Big data and human resource management: paving the way toward sustainability

The power of big data (BD) is growing within organizations. It is transforming the landscape of economic value creation (Blazquez and Domenech, 2018) while opening new and fascinating scenarios in human resource management (HRM) processes that are the highest value for any organization and a rare, inimitable and irreplaceable mine of production factors (Purcell et al., 2003; Pfeffer, 1994; Prahalad, 1983; Ratner, 2020).

The potentialities and benefits that these huge amounts of data originated are enormous (Wamba et al., 2017; Resch et al., 2021). Making good use of tetra-bite information requires enormous cultural efforts within the organization that only human resources, the driving force of the company's individual, interpersonal and organizational change, can support.

Nevertheless, few types of research have tackled the impact of BD in HRM and its various applications in organizations. Thus, exploratory empirical literature on the topic remains at a rudimentary state (Gupta and George, 2016; Wamba et al., 2017), reducing BD's application strategic potential in HRM (McAfee et al., 2012; Mikalef et al., 2018). Effectively, one of the most significant corporate applications of BD consists of contributing to a better understanding of changes and the making of predictions (Jun et al., 2018) within a multidimensional framework including “IT infrastructure, operational, organizational, managerial and strategic aspects” (Wang et al., 2016, 2020).

Throughout tools such as data mining and business analytics, techniques that use algorithms to analyze enormous quantities of data such as information on "employment history, skills and competencies, formal educational qualifications, and demographic information” (Angrave et al., 2016; Alaimo and Kallinikos, 2021), career and salary, BD are relevant in the transformational processes of companies, which aim to plan and create value. It is a new value that produces a responsible and sustainable competitive advantage.

However, also research on the impact of BD on sustainability (in terms of sustainable performance) is discontinuous (Dubey et al., 2017) and limited, leaving a largely unchartered territory of research, even if it is critical to grasp better how to use BD to render it an HRM's competitive advantage for managing structures, processes and organizational cultures.

Therefore, this paper intends to find out when and how the integration of BD technologies and HRM practices (in terms of training) enhances firms' operational and sustainable performance.

The conceptual idea of the paper starts from the Triple Bottom Line (TBL) model, which considers three components: the natural environment, society and economic performance (Elkington, 1994, 2004). According to this model, the sustainable advantage, which translates into economic, social and environmental benefits, considers employees as stakeholders and adopts an integrated vision of HRM with long-term strategic planning.

The primary purposes of this study are three: (1) describe and explain the existing scientific research, develop a systematic literature review to bring to light the ability of companies to integrate and improve HRM with BD technologies, as well as their influence on sustainable performance; (2) using the findings of the review for identifying a conceptual holistic research model; (3) testing empirically the model with a study: through a holistic approach, the document proposes empirical research employing Eurostat data to test the benefits of BD in HRM decisions that optimize the relationship between training, productivity and well-being.

In order to achieve this, the paper is structured in seven sections. After the introduction (Section 1), Section 2 describes the main steps adopted to conduct a systematic literature review. Section 3 offers the data sources and search strategy from which it is possible to deduce the main research areas observed by literature studies on BD and HRM. Section 4 develops empirical research on the VAR model's relationship between training, productivity and well-being. Section 5 shows the results of empirical research, while Section 6 discusses the results of systematic literature review and empirical research, as well as identifying avenues for further research. Lastly, Section 7 serves as the conclusion.

In this paper, we have used the methodology by Ngai and Wat (2002), Kitchenham (2004), Brereton et al. (2007), Kitchenham et al. (2009), for a systematic literature review. First, a revision protocol followed; then, the inclusion and exclusion criteria were established to reveal significant research for the analyzed topic. This was followed by the critical evaluation, the extraction of the data and the synthesis of the results. The following figure describes in detail the process (Figure 1).

A systematic literature analysis was developed in the Web of Science (WOS) and SCOPUS databases. The descriptors carefully chosen to examine the scientific production were “Big Data,” “Human management,” “Human resources management” and “Talent management.”

Investigation showed that scientific production started in 2013 and has 296 papers. Those written in “English” were 74 (70 in Journals and 3 in Trade Publication), all published between 2013 and 2020 (see Table 1).

Chinese, American and English authors (see Table 2 and Figure 2) mainly focused on research areas such as Engineering, Computer Science and Business, Management and Accounting, even though other regions were concerned (see Table 3) (see Figures 3–8).

