OVERVIEW OF THESIS
Research Background
The concept of ‘big data’ is generating tremendous attention worldwide The results of a Google search in mid-August 2014 on the phrases “big data” and
Big data analytics (BDA) has gained considerable traction in the corporate world, with a notable increase in investment among Fortune 1000 companies, rising from 85% to 91% in the past year This surge in interest is largely attributed to the potential for enhanced productivity and profitability, estimated at 5–6% According to Agarwal and Dhar (2014), the term "analytics" alone generates millions of search results, highlighting its growing importance in business strategies.
Notary Nguyen Kim Anh from the State Bank emphasized that data is crucial for progress, particularly in the monetary banking sector, which is central to transforming business models Effective data management is essential for financial sustainability (Anh, 2020) The United Nations' 2030 Agenda for Sustainable Development highlights the importance of meticulous data collection, storage, and consistent reporting to ensure no one is left behind (United Nations Report, 2019) Today, the influence of Big Data spans nearly every industry, offering diverse perspectives and significant implications across various fields.
Big Data, as described by Manyika et al (2011), refers to data sets that exceed the capabilities of traditional database software for capture, storage, management, and analysis It is characterized by three key elements: volume, indicating a large quantity of data; velocity, which denotes the rapid creation of data; and variety, representing diverse sources of information (DataStax, 2011) In finance and business, Big Data plays a crucial role, as it encompasses vast amounts of information from stock exchanges, banking, online activities, and in-store purchases, all of which are continuously processed and stored for analyzing stock trends, consumer behavior, and market dynamics (Halevi and Moed, 2012).
The era of Big Data presents both significant opportunities and challenges, particularly in data-intensive sectors like banking (Mayer-Schonberge and Cukier, 2013) As banks recognize that intelligence has become more valuable than financial capital (Kharote and Kshirsagar, 2014), the rise of e-banking and mobile banking has led to an exponential increase in real-time banking data To navigate these changes and leverage the vast availability of big data, mastering specific big data analytics tools has become crucial for the banking industry.
Research objectives
Big data mining is a transformative approach that many companies, including Vietnamese banks, are adopting to enhance their services and operations The Reserve Bank of India's new data collection method streamlines the analysis and tracking of financial institution customers, enabling banks to gather valuable insights This article examines the impact of big data on Vietnamese financial structures, discussing the results of big data analysis and exploring whether big data mining and investment can boost the financial sector's performance in Vietnam.
- To identify Big Data Analysis (BDA) related factors which influence banks' financial performance
- To evaluate the level of BDA related factors' influence on banks' financial performance
- To examine the relationship between bank performance and banks' financial performance
- To provide managerial implications to improve the applicability of BDA in managing bank performance.
Research questions
The advent of Big Data has introduced significant advantages and challenges across various research fields, particularly in banking, which has emerged as a prominent area for data-intensive exploration (Mayer-Schonberge & Cukier, 2013) In today's landscape, banks recognize that information has surpassed financial capital as the most valuable asset (Kharote & Kshirsagar, 2014) The rise of e-banking and mobile banking has further accelerated the demand for timely banking insights, emphasizing the necessity for financial institutions to effectively utilize big data analytics tools Despite the myriad opportunities and challenges presented by big data in the banking sector, several critical issues remain inadequately addressed in existing literature.
- Which BDA related factors that can influence banks' financial performanfe?
- Which managerial implications that can enhance the applicability of BDA in managing bank performance?
This research explores the intersection of big data analytics (BDA), financial performance, and bank performance, structured to address key research questions It begins with definitions of BDA, followed by an overview of relevant studies on IT capabilities and extensive data analytics The paper then introduces the research model and hypotheses, detailing the research design Subsequent sections cover data analysis and findings, leading to a discussion and concluding with implications for both research and practical applications.
Research subject and scope
Object of study: Recognizing the close association and certain influence between Big Data and Financial performance at banks in Ho Chi Minh City
Research Scope: The banks at Ho Chi Minh city
Research time: The research is going to be carried out from April, 2021 to June,
Survey audience: Stakeholders, leaders and officers of Banks at Ho Chi Minh city who are working and from fresher to senior.
Significant of study
This research employs a case study approach to investigate the benefits, challenges, and strategies related to Big Data adoption in banks The findings aim to provide Bank Managers with valuable insights into the factors influencing Big Data adoption and its outcomes, enabling them to make informed decisions that enhance their organizations' performance.
This research paper explores the impact of various research models on the relationship between big data and financial performance, enhancing understanding of how these variables interact The findings not only raise awareness among researchers about effective methodologies but also provide valuable insights that can be applied to further studies Ultimately, this work serves as a foundational resource for practical applications in the field.
Research Method
1.6.1 Expected layout of the thesis
Chapter 1 briefly introduces the overview and backdrop of this research; and argues the reasons why there is a need to acquire better knowledge of big data ideas within the banking industry in general and the Vietnamese banking sector in particular Finally, the study purpose and objectives are also presented
Chapter 2: Literature review and review research models
This chapter focuses on the Vietnamese banking sector, providing a literature review that examines big data analysis (BDA) and its impact on financial performance It highlights the challenges of implementing BDA in the banking industry and explores the relationship between BDA and financial outcomes Additionally, the chapter discusses key concepts and features of BDA within the sector and proposes a conceptual framework outlining variables that influence financial performance in Vietnamese banks A model illustrating the connection between BDA and financial performance is also presented, serving as a foundation for the research methodology outlined in Chapter 3.
