A Study on the Corporate ESG Impacts on the Company’s Internal Management Performance and ESG Management Types by Industry

2.1 ESG Framework

Fundamentally, ESG is generated as an investment principle. However, ESG investing focuses on non-financial dimensions of the business related to the environment, social, and government fields (Duuren et al., 2016). Compared to the traditional investment principle, which invests in financial profitability, ESG is considered a value-driven investment method that appeals company’s sustainable structure and relations. Beyond the range of the investment, ESG expands to the overall management practice of the company as a part of the global green innovation trend corresponding to the increasing ethical consciousness of the public and the global climate crisis. Thus, the current ESG principle is integrative since it covers not only management areas such as resources, products, energy, and other management part but also can be an attractive marketing tool and even have to be applied to accounting. However, as Li et al. (2021) indicated, it is obscure to confine the components of ESG since ESG is developing concepts proposed in 2004 in the United Nations report (2004) under the financial sector initiative to encourage sustainable investment. For instance, generally, ESG is used to measure corporate value from the sustainability perspective; however, this concept can also be applied to measure the nation’s performance. In the case of the World Bank, it categorizes the ESG components at the national level in <Table 1>.

Table 1.

World Bank ESG Framework

Also, to compare the ESG performance on the firm level, compared to the World Bank’s macro-approach at the national level, Refinitiv and S&P, which score companies’ ESG performance, utilize standards that focus more on the company side. However, unlike Refinitiv, S&P has different components in its ESG evaluation system, as shown in <Table 2>. Likewise, due to the importance of ESG ad sustainable development of the business, various institutions and governments establish various ESG evaluation frameworks. However, as LI et al. (2021) indicated, those various frameworks and standards can intensify the companies’ confusion and stress.

Table 2.

ESG Framework Comparison

Thus, to verify the effectiveness of ESG framework and management, its validness has been studied frequently, and various indicators such as financial profitability, stock valuation, risk, and so on have been examined as performance indicators with various variables (AYDOGMUS et al., 2022; Budsaratragoon and Jitmaneeroj, 2021). However, considering the effort of the company to enhance its ESG management for enhancing its valuation to attract investors, it is evident that ESG substantially influences internal stakeholders such as employees, shareholders, and customers (Zumente and Bistrova, 2021; Liang and Renneboog, 2020). Likewise, though the ESG influences various stakeholders as in <Figure 1>, the study on the impact of ESG components on the major stakeholders’ loyalty is limited (Tarmuji et al., 2016). Thus, this paper aims to analyze the impact of the company’s ESG performance on the major stakeholders’ loyalty.

Figure 1.

Stakeholder Relationships in ESG

2.2 ESG as an Innovation Management

ESG obligations require the company to improve its business in multiple aspects. However, the effectiveness of the ESG on the innovative management aspect has been suspicious since there are concerns, such as whether ESG can demotivate the employees by imposing a regulatory burden and deteriorate customer satisfaction by affecting the quality of its product and service. Based on the doubt about the validity of ESG, Tan and Zhu (2022) examines the impact of ESG as innovative management and verify that it can improve the quality and quantity of green innovation by increasing the consciousness of managers. Furthermore, Tan and Zhu (2022) indicate the effect of ESG, which is more effective for companies under high environmental regulation and market competition during their growth stage. Though Tan and Zhu define the quantity of green innovation as the number of green patents and measure its quality through citation with ESG ranking and governance variables mainly such as board (2022), based on the study, the components of ESG vary. The study of Simone et al. (2022) verifies the relationship between ESG factors and economic sustainability from the innovation perspective. This study utilizes the variable combining the company’s research and development (R&D) intensity and expected growth opportunity. Alareeni and Hamdan (2020) use the data of ESG Bloomberg, which include information about whether the company publicizes the environment, social, and government status transparently. The sub-variable of major ESG components consists of the company’s energy usage, resource treatment, pollution, business relationship, employee health, and others. Alareeni and Hamdan use the financial performance of the company, such as the return on asset (ROA) and equity (ROE), and attest that the company with higher asset and financial leverage tends to be more proactive in ESG management and ESG has positive influences on overall companies’ operation, financial, and market performance.

Compared to the studies which analyze the impact of ESG on the company’s financial performance, the study of Tarmuji et al. (2016) conducted a study about the ESG impact on economic performance using the stakeholders’ loyalty as a dependent variable, focusing on the case of Malaysia and Singapore. In this study, though the environmental variables positively influenced both Singapore and Malaysia companies, however, Singapore companies have more positive relations with social components, and the Malaysian companies are revealed to have more impact from governance components (2016). Similar to Tarmuji et al. (2016) study, this paper also utilizes the data from Refinitiv, named the ASSET 4 dataset, since it has comprehensive data about the ESG variables and the loyalty of the major stakeholders which directly related to the businesses. Based on the assumption that employees, shareholders, and customers are mostly influenced by a company’s ESG management (Zumente and Bistrova, 2021), in addition to Tarmuji et al. (2016) study, this paper aims to examine the relationship between sub-factors constructing ESG components and economic performance focusing on loyalty variables.

