An Analysis of the Relationship Between ESG Activities and the Financial Performance of Japanese Companies Toward Sustainable Development

Abstract

Demands for companies to comply with environmental, social, and governance (ESG) requirements are growing, and companies are also expected to play a role in promoting sustainable development. For companies to achieve sustainable growth while addressing ESG, it must be understood whether ESG activities promote improved corporate financial performance. We conducted a five-year panel data analysis of 635 Japanese firms from FY 2019 to FY 2023, using the PBR, PER, and ROE financial indicators as the dependent variables and CSR ratings in the human resource utilization (HR), environment (E), governance (G), and social (S) categories as the independent variables. The results revealed that, depending on the combination of ESG field and financial indicators, companies with advanced ESG initiatives had greater financial performance, with some cases showing a nonlinear relationship; differences in the results between manufacturing and nonmanufacturing industries were also observed. For companies to effectively advance ESG activities, it is important to clarify the objectives and results for each ESG category. For policymakers to consider measures to encourage companies’ ESG activities, it is also important to design finely tuned regulations and incentives according to the ESG category and industry characteristics.

1. Introduction

1.1. Background

The scope and content of ESG initiatives demanded by various stakeholders are rapidly evolving. In the global community, the role of the private sector in achieving the Sustainable Development Goals (SDGs) adopted by the United Nations [1], with 2030 as the target year, has been clearly stated. According to the Global Sustainable Investment Alliance [2], the global ESG investment market was worth USD 30.3 trillion by the end of 2022. As of 31 March 2024, the number of financial institutions that had signed the United Nations Principles for Responsible Investment to incorporate ESG issues into their investments totaled 5345, with assets under management totaling USD 128.4 trillion [3]. The concept of ESG investment has become widespread among many investors, and some studies have shown that it supports effective investment decisions. Teja and Liu [4] reported that high ESG risk scores are associated with low returns for investors, and that the strategy of investing in stocks with low ESG risk and shorting stocks with high ESG risk yields superior returns compared with the market portfolio.

In recent years, there has been a growing trend toward requiring the disclosure of the financial impact of nonfinancial ESG activities on corporate finances. In 2017, the recommendation of the Task Force on Climate-related Financial Disclosures [5] required companies to incorporate the financial impact of climate-related risks and opportunities into their financial reporting. In addition, interest in nature- and social-related financial impacts, such as the recommendations of the Task Force on Nature-related Financial Disclosures [6] in 2023 and the establishment of the Task Force on Inequality and Social-related Financial Disclosure [7] in 2024, is growing. In addition to stricter reporting requirements, the EU Corporate Sustainability Reporting Directive (CSRD) [8], which came into force in the EU in 2023, adopts the concept of double materiality, which states that “companies are obliged to report both the impact of their activities on people and the environment and the impact of sustainability aspects on the company”. Sato [9] noted that the EU CSRD may affect financial reporting systems, such as the linkage between financial and nonfinancial information. The EU CSRD also applies to companies outside the EU with significant sales; thus, many global companies are under pressure to disclose ESG information.

In addition, there is concern about ESG washing, where the content of ESG information disclosure is not accompanied by actual initiatives. Lagasio [10] revealed significant variation in ESG-washing practices across industries and geographical regions, in his analysis of sustainability reports from 749 listed companies with an ESG-washing severity index. Attig and Boshanna [11] suggested that the misalignment between “ESG talk” and “ESG walk” not only fails to serve shareholders’ best interests but also may undermine a firm’s social license to operate. Thus, companies are expected to make appropriate disclosures while meeting the increasingly rigorous demands of ESG activities. For companies to achieve sustainable growth while addressing ESG issues, it is important that ESG activities lead to increased corporate value.

In Japan, the interest in ESG is also growing, and the demand for companies to engage in ESG activities is increasing. At the end of 2022, the scale of ESG investment in Japan was JPY 494 trillion, an increase of 59% compared with that at the end of 2020 [2]. Japan’s Government Pension Investment Fund [12], which has one of the world’s largest assets under the management of JPY 260 trillion, selected its ESG indices in 2017 and managed JPY 17.8 trillion in ESG investments as of March 2024. In terms of ESG disclosure, the Sustainability Board of Japan issued its inaugural sustainability disclosure standards in March 2025, aiming to align with those of the International Sustainability Standards Board (ISSB) [13]. The Japanese government has also encouraged companies to engage in ESG activities, and the Ministry of Economy, Trada and Industry (METI) published guidance in 2022 to encourage constructive and substantive dialog between companies and investors with the aim of creating sustainable corporate value [14].