This chapter outlines the research techniques and methodologies employed in the study, highlighting the debate between qualitative and quantitative methods, along with their respective advantages and disadvantages The research adopts a mixed methodology, primarily focusing on qualitative research while incorporating quantitative methods for support This approach facilitates concurrent triangulation, allowing for cross-validation of findings aligned with the research objectives Data collection techniques include documentation and surveys, with the chapter providing a detailed overview of the research design and data collection strategies.
This chapter presents the findings of a quantitative study focused on the factors influencing financial activities within Vietnam's banking industry The results align with the study questions and the proposed framework, highlighting differences in Big Data Analytics (BDA) conceptualization and financial performance among various bank types Additionally, the chapter discusses actual BDA practices observed in the sector.
Chapter 5: Discussion of research findings
This chapter highlights the findings of a quantitative study that investigates the relationship between Big Data Analytics (BDA) and financial performance in the Vietnamese banking sector The research reveals that BDA capability, bank performance, technology, and talent capability all positively impact financial success By exploring how BDA practices affect various dimensions of financial performance, this study contributes valuable insights to the existing literature.
This chapter analyzes the research findings and their implications, followed by a self-critique and acknowledgment of limitations It concludes with practical managerial recommendations and suggestions for future research directions.
LITERATURE REVIEW AND REVIEW RESEARCH MODEL
The Big Data Analytics
Big Data Analytics (BDA) has become an essential asset for enhancing competitiveness, consistently ranking high on the agendas of senior executives in recent years Banks face challenges in navigating the immense influx of data generated from diverse internal and external sources, which come in various formats According to Sunset et al (2015), big data encompasses data sets from heterogeneous and autonomous sources, characterized by their complexity, dynamic interactions, and volume that surpasses the capabilities of traditional tools for collection, archiving, control, analysis, and utilization.
Figure 2.1 : The Gartner's Vector model
The financial industry generates an immense volume of data, often reaching big data levels of several terabytes (Tbytes) or even petabytes (Pbytes) For instance, the New York Stock Exchange (NYSE) produces over one terabyte of market quotes, historical trade data, and market data daily.
Velocity is crucial in big data environments, particularly when data collection or retrieval reaches a minimum of 105 transactions per second Financial markets routinely generate data at this high volume, and the efficiency of devices in processing exchange data directly enhances their trading capabilities.
Big data algorithms demonstrate their versatility by effectively handling diverse data types and sources In the banking and financial sectors, these algorithms manage reference data related to legal entities, trade and business information, customer demands through both electronic and voice channels, among other data sources.
Big data is characterized by five key dimensions: volume, variety, velocity, veracity, and value Volume pertains to the vast size of data sets generated from diverse applications, expanding from megabytes to petabytes Variety highlights the diverse types of data that comprise big data, including textual, social media, traffic, and health-related information Velocity emphasizes the rapid pace at which data is collected and generated in real-time Veracity addresses the trustworthiness of data sources, ensuring their reliability Lastly, value represents the insights and hidden opportunities that can be uncovered from extensive datasets (Hashem et al., 2015).
Traditional data warehouse technology struggles to manage vast data volumes, often reaching hundreds of terabytes (Grover et al, 2018; Manyika et al, 2011) Big data is inherently variable and not universally quantifiable, meaning that its value differs across organizations; larger datasets do not equate to better insights (Davenport et al, 2018; Noyes, 2018) Big data analytics introduces innovative tools for visualizing and manipulating data, enabling organizations to convert complex data into accessible formats like charts and graphs This suite of tools is designed to handle unstructured data that traditional database systems cannot manage, allowing businesses to identify changes and innovate in real-time (Davenport, 2018) Each company has unique data needs and use cases, making it crucial to understand the specific motivations behind technology choices and analytics strategies, as solutions effective for one organization may not be suitable for another While benchmarking against others can provide insights, it is essential to grasp your own business drivers to effectively apply big data solutions.
Banking and financial sector legislation requires the calculation of complex metrics like XVA, which accounts for counterparty credit risk and funding costs, influencing capital reserves and profitability Analyzing time-sequenced transactional data helps model business and consumer behavior, such as predicting credit defaults by charting trade volume over time, thus safeguarding loan capital Despite challenges, many financial institutions struggle with effective data management strategies, as highlighted by PwC's market analysis, which identified organizational and cultural barriers relevant across various industries.
Many financial sector executives believe that big data algorithms primarily address technological challenges rather than market issues However, as companies collect and analyze data, it becomes clear that technology plays a crucial role in their success While some organizations struggle to extract value from their data, others argue that realistic data methods enhance technological productivity without significantly impacting profits Nevertheless, in-depth research through advanced data techniques can lead to improved market success and operational efficiency Additionally, the finance industry has traditionally been less appealing to data scientists, resulting in challenges for institutions in attracting and retaining the necessary talent.
