042 1.651 0.099* H12: BDAC*DDC ->IP 0.090 0.078 1.147 0.251 Note: *** , ** , *represent the significance level at 1%, 5% and 10% respectively. A. Khan and M. Tao Journal of Innovation & Knowledge 7 (2022) 100201 8