In response to these trends, major companies in Japan are actively engaged in ESG activities. In 2024, KPMG International [15] reported that 100% of Japan’s largest companies published sustainability reports. However, it is unclear whether Japanese companies, which proactively disclose their ESG information, have been able to link their ESG initiatives to concrete actions to increase their corporate value. The value creation process proposed by the IFRS Foundation [16] is recognized as a useful tool that clearly shows how the company pursues both sustainability and financial performance, and is mentioned by many companies in their integrated reports. According to a survey by PwC Japan [17], of the 200 listed companies in Japan, 147 (73%) disclosed their value creation stories. However, only 74 companies (37%) had solid stories describing all the elements from inputs to outcomes at the business level.

1.2. Objectives and Framework of This Study

For companies to achieve sustainable growth while addressing ESG issues, it is desirable that a company’s ESG activities lead to an increase in corporate value. In particular, if it becomes clear that ESG activities lead to improvements in not only nonfinancial performance but also financial performance, it will be easier for companies to communicate with stakeholders and make internal decisions effectively and smoothly.

Therefore, this study is designed to analyze the relationship between ESG activities and the financial performance of major companies in Japan by combining different ESG categories and multiple financial indicators. Our research questions focus on how different ESG activities are related to the financial performance of Japanese companies and how this relationship varies considering differences in the industry sector. We aim not only to contribute to the formulation of effective ESG management strategies for companies but also to help policymakers consider effective measures to encourage companies’ ESG activities. The reason for focusing on Japanese companies as the subject of research is that, as mentioned above, many companies are proactive in responding to requests and regulations from stakeholders related to ESG, but do not have sufficient strategic associations with their core businesses. We hope that the suggestions obtained from this research will lead to the consideration of measures to strengthen the sustainable competitiveness of Japanese industries. As an academic contribution, this analysis of Japanese companies serves as a useful case study because the accumulation of research on the relationship between ESG activities and financial performance is still in its early stages.

Figure 1 illustrates the framework of this study. We conducted a panel data analysis to determine the impact of nonfinancial performance due to ESG activities on financial performance. The financial indicators used as the dependent variables were the price-book value ratio (PBR), the price earnings ratio (PER) and return on equity (ROE), which are components of the PBR. For the nonfinancial indicators used as the independent variables, we focused on the corporate social responsibility (CSR) ratings in four categories, namely, “human resource utilization”, “environment”, “governance”, and “social”, in Toyo Keizai Inc.’s CSR Companies Complete Guide [18], as it includes an evaluation of the results of ESG activities. In addition, the number of employees, capital equipment ratio (CER) and equity multiplier (EM) were used as control variables. In our analysis, we focused on the possibility of a nonlinear relationship between the financial indicators and the CSR ratings; we compared a linear model with only a linear term in the CSR ratings and a quadratic model with both a linear term and a quadratic term in the CSR ratings in each of the four CSR rating categories. Furthermore, in addition to analyzing the entire sample of companies, we classified the sample companies into manufacturing and nonmanufacturing sectors to uncover potential differences in industry characteristics.

This paper is structured as follows. Section 2 presents the research hypotheses on the basis of a literature review of the relationship between ESG activities and financial performance. Section 3 describes the materials and methods of this study. Section 4 presents the financial data of the analyzed companies and the CSR rating data used to evaluate ESG activities. Section 5 discusses the results of the analysis, and Section 6 closes the paper with considerations for testing the hypotheses and the policy implications of this study.

2. Literature Review and Hypotheses

Theories and concepts regarding corporate efforts to build a sustainable environment and society and the enhancement of corporate value have been proposed and discussed from a variety of perspectives. Porter and Kramer [19] proposed the concept of creating a “shared value” that generates economic value in a way that also produces value for society. Polman and Winston [20] argued that the concept of “net positive” helps companies thrive by giving back more to the world than they take. Net positive companies profit by fixing the world’s problems, not creating them. One of the most frequently mentioned theories in the research on ESG and corporate value is stakeholder theory, which was initially developed by Freeman [21]. Stakeholder theory states that managers have a fiduciary responsibility to all stakeholders, not just shareholders. Kong et al. [22] noted that, from an ESG standpoint, ESG performance optimizes a firm’s operating and market performance since the firm will appear more responsible and be viewed favorably by stakeholders. In contrast to the external environmental model of competitive advantage, such as stakeholder theory, the resource-based view focuses on the internal strengths and weaknesses of companies. Under the resource-based view, ESG activities can improve the management team’s capabilities and the company’s potential to attract qualified employees. In addition, such activities can enhance corporate reputation and strengthen interactions with its stakeholders (Branco and Rodrigues [23]). It is important to consider a company’s sustainable competitiveness from both the perspective of its external environment and its internal resources. Freeman et al. [24] encouraged researchers to merge stakeholder theory and the resource-based view.