Despite the strong desire for reform, the timeline and methods for transforming organizations to adopt realistic data strategies can be ambiguous A recent IDC survey revealed that the banking sector is among the largest investors in big data analytics Furthermore, FinTech companies are developing innovative technologies and products tailored to diverse banking requirements, particularly in asset and wealth management This trend is accompanied by a growing volume of analysis and algorithm development focused on financial data, aimed at enhancing market efficiency.
Navigating BDAs in the Financial Sector
Big data, as defined by Goes (2014), refers to vast amounts of diverse observational data that aid in decision-making Schroeck et al (2012) emphasize the extensive range of information encompassed by big data, including real-time data, non-traditional media, technology-driven data, and social media insights While 'volume' and 'variety' are commonly highlighted in big data discussions (Davenport et al., 2012; IBM, 2012), factors like velocity and veracity also play crucial roles (Beulke, 2011; Gentile, 2012) The advent of big data analytics (BDAs) has transformed how companies operate, with many large firms leveraging BDAs to enhance supply chains and maintain a competitive edge (Jun et al., 2015) BDAs empower organizations to utilize advanced analytical tools to harness high-quality data, boosting productivity and ensuring sustainable success According to McKinsey's 2018 Global Banking Report, analytically driven companies grow three times faster than their less data-oriented peers, with the banking sector leading in BDA utilization However, the integration of BDAs into traditional decision-making processes within the banking industry remains inadequate (McKinsey Company Global Banking Report, 2018).
Big Data Analytics (BDAs) have become essential for the banking sector, aiding in fraud detection, financial crime prevention, credit risk management, and targeted marketing Banks are shifting from product-centric to customer-centric models, a transition facilitated by BDAs For example, the Overseas Chinese Banking Corporation (OCBC) in Singapore and Malaysia effectively utilizes historical customer data to understand client needs (IBM Global Business Services Report, 2013) Additionally, BDAs play a vital role in promoting financial inclusion, enhancing microfinance initiatives, and evaluating creditworthiness by analyzing both organized and unstructured consumer data from diverse platforms (The ASEAN Post, 2020).
The theoretical background
The author employ a broad theoretical framework based on three complimentary theories: institutional theory (DiMaggio and Powell, 1983), resource-based view theory (RBT) (Barney, 1991), and organizational culture (DiMaggio and Powell,
Institutional theory sheds light on the implementation of Big Data Analytics (BDA) by examining the interrelationships and coordination among stakeholders and the focus organization According to Resource-Based Theory (RBT), internal resources significantly influence organizational strategy and performance Previous studies have integrated institutional theory with RBT to elucidate organizational decision-making, highlighting the distinct reasons for organizations and the interplay between external pressures and internal resources.
2.3.1 The Resources-based view theory (RBT)
Resources-based view theory (RBT) explains how an organization can achieve competitive advantage by creating bundles of strategic resources and/or capabilities
In the context of Big Data Analytics (BDA), the integration of tangible resources, such as technology and data, along with intangible resources like information sharing, plays a crucial role in developing BDA capabilities (Gunasekaran et al., 2017) Additionally, Gupta and George (2016) emphasize that the combination of these resources with human skills—both technical and managerial—further enhances BDA capabilities, particularly when supported by a culture driven by big data.
This study is situated within the Resource-Based Theory (RBT) domain, which analyzes the elements affecting the performance of small-scale firms It is posited that a company's internal resources and competencies yield competitive advantages, as suggested by Barney (1991) Firm performance is derived from a unique set of resources that are challenging to replicate or substitute The RBT framework effectively elucidates the various factors influencing business performance, as noted by Rahman and Ramli (2014).
In the accounting and finance literature, firm resources are categorized in various ways, with Barney (1991) identifying three main types: human, physical, and organizational capital Human capital encompasses the experience, training, technical skills, relationships, and expertise of a firm's management and staff, while physical capital includes technology, machinery, location, and access to productive inputs (Essel et al., 2019) Additionally, attributes such as commitment, honesty, and competence are recognized as valuable assets According to Radzi et al (2017), human, physical, financial, and technical resources are classified as the only tangible assets within a firm.
Intangible assets, such as knowledge, skills, and reputations, play a crucial role in a company's competitive advantage Radzi et al (2017) highlight that businesses aim to access and manage these resources effectively for both short-term and long-term benefits Variations in resource access and control can lead to unique products and services, making a firm's success dependent on its ability to leverage its resources and competencies (Saffu et al., 2012) Ultimately, these intangible resources are vital for fostering innovation and sustaining a competitive edge in the market (Radzi et al., 2017).
The Resource-Based Theory (RBT) offers a valuable framework for assessing the performance of small businesses in emerging markets, where resource limitations and lack of experience are prevalent (Saffu et al., 2012) This study adopts RBT's premise that a firm's resources and capabilities are fundamental to its performance Furthermore, RBT underscores that companies can achieve a competitive advantage through resources that are valuable, rare, inimitable, and non-substitutable (Barney et al., 2011), which also applies to inherent talent.