Selected journals and highly cited articles are in the following Tables 4 and 5:

In the following figures are visible the maps of the citation by countries, citations by authors and coauthorships:

Following this classification, all papers were examined, and those out of our study's scope were excluded. This second screening process was based on the analysis of “Abstract and Keywords,” and the result gave 45 relevant articles (Figures 9 and 10) using a combination of the following keywords: “Big Data,” “Human Management,” “Information Management,” “Data Analytics,” "Big Data Analytics,” “Decision Making,” “Human Resources Management,” “Human,” “Resource Management,” “Human Resource,” “Machine Learning,” “Personnel,” “Sustainable Development,” “Artificial Intelligence,” “Innovation,” “Internet Of Things,” “Managers,” “Natural Resources Management,” “Performance Assessment,” “Big Data Era,” “Employee,” “Employment,” “Enterprise Human Resource,” “Environmental Management,” “Green Innovations,” “HR Analytics,” “Industry 4.0,“ “Information Management Systems,” “Information Systems,” “Management,” “Organizational Performance,” “Performance Management,” “Quality Management,” “Talent Management,” “AI,” “ Environmental and Organizational Performance,” “Behavioral Research,” “Big Data Analysis,” “Big Data Analytics (BDA),” “Big Data Applications,”“Big Data Decision-making Capability,” “Blockchain,” “Business Intelligence,”Artificial Intelligence and Talent Management.”

The third phase of the process, assessed based on subjects, consisted in identifying relevant articles that allowed focus only on 25 papers, classifying them by research approach into seven categories: Conceptual, Review, Data Analysis, Survey, Experimental, Case Study and Developmental (see Table 6).

Our investigation revealed that BD usage could benefit different categories of HRM (see Table 7).

At the fourth step of our investigation, it became necessary to introduce new descriptors, such as “sustainability,” “environmental management” and “Triple Bottom Line.” From this further evaluation, only four papers appeared (3 articles and 1 article in press) and were published between 2016 and 2018 in the following journals: “Human Development and Interaction in the Age of Ubiquitous Technology,” “International Journal of Information Management,” “Journal of Cleaner Production” and “Technological Forecasting and Social Change.”

Following their classification by research areas, two main categories were identified: Business, Management and Accounting, and Computer Science, even though other regions were treated, as demonstrated in Table 8.

At the final phase of the process, the four selected papers were crossed with the other previously analyzed documents. Two papers were finally excluded because they were out of the scope of our study. In the end, only two articles were in line with the research topic (El-Kassar and Singh, 2019; Singh and El-Kassar, 2019). Both papers highlight the links between HRM, sustainability, BD and performance (as shown in Table 9).

This paper has developed a literature review of BD, HRM and sustainability in the largest abstract and citation database of peer-reviewed publications called SCOPUS. The literature review found 25 relevant articles published on these topics. The analysis of the related papers shows that since 2013, new literature has continued to emerge, highlighting how BD are being applied to revolutionize HRM, indicating that the transformation is underway. However, there has been little indication (only two papers) on the impact of environmental sustainability on BD and HRM and vice versa, even if the “green economy” is imposing a profound rethinking of the HRM.

Findings indicate that, at varying levels, scholars from all research areas who are particularly interested in Engineering, Computer Science and Business, Management and Accounting fields are demonstrating a growing interest in BD and HRM, especially those from China, the USA and the UK. BD research is emerging as an essential discipline in the HRM field. Virtually all HRM subareas deserve special attention, including the following: people resourcing, learning and development, career, employee relations, performance and awards and a flexible work system. Investigations are mainly conducted through case studies and conceptual papers, meaning an increasing need to develop an integrated and standardized method for managing human resources in organizations.

For the conceptual framework, this paper draws upon the dynamic capabilities (DC) approach, such as in Singh and El-Kassar's (2019) studies put in light of the literature review. According to this theoretical lens, companies need to constantly develop resources able to integrate and/or reconfigure competencies internally and externally (Teece et al., 1997). Moreover, in this perspective, the company's ability to renew its skills needs to be in tune with environmental changes. In this regard, several BD studies have demonstrated the applicability of the dynamic capabilities view, arguing that BD technologies are resources that support firms operating in turbulent contexts (Wamba et al., 2017; Dubey et al., 2017) and examining their influence on firm performance.

Considering the context of sustainability, dynamic capabilities can be referred to as the ways in which companies implement and reconfigure BD technologies and HRM practices to achieve sustainable performances (e.g. Eisenhardt and Martin, 2000; Teece et al., 1997). It's evident, therefore, that a strong accent needs to be laid on human capital adapting to the company's new requirements.

Based on these observations, this paper develops a conceptual model that focuses on HRM decisions aiming to test the relationship between training, productivity and well-being. It consists of 6 factors, namely: education (E), IT training (ITT), IT personnel (ITP), productivity (P), well-being (WB) and social exclusion (SE).