Studies analyzing the relationship between ESG activities and financial performance have increased significantly since around 2019 (Ikuta and Fujii [25]). Jung [26] reported that research on the relationship between a company’s ESG performance and its financial outcomes is growing richer, but empirical findings on the relationship between ESG performance and corporate value or financial performance have yet to converge to a consistent conclusion. Tanahashi [27] noted that there are many studies related to the impact of ESG disclosure information on corporate performance. On the basis of more than 1000 studies published between 2015 and 2020, Whelan et al. [28] reported that 58% of the studies found a positive relationship between ESG performance and financial performance. Through the analysis of 3332 listed organizations worldwide over an interval of 10 years (2011–2020), Chen et al. [29] showed that the influence of ESG ratings on corporate financial performance is significant for large-scale companies. In contrast, on the basis of a panel analysis of 5540 listed companies from 43 countries between 2018 and 2022, Gawęda [30] revealed a negative association between ESG performance and firm value.

At the country and regional levels, studies have also shown that ESG activities positively affect corporate management. Using data from 367 firms listed on the London Stock Exchange, Li et al. [31] reported a positive association between the level of ESG disclosure and firm value. Jung [26] discovered that ESG scores have a significant positive effect on the financial performance of Chinese A-share listed companies. Jo and Harjoto [32] showed that US firms’ CSR engagement positively influences their financial performance and that their CSR engagement with the community, environment, diversity, and employees plays a significantly positive role in enhancing corporate financial performance. Through a panel analysis of 1144 companies in the EU over an 11-year period (2013–2023), Ahmed and Khalaf [33] showed that the more efficient and effective the ESG performance is, the higher the company value is. In the case of Japan, on the basis of a multiple regression analysis of 533 Japanese manufacturing firms, Ikuta and Fujii [23] showed that firms with high financial performance, such as ROA and ROE, tended to be ahead in their SDG efforts. In addition, Kuang [34] analyzed the relationship between ESG performance and corporate financial performance, focusing on the Japanese ownership structure. However, few studies have been conducted on Japanese companies.

Based on these previous studies, this study first proposes the following hypothesis.

Hypothesis 1: 

Companies with advanced ESG activities have greater financial performance.

To elaborate on the relationship between ESG activities and financial performance, analyses have been conducted from various perspectives. A growing number of studies point to a nonlinear relationship between ESG activities and financial performance. In terms of quadratic curves, relationships can be classified as U-shaped or inverted U-shaped, and cases of both have been reported. Possible reasons for the appearance of a U-shaped curve include the high initial costs of activities or the time it takes to realize the benefits. In addition, proceeding with the activities will result in the deterioration of financial performance until a turning point is reached, but once the turning point is exceeded, financial performance will improve. Nollet et al. [35] provided evidence of a U-shaped relationship between the ESG disclosure score and financial indicators, such as ROA and ROC, among all firms listed in the S&P 500 stock market. On the basis of the analysis of 350 European listed companies, Bruna et al. [36] showed that there is a nonlinear positive relationship between ESG performance and financial performance, which is influenced by ESG content and company size. Zhang and Xiong [37] revealed a U-shaped nonlinear pattern in the impact of ESG performance on financial performance for 114 publicly traded agricultural and food firms in China. Bhandari et al. [38] revealed that the relationship between sustained competitive advantage and the ESG footprint for 1208 U.S. manufacturing firms is concave-shaped.