Environmental regulations in the banking sector, as highlighted by structural theory (Campbell, 2007), influence bank performance and financial outcomes by subjecting institutions to coercive and normative pressures (Ameer and Othman, 2012).
Institutional theory delves into the enduring aspects of social structure, exploring how constructs like schemas, laws, norms, and routines are established as authoritative guidelines for social behavior It analyzes the processes through which these elements are created, spread, embraced, and modified over time and across different contexts, as well as their eventual decline and loss of relevance.
Institutional theory has its roots in the early social sciences, drawing on revolutionary ideas from diverse thinkers such as Marx, Weber, Cooley, Mead, Veblen, and Commons Although much of this foundational work was overshadowed by the rise of neoclassical economics, behavioral political science, and positivist sociology in the late 19th and early 20th centuries, it has recently experienced a significant resurgence This revival underscores the relevance of institutional theory in analyzing bank and financial performance, highlighting its effectiveness in motivating and satisfying financial needs, which ultimately enhances overall organizational performance.
Strong management theorists assert that prioritizing relationships with key stakeholders enhances overall performance, linking bank success to sustainability efficiency (Hackman, 1980) Effective management practices can positively influence financial outcomes, enabling banks to reduce costs while enhancing their reputation (Weber, 2017; Deephouse, Newburry, and Soleimani, 2016) Consequently, leaders in banking are motivated to implement sound management strategies within their organizations.
Management is the process through which organizational leaders efficiently utilize resources to achieve the company's goals at minimal costs Collectively, management refers to the team responsible for guiding the workforce and ensuring the organization meets its objectives The key activities involved in this process are known as the "functions of management."
• Planning- the process of setting the objectives to be accomplished by an organization during a future time period and deciding on the methods of reaching them
• Organizing- the process of grouping and assigning activities and providing the necessary authority to carry out the activities
• Staffing- the process of filling positions in the organizational structure with the most qualified people available
• Motivating- the process of getting people to contribute their maximum effort toward the attainment of organizational objectives
• Controling- the process of ensuring the achievement of an organization’s objectives
• Two essential processes are involved in all these five managerial functions: decision making and communicating
• Decision making- this is the process of choosing form two or more alternatives In planning, for instance, the manager decides among alternative ways of accomplishing objectives
Effective communication is essential for successful management, as it involves the exchange of facts, ideas, opinions, and emotions among individuals Managers rely on accurate information to plan effectively, and clear communication is crucial for implementing those plans Additionally, organizing, staffing, motivating, and controlling processes cannot occur without effective communication It is important to note that many management issues stem from communication breakdowns.
Thus, banks regard the Central Bank's environmental and social risk management guidance as a form of structured pressure (Global Climate Partnership Fund, 2018)
Banks are increasingly pressured to adopt sustainable practices and invest in the green economy due to societal expectations (Ahmed, Alam, and Rahman, 1999) Consequently, these financial institutions are compelled to align with these demands Research indicates that strong performance in sustainability positively influences the financial outcomes of banks.
The transfer of a talented employee's skills and abilities is contingent upon suitable environmental contexts, aligning with management theory This reinforces the importance of Big Data Analytics (BDA) capabilities in succession planning, highlighting their rationality within this theoretical framework.
Organizational culture is a term in organizational theory that has many conflicting definitions, as highlighted by Smircich (1983), who identified five distinct types Instead of trying to reconcile these varied definitions, this paper adopts a definition that aligns with most research linking organizational culture to business performance, as seen in the works of Deal and Kennedy (1982) and Peters and Waterman (1982).
Theoretical Model and Hypotheses
In our study, we utilized SPSS 20.0 for structural equation modeling, analyzing data from a sample of 250 bank officials By integrating institutional theory with big data culture, we provide a comprehensive framework that underpins our empirical findings, as neither perspective alone sufficiently accounts for the direct impact of big data analytics on financial performance.
Institutionalists assert that organizations independently determine their responses to external influences, as supported by various studies (Oliver, 1991; Demirbag et al., 2007; Zhang and Dhaliwal, 2009; Liu et al., 2010; Zheng et al., 2013; Braganza et al., 2017).
In 1991, it was asserted that institutional theory can encompass proactive organizational behavior, as long as organizations do not always respond passively to institutional constraints in every situation.
Organizational culture significantly impacts human behavior, motivation, teamwork, and leadership within a company Defined as a collection of shared assumptions, values, and practices, organizational culture helps individuals navigate how a company operates Numerous studies highlight its role as a critical source of competitive advantage, emphasizing its importance in shaping organizational effectiveness and success.
BDAMAC plays a crucial role in Big Data Analytics (BDA) by facilitating informed business decisions through an effective management framework It encompasses four key themes: BDA planning, investment, coordination, and control The BDAMAC process begins with strategic BDA planning, which identifies business opportunities and assesses how big data models can enhance financial performance (FPER).
Investment decisions in Big Data Analytics (BDA) are crucial for Business Development and Management Accounting (BDAMAC), as they involve thorough cost-benefit analyses Research by Ramaswamy (2013) indicates that companies heavily investing in Big Data are achieving substantial returns and securing competitive advantages, thereby putting firms with minimal investment in Big Data at a significant disadvantage.