3.1 Data Description

This paper utilizes the data provided by Refinitiv. Refinitiv is a company that manages and services financial market data. Refinitiv collects the companies’ ESG data from more than 76 countries. ESG data is collected comprehensively by referring to corporates’ annual reports, websites, NGO websites, stock market information, CSR reports, and media sources.

From this database, this paper uses data with 130983 observations collected from the 2002 to 2021 fiscal year and evaluates the ESG performance of the 12156 companies through their released reports under ten major standards with 719 variables. The variables in this dataset are considered sufficiently consistent and reliable by having Cronbach’s Alpha value as 0.67, 0.77, 0.87, 0.85, 0.71, 0.60, 0.61, 0.68, 0.87, 0.61, 0.68, 0.87, 0.83, 0.79, 0.90, 0.89, 0.91, 0.75, and 0.90 for each ER, PI, RR, BD, IN, SH, Cu, So, Wo, EL, SL, CL, ENV, SOC, GOV, and EC. As <Table 3>, this dataset consists of three levels. The first level (L1) consists of ESG factors and performance variables, and the second level (L2) categorizes variables based on which stakeholders will effect. The third level (L3) variables consist of business practices for ESG management.

Table 3.

Variable Description

The Environment (ENV) has Emission Reduction (ER), Production Innovation (PI), and Resource Reduction (RR). Social (SOC) is constructed with two levels, it has Product Responsibility (PR) for the Customer (CU) dimension, Community (CO), and Human Rights (HR) for the second level (L2) Social (SO) dimension, and Diversity & Opportunity (DO), Employment Quality (EQ), Health and Safety (HS), Training & Development (TD) under Employee (EE) dimension. Corporate Governance (GOV) has Board Function (BF), Board Structure (BS), and Compensation Policy (CP) under the Board dimension. CG also has two other dimensions in the second level: Integration (IN) and Shareholders (SH), which have Vision & Strategy (VS) and Shareholders’ Rights (SR) as sub-variables for each. For the Economic (EC) variable, it has Margins (MA), Profitability (PR), and Revenue (RE). PR includes shareholder loyalty (SL), and RE includes client loyalty (CL). Thus, MA consists of ESG management commitment and effectiveness on employees.

Though the exogenous variable such as country and industry is attested as having an influence on companies’ ESG performance (Shinghania and Saini, 2023; Kumar et al., 2016), since this paper aims to verify the influence of ESG on major stakeholders, it uses company’s industrial information on the clustering analysis to define industrial differences in ESG performance features.

3.2 Research Model

To examine the impact of ESG components on the Economic factor among the various variables, it is required to select significant and distinctive variables. Thus, as in <Table 4>, this paper conducts the data preparation process, Explanatory (EFA), and Confirmatory factor analysis (CFA) prior to forming the Structural Equation Model (SEM) to analyze the causal relationship between ESG and the internal economic performance of the company. Internal economic performance is defined as variables related to employees, customers, and shareholders’ values.

Table 4.

Research Process

3.3 Exploratory Factor Analysis (EFA)

This paper conducts the Exploratory Factor Analysis (EFA) to select the significant variable among various variables. Prior to the Confirmatory Factor Analysis for measuring the consistency of the structure and factors, EFA based on the Maximum Likelihood (ML) method denotes the proper number of factors. However, before conducting EFA, this paper conducts Kaiser-Meyer Olkin (KMO) model to measure sampling adequacy through the variance proportion in the items, measuring commonality (1). As Li et al. (2020) indicated, KMO should be over 0.5 to be acceptable. Furthermore, on the EFA, the fitness of the model can be verified through the Tucker ＆ Lewis Index (TLI) (2) and Root Mean Square Error of Approximation (RMSEA) which is acceptable when it is over 0.9 and less than 0.08 per each (Xia and Yang, 2019) (3).