An inverted U-shape may occur for the following reasons: once a turning point is exceeded, additional costs increase, leading to poor financial performance; moreover, a situation of diminishing returns occurs where the benefits of investment are not commensurate with the business. In the analysis of the relationship between nonfinancial disclosure and corporate financial performance using data from 6631 samples from 74 countries and 11 industries, Xie et al. [39] reported that an inverted U-shaped relationship may exist between nonfinancial disclosure and efficiency. Meier et al. [40] analyzed 591 European companies and demonstrated a significant quadratic (inverted U-shaped) relationship between the corporate social performance of human resource management and corporate financial performance. Kuang [34] analyzed more than 4000 firm-year observations in Japan and showed that ESG performance is linked to corporate financial performance through an inverted U-shaped pattern, and that such relationships result mainly from the social pillar and the governance pillar when the dependent variables are firm value and profitability, respectively.

On the basis of these previous studies, we formulate the following hypothesis regarding the relationship between ESG activities and financial performance.

Hypothesis 2: 

There is a nonlinear relationship between ESG activities and financial performance.

Previous studies have also noted differences in the relationship between ESG categories and financial performance, and these studies have shown that the relationship between ESG categories and financial performance is not uniform. Ming et al. [41] examined the impact of ESG scores on the financial performance of Malaysia’s leading companies and reported a positive effect of the overall ESG score on ROA, but negative effects of individual E and S scores and adverse effects of corporate governance and debt ratios. Velte [42] reported that governance performance has the strongest impact on financial performance in comparison to environmental and social performance for 110 companies listed on the German Prime Standard. Lee et al. [43] showed that only the social score has a significant effect on firm performance measured by Tobin’s Q, whereas the environmental and governance pillars are insignificant for 59 listed firms under the FTSE4Good Bursa Malaysia. Aydogmus et al. [44] conducted further analysis on the impact of ESG performance on corporate value and profitability for 5000 listed companies worldwide, and reported that individual social and governance scores are positively and significantly related, whereas environmental scores are not significantly related to corporate value. Wang [45] examined 4660 A-listed Chinese corporations and reported that, among ESG factors, governance performance has the most significant influence on ROE. Dogan et al. [46] analyzed 5450 firms globally using a machine learning model and found that environmental and ESG controversies scores are important predictors of market value. Ikuta and Fujii [47] noted that there are differences between industries in the relationship between financial performance and the status of SDG efforts in the Japanese manufacturing sector. Brogi and Lagasio [48] analyzed the relationship between nonfinancial disclosure and ROA for U.S. firms, and reported that the disclosure of social information in the manufacturing industry, environmental and social information in the banking industry, and environmental, social, and corporate governance information in the insurance industry increased ROA more effectively than other types of disclosure.

Based on previous studies, we further developed the following hypotheses regarding the differences in the relationship between ESG activities and financial performance.

Hypothesis 3: 

The relationship between ESG activities and financial performance varies by ESG category.

Hypothesis 4: 

The relationship between ESG activities and financial performance varies by industry sector.

3. Materials and Methods

3.1. Materials

In this study, we conducted a panel data analysis using five years of firm data for Japanese companies to determine the impact of nonfinancial performance due to ESG activities on financial performance. As the financial indicator for the dependent variables, we employed the PBR, which investors value in evaluations of corporate value. In Japan, the PBR is an indicator that has attracted attention since the Tokyo Stock Exchange [49] requested that listed companies improve their PBR by 2023. Yanagi [50] used internal data from the Japanese pharmaceutical company Eisai to analyze the relationship between ESG key performance indicators (KPIs) and the PBR, and reported a significantly strong positive relationship with the PBR, especially for human capital KPIs. This also drew attention to the PBR as an ESG KPI in Japan. The PBR can be calculated by multiplying the PER by the ROE; thus, the PER and ROE are indicators that are positioned as components of the PBR. For nonfinancial independent variables, we focused on CSR ratings in four categories, namely “human resource utilization (HR), “environment (E),” “governance (G),” and “social (S)”, in Toyo Keizai Inc.’s CSR Companies Complete Guide as the results of ESG activities.

The analysis was conducted for companies listed in the 2019–2023 editions of the CSR Companies Complete Guide for five consecutive years, for which CSR ratings in all four categories were given. To support comparison with the financial data of the same period, the target companies were general business companies (excluding banks, securities, and insurance) listed in the Nikkei NEEDS-Financial QUEST Micro Comprehensive Database (hereafter, Nikkei NEEDS-FQDB) for the same period. Therefore, the target companies were listed companies and nonlisted companies that submitted securities reports, which can be regarded as major companies in Japan. In addition, while general business companies generate profits from sales of goods and services, financial companies generate profits by managing funds. As their business models are significantly different, it is difficult to compare their relationships with financial performance using the same criteria; thus, financial companies were excluded from the analysis of this study. Furthermore, the data published in the CSR Companies Complete Guide are based on survey responses from the previous year, so the actual CSR rating period is from 2018 to 2022. The 635 companies for which financial data were available for the five consecutive years from 2019 to 2023 were among the companies with listed CSR ratings, according to the assumption that the results would emerge in the year following the ESG activities, as shown in the studies by Velte [42] and Aydogmus et al. [44].