H1: BDA management capability will have a positive impact on financial performance
Big data requires innovative technologies to effectively manage its volume, diversity, and velocity, enabling the extraction of valuable and accurate insights (Gupta and George, 2016) Notably, relational database management systems (RDBMS) hold approximately 80% of their data in an unstructured format.
Enterprises are increasingly moving away from traditional RDBMS for data storage and analysis, opting instead for advanced technologies like Hadoop, which facilitates distributed storage and concurrent processing of unstructured files Additionally, Not Only SQL (NoSQL) databases have emerged as a popular solution for efficiently storing and retrieving non-relational unstructured data.
BDATEC highlights the adaptability of the BDA platform, emphasizing its connectivity of cross-functional data, compatibility with various platforms, and modularity in model building These features empower data scientists to swiftly develop, deploy, and manage a bank's resources The three key themes of BDATEC—connectivity, compatibility, and modularity—are essential for addressing fluctuating business conditions, such as shifts in competition, market dynamics, and consumer behavior Aligning resources with both long-term and short-term business strategies, including new product development and diversification, is crucial for success in this evolving landscape.
A flexible BDATEC enables banks to efficiently source and connect diverse data points from remote, branch, and mobile offices, facilitating the creation of compatible data-sharing channels across various functions The adaptability of a bank's Big Data Architecture (BDA) relies on two key components: the first being the connectivity among different business units to source and analyze a wide range of data from various functions, such as supply chain management and customer relationship management.
H2: BDA technology capability will have a positive impact on financial performance
BDATLC denotes the proficiency of analytics professionals in executing tasks within a big data environment This expertise, along with various other forms of knowledge, is categorized as capabilities, which can foster or maintain a competitive edge in the industry (Constantiou and Kallinikos).
The study emphasizes that analysts must possess four crucial skill sets: technical knowledge, technology management knowledge, business knowledge, and relational knowledge Technical knowledge encompasses expertise in operational systems, statistics, programming languages, and database management systems Technology management knowledge involves proficiency in visualization tools and techniques for effective management and deployment Business knowledge is essential for understanding both short-term and long-term organizational goals, while relational knowledge focuses on fostering cross-functional collaboration through effective information sharing.
To thrive in a data-driven world, new skills and management styles are essential (McAfee et al., 2012) Waller and Fawcett (2013) emphasize that Big Data Analytics (BDA) necessitates a diverse skill set, including statistics, forecasting, optimization, applied probability, and financial acumen Additionally, previous research highlights the importance of both technical and management skills as critical human resource qualities in the IT industry (Bharadwaj, 2000; Chae et al.).
According to Gupta and George (2016), recent IT competency studies highlight the essential role of human resources in developing Big Data Analytics (BDA) competence Human resources encompass various attributes, including experience, knowledge, business acumen, problem-solving skills, leadership qualities, and interpersonal relationships (Wade and Hulland, 2004; Akhtar et al., 2018).
Technology management knowledge encompasses the essential skills for managing big data resources to align with business objectives For instance, analytics experts at Netflix utilize visualization and demand analytics tools to gain insights into consumer behavior and preferences, which has significantly contributed to their success.
“House of Cards” program in the United States (USA) (Ramaswamy, 2013)
RESEARCH METHODOLOGY
Research Process
The analysis sample is crucial for accurately representing the relationship between variables in research, as it helps draw meaningful conclusions about the study's characteristics Sample collection plays a vital role, with larger samples enhancing precision but requiring more time and resources To address these limitations, the survey author adopted a convenient polling strategy, selecting readily available and willing respondents to optimize time and cost According to Tabachnick and Fidel (2007), the sample size should meet the formula n >= 50 + 8*p, which for a study with five variables translates to a minimum sample size of 90.
In which: n is Sample size p is number of independent variables of the model
To develop an effective regression model, a minimum of 170 samples is essential; therefore, the research team conducted a survey involving 250 banking professionals in Ho Chi Minh City, encompassing various levels from junior staff to senior management, to enhance the reliability of the findings The study also incorporated an Exploratory Factor Analysis (EFA), which, as noted by Hair et al (1998), necessitates at least 50 samples, ideally 100, with a variable ratio of 5:1, following the formula n >= 5 * q.
In which: n: Sample size q: Number of questions
The thesis has a total of 28 questions As a consequence of the formula (3.2), the minimum sample size for EFA discovery factor analysis is n > = 5 * 30 = 150 samples
As a result, the minimum number of samples needed to integrate EFA and regression analysis is 150
The researcher utilized questionnaires to elucidate scientific principles and conducted preliminary testing of these questionnaires This process allowed for necessary adjustments and enhancements before the official interviews were carried out.