Understanding the effects of these factors can support managerial strategies in effectively implementing sustainable HRM practices to perform better, be more competitive, and gain more market quotas, despite the turbulence. Adaptability theory proposes that organizations must possess the ability to adapt to changing circumstances to achieve sustainable success. Integrating BD analytics (BDA) into HRM, practices offers a valuable tool for enhancing organizational adaptability and ultimately fostering sustainability. This article explores how BD and HRM can be combined using the lens of adaptability theory to pave the way toward sustainable practices. Adaptability theory provides a valuable framework for understanding the role of BDA in HRM and its influence on organizational sustainability. By integrating BDA into HRM practices, organizations can enhance their adaptability by improving environmental scanning, sense-making, flexibility, learning, employee empowerment and change management. This integration ultimately paves the way toward sustainable practices by enabling organizations to make data-driven decisions, respond to emerging trends and align HRM strategies with sustainability goals.

The link between training, performance and wage is one of the most researched topics in the field of economics (Bartel, 1995). However, the implications of these variables are far-reaching, with echoes in macroeconomic variables such as GDP per capita and social exclusion.

To estimate the impact of HRM decisions on employees' performance and observe the implication of these decisions in a larger picture from a macroeconomic point of view, we have considered both microeconomic and macroeconomic variables. Thus, having in mind the implications, both from macro and microeconomic points of view, and in order to test and contextualize these relationships, we have developed empirical research, implementing a VAR model and taking into consideration two variables that are responsible for BD generation: IT training offered by enterprises and IT personnel, focusing upon the HRM decisions that are aiming to optimize the relationship between training, productivity and well-being.

Our hypotheses are based on the augmented Solow neoclassical and the “new growth theories,” and the role of education on productivity, well-being or social exclusion is almost axiomatical. Thus, human capital, as an extra input, directly influences productivity and hence well-being and social exclusion. According to Sianesi and Reenen (2003), a one-year increase in average education is found to raise the level of output per capita by between three and six percent according to augmented neoclassical specifications, while leading to an over one percentage point faster growth according to estimates from the new-growth theories.

Thus, starting from the findings of the literature review and the macro and microeconomic implications, we were able to formulate the following hypothesis:

Hypotheses 4a and 4b are related to training and its impact on productivity. Training increases the marginal product of labor more than the wage, which creates incentives for the firm to invest in general training (Konings and Vanormelingen, 2015). In addition, taking into consideration the fact that the industry is full of 4.0 transformation, it makes sense the need to invest in IT&C personnel and the training by which they can transfer know-how and skills to other workers. However, IT&C personnel, considering their average salary, imply additional costs for the entity. Therefore, a cost-benefit analysis is required on this issue. In order to perform the case study, the following variables have been considered: GDP per capita, employees with tertiary education, employees in the IT field, real labor productivity per person, people at risk of poverty or social exclusion, enterprise-provided training to their personnel to develop their ICT skills.

For our case study, we have used an unrestricted panel VAR, one of the most flexible and versatile research instruments, as we can test different hypotheses. Thus, because of the interdependencies between the variables selected, and the bidirectional impact between them, we have concluded that a VAR panel model is the optimal type of research methodology. The model that we have opted for is original and, since the variables are integrals of different orders, is taking into consideration the recommendations of Harvey (1990). According to Canova and Cicarelli (2013), panel VAR models are best at determining the explicit microstructure present in DSGE models and, as their VAR counterparts, attempt to capture the dynamic interdependencies present in the data using a minimal set of restrictions.

Thus, because of the interdependencies that are common between our variables and the impact between them is mutual, we have opted for this unrestricted VAR Panel model, which because of its versatility, can be easily used for testing the hypothesized relationships between the variables considered. Thus, using the Eurostat database, we have taken 28 cross-sections regarding the European Union member states from 2012 until 2018.

Model:

LEGEND:

• Variable Explanation

• GDP GDP per capita

• Employees3 Employees with tertiary education

• Itemployees Employees in the IT field

• Productivity Real labor productivity per person

• Exclusion People at risk of poverty or social exclusion

• Ittraining Enterprise provided training to their personnel to develop their ICT skills

After the estimation of the VAR, all the necessary tests were performed in order to continue further with the VAR methodology. As observed in Table 10, the number of lags is 1.

Also, in Table 11, it can be observed that the VAR satisfies the stability condition after performing the stability test.

The VAR satisfies the autocorrelation condition, as presented in Table 12. For assuring the autocorrelation, it used the LM test, and it is possible to observe that for lag 1, the condition is satisfied.