3.4 Confirmatory Factor Analysis (CFA)

After deciding the number of the factor and selecting the variables which can fit into the model, this paper conducts the Confirmatory Factor Analysis (CFA) to verify the fitness of the measurement variable corresponding to the structured model. Through the CFA, this paper can attest the whether the measurement model fits the data. Furthermore, since CFA indicates the comprehensive relationships between the variable and factors, through CFA, it is available to identify the potential inconsistencies of the model. Also, by examining the individual factor loadings prior to the SEM, CFA can identify specific variables which contribute more to the factors (Woods & Edwards, 2011). On the CFA, to measure the goodness of the fitness, in addition to the TLI and RMSEA, Chi-square (4), Standardized Root Mean square Residual (SRMR) (5), Goodness of Fit Index (GFI) (6), and Adjusted GFI (AGFI) (7) would be analyzed with Chi-square test (Schermelleh-Engel et al., 2003). Fundamentally, the Chi-square test identifies the difference between observed and expected covariance, which implies that the lower value has a better fit (Alavi et al., 2020). SRMR denotes the discrepancy between the sample covariance and the model covariance. SRMR less than 0.08 is perceived as acceptable (Vellone et al., 2014). AGFI is the corrected indicator of GFI, reflecting the number of indicators per latent variable. GFI is the fit between observed variance and the hypothesized model, and GFI over 0.9 is considered acceptable (Huang et al., 2019). Similar to GFI, Comparative Fit Index (CFI) (8) compares the hypothesized model with a null model where none of the variables are related to one another (Alavi et al., 2020).

3.5 Structural Equation Modeling (SEM)

To examine the causal relationships between multiple variables, SEM can be applied since it is a multivariate technique that combines regression and factor analysis. Through SEM, latent variables can be measured through observed variables, and their relationships can be modeled using structured equations. Similar to CFA, the fitness of SEM estimated through ML can be evaluated through Chi-square, RMSEA, TLI, SRMR, GFI, and AGFI. Though the SEM is a comprehensive statistical model available for measuring direct and indirect impact with factor analysis, linear regression, path, mediation, and other analysis (Kline, 2011), this paper only aims to measure the direct relationship between stakeholders’ performance value with ESG factors.

3.6 K-mean Clustering

After model construction through SEM, this paper utilizes K-mean Clustering to categorize the ESG performance by industry. Through the clustering method which is one of the machine learning methods based on unsupervised learning, this paper expects to define an industry that has similar ESG status and detailed ESG features. For clustering, kmeans function in R, stats packages are used based on the Euclidean distance. Based on the algorithm, the position of the K-mean centers becomes optimized as the following equation (Sinaga and Yang, 2020).

4.1 Factor Analysis

Refer to <Figure 2>, since the eigenvalue of the EFA scree plot converges on factor around 12, this paper utilizes CU, PI, IN, ER, SH, So, Wo, RR, and BD as independent factors and EL, SL, and CL as dependent factors referring to the conceptual framework provided by Refinitiv as <Table 3>. However, though EFA analysis denotes similar factors with the Refinitiv framework by having its L2 factors, for ENV, L3 factors have statistically heterogeneous to the others by satisfying the 0.4 cutoff standard on its loading in Appendix 1 (Wu et al., 2014). This EFA analysis also implies that it is not necessary to utilize detailed standards but factors on L2 can represent the sub-variables sufficiently.

Figure 2.

EFA Scree Plots

Its overall KMO is over 0.977, which is over 0.5, and implies having a proper correlation matrix. EFA has RMSEA and SRMR as <Table 5>, which satisfies the cutoffs. Therefore, replacing the GFI and other indexes in CFA, this paper analyzes fit, which is fair as much as it is close to 1 and complexity level with a degree of freedom (df) in EFA. The study of Pettersson and Turkheimer (2010) describes Hofmann’s complexity index as when it is 1, it indicates the perfect simple structure, and when it is over 1, it means evenly distributed items that are sufficient to further analysis.

Table 5.

EFA Good of Fitness Summary

Also, as in <Table 6>, GFI in CFA also satisfies its cutoffs. Through CFA, denoted in Appendix 2, under the ER factor, the variables related to emission, and waste reduction have significant impacts. Also, companies’ efforts to reduce the impact on biodiversity and construct proper environmental management systems are also considerable variables. In the case of PI, companies’ policies for eco-friendly designed products, eco-labeling, life-cycle assessments, and product innovation also denote a high relationship with their investment and consideration of product sustainability and de-materialization. For RR, the variables related to resource efficiencies, such as water, energy, packaging, and others are considered important variables, however, imposing a resource efficiency policy to the supply chain has fair significance to the model. Under the governance factor, BD has auditing, compensation, and nomination as sub-variables with improvement tools. Though there are various variables in IN factor, including compensation, committee, global compact, and others, only integrated strategy policy and CSR-related variables are having a significant impact. Under the SH factor, compared to the other variables such as voting procedures, controversies, and else, their voting rights, sense of engagement, and transparency indicate more impacts. For the social factors, from the Cu perspective, product access and lower price are denoted as important variables, and others that indicate the responsibility for the product such as consideration of quality, customer safety, responsible marketing, and fair trade are followed. Policies and commitment on community involvement, ethics, rights, suppliers, and economy construct So factor. Furthermore, culture and trust between employees, trade unions, training, and health & safety concerns construct the Wo factor. In the case of EC, EL, and CL have employee and client satisfaction as measuring variables, however, the SL factor has insider trading, and transparency as significant variables which impact their loyalty level. Thus, based on the CFA model, the causal relationship can be examined through SEM.