Adjust the Scale
Table 3.1 The BDA Management Capability scale (BDAMC)
We continuously examine the innovative opportunities for the strategic use of big data analytics
Boynton et al (1994), Karimi et al (2001), Kim et al (2012), Sabherwal, (1999), Segars and Grover (1999)
When we make big data analytics investment decisions, we consider and project about how much these options will help end-users make quicker decisions
Kim et al (2012), Ryan et al
In our organization, business analysts and line people meet frequently to discuss important issues both formally and informally
Boynton et al (1994), DeSanctis and Jackson (1994), Karimi et al (2001), Kim et al
We are confident that big data analytics project proposals are properly appraised
Karimi et al (2001), Kim et al
Table 3.2 The BDA Technology Capability scale (BDATEC)
Compared to rivals within our industry, our organization has the foremost available analytics systems
All remote, branch, and mobile offices are connected to the central office for analytics
Software applications can be easily transported and used across multiple analytics platforms
Reusable software modules are widely used in new analytics model development
Broadbent et al (1999), Duncan (1995), Kim et al
Table 3.3 The BDA Talent Capability scale (BDATLC)
Our analytics personnel are very capable in terms of programming skills
(1999), Kim et al (2012), Lee et al (1995), Terry Anthony Byrd (2000)
Our analytics personnel show superior understanding of technological trends
Kim et al (2012), Terry Anthony Byrd (2000), Tippins and Sohi (2003)
Our analytics personnel understand our organization’s policies and plans at a very high level
Our analytics personnel are very capable in terms of planning, organizing, and leading projects
Boar (1995), Duncan (1995), Jiang et al (2003), Kim et al
(2012), Lee et al (1995), Terry Reflective Byrd (2000)
Table 3.4 The BDA Capability scale (BDAC)
We perform big data analytics planning processes in systematic and formalized ways
Boynton et al (1994), DeSanctis and Jackson (1994), Karimi et al (2001), Kim et al
We frequently adjust big data analytics plans to better adapt to changing conditions
Boynton et al (1994), DeSanctis and Jackson (1994), Karimi et al (2001), Kim et al
In our organization, business analysts and line people coordinate their efforts harmoniously
Boynton et al (1994), DeSanctis and Jackson (1994), Karimi et al (2001), Kim et al
BDAC4 Our analytics department is clear about its performance criteria
Karimi et al (2001), Kim et al
Table 3.5 The Bank Performance scale (BPER)
Bank performance (BPER): Using analytics improved during the last 3 years relative to competitors:
BPER1 We have entered new markets more quickly than our competitors
Tippins and Sohi (2003), Wang, Liang, Zhong, Xue, and Xiao (2012)
We have introduced new products or services to the market faster than our competitors
Our success rate of new products or services has been higher than our competitors
BPER4 Our market share has exceeded that of our competitors
Table 3.6 The Financial Performance scale (FPER)
Financial performance (FPER): Using analytics improved during the last 3 years relative to competitors:
Tippins and Sohi (2003), Wang, Liang, Zhong, Xue, and Xiao (2012)
Data Processing Methods
The Cronbach’s Alpha coefficient is essential for assessing the reliability of a scale, with higher consistency indicating greater reliability This method helps eliminate inappropriate variables, as pseudo-factors may arise without its application (Nguyen Dinh Tho, 2011) Prior to conducting exploratory factor analysis (EFA), the assessment of the scale using Cronbach's Alpha coefficient is crucial.
Variables with a total correlation below 0.3 are deemed irrelevant and should be eliminated, as they do not align with the research concept, provided that the Cronbach's Alpha coefficient is satisfactory (Nguyen Dinh Tho, 2011) The assessment of the scale is based on specific criteria.
- Cronbach’s Alpha reliability coefficient with reliability greater than 0.8: Good measurement scale
- Cronbach’s Alpha reliability coefficient with a reliability of 0.7 to 0.8: Usable scale
- Cronbach’s Alpha reliability coefficient is 0.6 or higher: The scale can be used.51
- Cronbach’s Alpha reliability coefficient is less than 0.6: Calibration of the scale is based on the elimination of low total correlation variables, thereby increasing the Cronbach’s Alpha coefficient
Before assessing a scientific theory, it's essential to determine the reliability and value of the measurement scale The reliability is evaluated using the Cronbach Alpha method, while the Explanation Factor Analysis (EFA) method assesses two key values: convergence and discriminant validity EFA is an interdependence technique that does not distinguish between dependent and independent variables, focusing instead on the relationships among them This method helps to condense a set of observations into meaningful factors (F < k), based on the linear relationships between these factors and the original observed variables.
Authors Mayers, L.S., Gamst, Guarino A.J (2000) mentioned that: In factor analysis, the method of extracting Principal Components Analysis coupled with Varimax rotation is the most commonly used method
According to the Hair and Ctg (1998), factor loading is an indicator to preserve the practical significance of EFA:
- Factor loading > 0.3 is considered to be the minimum
- Factor loading > 0.4 is considered important
- Factor loading > 0.5 is considered to be practical
The requirement for factor analysis is to satisfy the following requirements:
- 0.5 ≤ KMO ≤ 1: The KMO (Kaiser - Meyer - Olkin) coefficient is used to determine the suitability of factor analysis Large KMO values have factorial analysis as appropriate
Barlett testing is statistically significant (Sig.< 0.05): This is a statistical quantity used to consider the hypothesis that the variables have no correlation overall If this test is statistically significant (Sig < 0.05)
This statistical measure evaluates the hypothesis of no overall correlation between variables A statistically significant result (p < 0.05) indicates that the observed variables are correlated.