3.2. Methods

The CSR ratings in the CSR Companies Complete Guide are based on a five-point scale (AAA, AA, A, B, and C) for each of the following four categories: “HR” “E,” “G” and “S”. Toyo Keizai Inc.’s (Tokyo, Japan) financial and corporate evaluation team, which includes academic experts as advisors, has been conducting this rating since the 2007 edition. These ratings are relative ratings based on the scores and distribution of evaluation items in a questionnaire survey of companies listed in the CSR Companies Complete Guide (50 items for “HR,” 32 items for “E,” 41 items for “G,” and 31 items for “S”). For the purpose of analysis, the five-level CSR ratings were quantified as follows: AAA: 5 points, AA: 4 points, A: 3 points, B: 2 points, and C: 1 point. Financial data were winsorized by setting thresholds at the top and bottom 1% of the data for the PBR, PER, and ROE to reduce the impact of outliers.

In this panel data analysis, which focuses on the possibility of a nonlinear relationship between the financial indicators and CSR ratings, we compared a linear model with only a linear term for the CSR ratings and a quadratic model with both a linear term and a quadratic term for the CSR ratings in each of the four CSR categories. In cases where the nested model test could not confirm the effectiveness of adding a quadratic term, we adopted only the first-order model and did not conduct any analysis of nonlinear relationships. Furthermore, to identify differences in industry characteristics, we classified the sample companies into manufacturing (375 companies) and nonmanufacturing (260 companies) sectors for comparison, in addition to analyzing all target companies (hereafter, all companies). As we included data for five years, the number of observations is 635 companies × 5 years = 3175 for all companies, 375 companies × 5 years = 1875 for the manufacturing sector, and 260 companies × 5 years = 1300 for the nonmanufacturing sector. When conducting the panel data analysis, we adopted robust standard errors to eliminate problems of autocorrelation and heterogeneity and selected between a fixed effects model and a random effects model for each case on the basis of the results of the F test and Hausman test.

Table 1. Variables used in this analysis.

	Indicator		Abbreviation	Source
Dependent variable	Price-book value ratio		PBR	Nikkei NEEDS-Financial QUEST Micro Comprehensive Database
	Price earnings ratio		PER
	Return on equity		ROE
Independent variable	CSR rating	Human resource utilization	HR	Toyo Keizai Inc.’s CSR Companies Complete Guide
		Environment	E
		Corporate Governance	G
		Sociality	S
Control variable	Natural logarithmic number of employees		Ln (Emp)	Nikkei NEEDS-Financial QUEST Micro Comprehensive Database
	Capital equipment ratio		CER
	Equity multiplier		EM

Source: Created by the authors.

summarizes the variables used in this panel data analysis. In addition to the three financial indicators (PBR, PER, and ROE) as the independent variables and the four categories of CSR ratings (“HR,” “E,” “G,” and “S”) as the dependent variables, the number of employees, capital equipment ratio (CER), and equity multiplier (EM) were used as the control variables. The number of employees, which is the number at the end of the fiscal year, was used to account for the effect of firm size. Since the number of employees varies greatly across companies and is not normally distributed, the number of employees converted to the natural logarithm (Ln(Emp)) was used in the analysis. The CER indicates tangible fixed assets per employee and was used to account for the impact of the level of equipment and other facilities of a company. The EM indicates the ratio of debt to equity and was used to consider the impact of a company’s management strategy—either conservative management or growth-oriented. When the variance inflation factor (VIF) was also calculated for each case analyzed in this study, multicollinearity between variables was not detected in any case.

4. Data

Table 2. Data description of the analyzed companies.