Table 6.

CFA Good of Fitness Summary

4.2 SEM Analysis

The hypothesized SEM model used in this paper can be denoted as <Figure 3> and <Table 7>.

Figure 3.

SEM Diagram

Table 7.

Result Summary of SEM Analysis

Since the p-values of the hypotheses are less than the confidence level of 95%, all the hypotheses are significant. However, H1, H7, and H9 are rejected since though the environmental factor has a positive influence on the EL (H1) and CL (H4), it can deteriorate the SL (H7). Though previous studies verify that the burden to satisfy environmental standards required in ESG practice can have a negative impact on EL (Piao et al., 2022), similar to McKinsey’s analysis which postulates the superior ESG proposition can enhance employees’ motivation by stimulating the sense of purpose (Henisz et al., 2019) the sense of contributing to the community derived from environmental practice in ESG initiative seems like improving employee motivation in this study. Furthermore, compared to the initial hypotheses that EN and GO have a positive impact on SL by increasing companies’ value and organizing systems for the shareholders, the SEM analysis indicates a negative relationship (Dahlberg & Wiklund, 2018). These results can be assumed by a similar context with H3. In Menner and Menningers’ study (2018), the negative causal relationship between governance and employee and shareholder satisfaction since intensified governance can strengthen managers’ interest in short-term performance but decrease interest in long-term investments such as human resources. Likewise, the burden to sustain transparency, fairness of the voting process, and other governance factors can cause a burden to the employees and shareholders. Moreover, this sort of burden leads to negligence of shareholder management the can deteriorate shareholder value in the long-term aspect. However, as it initially assumed in the previous study, SEM analysis denotes the positive impact of ESG on ESG to CL through satisfying customers’ needs for ethical consumption (Ismail et al., 2019).

4.3 K-mean clustering

Based on the SEM, using the elbow method on the K-mean clustering, which set the proper cluster number where the point considerably decreases through the scree plot as it is on the left side of <Figure 4>. Thus, this paper set 5 as the cluster number and has cluster as on the right side of <Figure 4>.

Figure 4.

Scree Plot & Result of K-means Clustering

Referring to the clustering analysis, the fundamental industries related to advanced resources, materials, parts, and equipment are categorized into cluster 1. However, it also includes infrastructural services such as transportation, retailing, etc. Cluster 2 comprises traditional resources such as fossil fuels, minerals, and general utilities. As the bank classifies into cluster 3 independently, it indicates distinctive ESG features compared to the others. Cluster 4 can be considered as formed with the manufacturing industry, and Cluster 5 seems like specified to the customer-oriented services (<Table 8>).

Table 8.

Cluster Categorization Details

As in <Table 9>, cluster 2 and 4 seems to have superior performance in overall ESG factors. This can be assumed since, due to the emission and environmental impact caused during production, manufacturing, and resource industries sincerely consider the ESG factors more than other industries. Also, cluster 1, which has adhesive relation with public and market infrastructure, indicates fair ESG performance except for the environmental field. However, it seems obvious that such industries as logistics, retail, academic services, and others are hard to improve emission reduction and product innovation significantly. This situation is also reflected in clusters 3 and 5, banking and advanced service sectors. Those industrial sectors consider the shareholders and fair governance structure significantly; however, since those industries have limited visible production, it is hard to satisfy environmental factors. However, despite the industrial specification, refer to tables 9 and 10, they denote inferior social consideration compared to the other clusters, and it seems like it resulted in low satisfaction of stakeholders.

Table 9.

Cluster summary (L1)

Table 10.

Cluster summary (L2)

To elaborate on the result, cluster 4 focusing on manufacturing-related industries such as chemicals, goods, and so on, denotes superior performance on EN having a higher value of 17.55 compared to the other clusters. As in <Table 3>, this cluster has tangible products which should conform to emission, product life cycle, and other various standards during their regular operation. Similar to EN, as SO is also adhesively related to its supply chain relationship and production impact on society, those clusters including banking and service denote inferior performance under the current framework. Compared to the EN and SO, GO has improved performance within clusters 3 and 5, however, compared to the other clusters, it also denotes a relatively low performance in GO. Likewise, those industries which do not perform practical production can be faced with comparative disadvantages when applying universe standards. Thus, in the context of practical equity, the further ESG framework and policies should consider different industrial attributes of the corporation.