Regression is a statistical modeling technique used to predict the value of a dependent variable based on one or more independent variables In simple regression, the model examines the relationship between a single independent variable and the dependent variable, while multiple regression involves multiple independent variables Linear regression specifically represents these relationships with a straight line, known as the line of best fit, which effectively summarizes the data.
Simple linear regression demonstrates the relationship between variables dependent on a single variable The regression model is as follows (3.3): γ = α + βX + ɛ γ: the dependent variable α: the intercept β: slope coefficient
X: the independent variable ɛ: random error component
One-way ANOVA, or one factor analysis, is employed to evaluate the average hypothesis among sample groups, with a significance level set at 5% When conducting ANOVA, certain assumptions must be considered to ensure the validity of the results.
- Comparative groups must be independent and randomly selected
- Comparative groups must have a standard or sample size that is large enough to be considered as a standard approach
- Variance of comparative groups should be uniform The ANOVA test results consist of two parts:
Part 1: Levene tests are used to test variances equally between groups
- Sig ≤ 0.05: accept Ho, enough condition to analyse ANOVA
Part 2: ANOVA test - Ho: average equal
- Sig ≤ 0.05: reject Ho à eligible to confirm there is a difference between groups for dependent variables
When the significance level (Sig.) is greater than 0.05, we accept the null hypothesis (Ho), indicating that there is insufficient evidence to assert differences between groups for the dependent variables If differences are detected, further analysis can be conducted using Tukey, LSD, Bonferroni, or Duncan tests, collectively known as Post Hoc testing in the context of ANOVA.
RESULT AND DISCUSSION
Overview of the research samples
In chapter 4, the results of the study will be presented after the analysis on SPSS
This article discusses 20 software tools that facilitate various statistical analyses, including the evaluation of sample data, reliability assessment using Cronbach’s alpha coefficient, exploratory factor analysis (EFA), ANOVA testing, linear regression, and hypothesis testing for research models.
The research involved surveys conducted at banks in Ho Chi Minh City using a soft questionnaire rated on a scale of 1 to 5 Initially, 250 responses were collected through Google Forms; however, 35 questionnaires were excluded due to incomplete answers or uniform responses throughout Consequently, the final dataset for the study comprised 215 valid questionnaires.
Table 4.1 Descriptive statistics for Type of Banks (Type) variables
The study involves a total of 215 banks, with domestic banks representing 55.8% and foreign banks making up 44.2% This skew towards domestic banks may lead to a biased perspective, particularly since the survey was conducted in Ho Chi Minh City, where domestic banks dominate the market This disparity highlights the need for a more balanced representation of both domestic and foreign banks in the study.
Table 4.2 Descriptive statistics for Data usage frequency variables (UF)
Based on Table 4.2, the author find that there are 86 answers of "Having used but not handled" held the lion’s share, with 40%, followed by that of "Occasionally" and
The research reveals that "No use" accounts for 42 responses (19.53%) and 27 responses (12.56%), highlighting a significant focus on the "Occasionally" and "Having used but not handled" groups Notably, the study emphasizes the perspectives of the "Having used but not handled" group, as the majority of survey participants are employees who frequently engage with, process, and utilize data in their work, indicating a distinct difference in their experiences.
4.1.2 Descriptive statistics for independent variables
Table 4.3: Descriptive statistics for variables describing BDA management capability (BDAMC) BDA management capability
The results from Table 4.3 indicate that among the 215 participants, opinions varied significantly, with responses ranging from a minimum of 1 (strongly disagree) to a maximum of 5 (strongly agree) While the initial research hypothesis anticipated a positive response within the range of 3 to 5, the findings reveal a standard deviation from this expectation Nonetheless, the average scores for each scale fall within the 3 to 5 range, suggesting that the research data regarding BDA management capability, BAD technology capability, BDA talent capability, BDA capability, bank performance, and financial performance remains acceptable.
Cronbach’s Alpha Analysis
The analysis of Cronbach's Alpha for the independent variables reveals key components of Big Data Analytics (BDA) capabilities The BDA management capability consists of four observed variables: BDAMC1, BDAMC2, BDAMC3, and BDAMC4 Similarly, the Bank Technology Capability is represented by four variables: BDATEC1, BDATEC2, BDATEC3, and BDATEC4 The BDA Talent Capability also includes four observed variables: BDATLC1, BDATLC2, BDATLC3, and BDATLC4 Furthermore, the overall BDA Capability comprises four variables: BDAC1, BDAC2, BDAC3, and BDAC4 Lastly, the Bank Performance is assessed through four observed variables: BPER1, BPER2, BPER3, and BPER4.
Table 4.4 Cronbach’s Alpha coefficient of Indepenent Variables
Scale Mean if Item Deleted
Scale Variance if Item Deleted
Cronbach's Alpha if Item Deleted BDA Management Capability (BDAMC)
The analysis revealed that the Cronbach's Alpha coefficients for BDAMC (0.796), BDATEC (0.859), BDATLC (0.887), BDAC (0.803), and BPER (0.842) all exceed the acceptable threshold of 0.7, indicating strong reliability Additionally, the Corrected Item-Total Correlation for the scale's variables is above 0.3, further confirming the reliability of the subjective variables Consequently, the measurement variables for this component have been retained for the Exploratory Factor Analysis (EFA).
Table 4.5 Cronbach’s Alpha coefficient of Indepenent Variable
The findings indicate that the Cronbach's Alpha coefficient for financial performance exceeds 0.7, demonstrating a high level of reliability Additionally, the Corrected Item-Total Correlation for the variables in the scale is above 0.3 Consequently, the subjective variable scale is deemed reliable, and the measurement variables for this component were retained for the EFA analysis.
Exploratory Factor Analysis (EFA)
4.3.1 EFA analysis for Independent Variable Scale
The reliability of the scales was assessed using Cronbach’s alpha coefficient, followed by an evaluation through Exploratory Factor Analysis (EFA) This method focuses on two key value types: convergence and discriminant values The evaluation criteria include factor load values, the KMO coefficient, Bartlett's test, and the percentage of total variance explained The Independent Variable Scale comprises five components with 20 observed variables After confirming the scale's reliability via Cronbach’s Alpha, all variables were subjected to EFA analysis.
Table 4.6 KMO and Bartlett's Test KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling
The factor analysis results indicate a KMO coefficient of 0.810, which exceeds the threshold of 0.5, and a significance level of 0.000, which is less than 0.5 This leads to the rejection of the null hypothesis, suggesting that the observed variables are correlated Consequently, the factor model hypothesis is deemed inappropriate and rejected, confirming that the data utilized for the factor analysis is highly suitable.
Table 4.7 Total Variance Explained of Independent variables
Initial Eigenvalues Total % of Variance Cumulative (%)
Based on the results table, at the Eigenvalue value = 1.444 > 1, the total extracted variance is 67.569% > 50% Thi s means that these 20 observed variables can explain 67.569% of the data variability
Table 4.8 Rotated Component Matrix a of Independent variables
Using Principal Components extraction method with Varimax procedure rotation, loading coefficient greater than 0.5 is considered to be of practical significance The result shows that all variables are accepted
4.3.2 EFA analysis for Dependent Variable Scale
The reliability of the scales was assessed using Cronbach’s alpha, followed by an Exploratory Factor Analysis (EFA) to evaluate the convergence and discriminant values Key evaluation criteria included factor load values, KMO coefficients, Bartlett's test, and total variance percentages The Dependent Variable Scale, which measures financial performance, comprises four observed variables After confirming the scale's reliability with Cronbach’s Alpha, all variables were included in the EFA analysis.
Table 4.9 KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling
The factor analysis results indicate a KMO coefficient of 0.764, which exceeds the threshold of 0.5, and a significance level of 0.000, which is below 0.5 This leads to the rejection of the null hypothesis that the observed variables are uncorrelated, confirming that the factor model is appropriate Therefore, the data utilized for the factor analysis is deemed highly suitable.
Table 4.10 Total Variance Explained of Dependent variable
Initial Eigenvalues Total % of Variance Cumulative %
Based on the results table, at the Eigenvalue value = 2.279 > 1, the total extracted variance is 56.977% > 50% This means that these 4 observed variables can explain 56.977% of the data variability
Table 4.11 Rotated Component Matrix a of Dependent variables
Using Principal Components extraction method with Varimax procedure rotation, loading coefficient greater than 0.5 is considered to be of practical significance The result shows that all variables are accepted.
Linear Regression and ANOVA Analysis
Std Error of the Estimate
Total 97.086 214 a Dependent Variable: FPER b Predictors: (Constant), BPER, BDAMC, BDATEC, BDAC, BDATLC
The analysis yielded a statistically significant coefficient of determination (R² = 0.593; F = 60.926; Sig = 0.00 < 0.05), indicating that the five independent variables—BDA management capability, BDA technology capability, BDA talent capability, BDA capability, and bank performance—explain 59.3% of the total variation in financial performance This suggests that the remaining 40.7% is influenced by extrinsic variables and random errors Consequently, the model effectively demonstrates the relationship between big data analysis and financial performance through the identified predictor variables, confirming its validity as a predictive model.
Hypothesises testing
The regression coefficient analysis and "t" test at a significance level of 0.05 indicate that five out of six factors were significant, while the variable BDAMC, with a coefficient of -0.010 and a significance level of 0.875, did not meet the criteria for inclusion Consequently, it is essential to exclude this variable from the regression model.
The normalized regression coefficient shows that the important variable in the model is the BDAC variable, the second is the BDATLC variable and the third is the
❖ Dependent Variable of financial performance
According to table 4.14, it can be seen that:
• BDAC: Beta = 0.369 (t = 6.977, sig.= 0.000 < 0.05) As a result, BDAC variable is statistically significant
• BDATEC: Beta = 0.122 (t = 2.703, sig.= 0.007 < 0.05) As a result, BDATEC variable is statistically significant
• BDATLC: Beta = 0.292 (t = 5.277, sig.= 0.000 < 0.05) As a result, BDATLC variable is statistically significant
• BPER: Beta = 0.124 (t = 2.474, sig.= 0.014