		PBR	PER	ROE	Ln (Emp)	CER	EM	CSR Rating
								HR	E	G	S
All companies (Sample = 635)	Mean	1.68	21.75	7.87	6.96	56.24	2.23	3.80	3.78	3.79	3.77
	SD	1.38	51.52	8.76	1.48	318.44	1.42	0.87	0.91	0.89	0.91
Manufacturing sector (Sample = 375)	Mean	1.70	21.83	7.57	7.16	34.78	2.23	3.91	4.00	3.91	3.93
	SD	1.41	50.60	8.92	1.41	88.04	1.44	0.84	0.86	0.86	0.87
Nonmanufacturing sector (Sample = 260)	Mean	1.66	21.65	8.31	6.67	87.18	2.23	3.64	3.47	3.61	3.53
	SD	1.34	52.84	8.51	1.54	484.74	1.39	0.87	0.89	0.89	0.92
T-value of the t-test of the mean of manufacturing and nonmanufacturing sectors		0.71	0.10	−2.36 ***	9.06 ***	−3.85 ***	−0.03	8.98 ***	16.66 ***	9.42 ***	12.54 ***

Source: Toyo Keizai Inc. [16] and Nikkei NEEDS-FQDB. Note 1: Means and variances are calculated over 5 years of data for the target companies (financial indicators and control variables: FY 2019–2023; CSR ratings: 2018–2022). Note 2: The CER is shown in millions/person. Note 3: *** indicates significance at the 1%, 5% and 10% levels, respectively.

shows the means and standard deviations (SDs) of the financial indicators: the PBR, PER, and ROE; the control variables: Ln(Emp), CER (shown in millions/person) and EM; and the four categories of CSR ratings (“HR,” “E,” “G,” and “S”) for all companies and for the manufacturing and nonmanufacturing sectors. The means and SDs were calculated for each of the five years of data, from FY 2019 to FY 2023 for the financial indicators and control variables and from 2018 to 2022 for the CSR ratings.

With respect to the means of the financial indicators, there was no significant difference in the PBR, with values of 1.70 and 1.66 for the manufacturing and nonmanufacturing sectors, respectively, and 1.68 for all companies. Similarly, for the PER, there was almost no difference, with a value of 21.83 for the manufacturing sector and 21.65 for the nonmanufacturing sector, compared with 21.75 for all companies. However, there was a difference in the ROE between manufacturing and nonmanufacturing sectors—7.57 for the manufacturing sector and 8.31 for the nonmanufacturing sector—compared with 7.87 for all companies.

The means of the control variables showed that companies in the manufacturing sector tended to have more employees than those in the nonmanufacturing sector, as in the case of Ln(Emp), which was 7.16 for the manufacturing sector and 6.67 for the nonmanufacturing sector, compared with 6.96 for all companies. The CER tended to be higher in the nonmanufacturing sector than in the manufacturing sector, as the CER for all companies was JPY 56.24 million/person, compared with JPY 34.78 million/person for the manufacturing sector and JPY 87.18 million/person for the nonmanufacturing sector. With respect to the EM, the mean was 2.23 for all companies, whereas the mean for the manufacturing and nonmanufacturing sectors was also 2.23, indicating no difference.

In terms of the means of the CSR ratings by category, the means for all companies, in descending order, were 3.80 for “HR,” 3.79 for “G,” 3.78 for “E,” and 3.77 for “S,” but the difference between the largest and smallest means was only 0.03. The means of the CSR ratings of the manufacturing sector were higher than those of all companies in all four categories, with “E” at 4.00, “S” at 3.93, and “HR” and “G” at 3.91, and the difference between the maximum and minimum means was 0.09, which was greater than the difference for the overall sample. Conversely, the means of the CSR ratings of the nonmanufacturing sector were lower than those of all companies in all four categories, with 3.64 for “HR”, 3.61 for “G”, 3.53 for “S”, and 3.47 for “E”, showing a larger difference between the maximum and minimum means of 0.17.

Thus, a comparison of the mean CSR ratings between the manufacturing and nonmanufacturing sectors shows that the manufacturing sector had greater values than the nonmanufacturing sector in all the CSR categories; this is supported by the t-test results, which were significant at the 1% level. This finding indicates that, overall, ESG activities are more highly evaluated in the manufacturing sector than in the nonmanufacturing sector. This trend toward higher ratings of ESG activities may be because the manufacturing companies have more employees than nonmanufacturing companies, and thus have more personnel available to engage in ESG activities. Moreover, many manufacturing companies are embedded in global supply chains with strong ESG requirements.

In addition, a comparison of the difference in the mean CSR ratings between the manufacturing and nonmanufacturing sectors for each of the CSR categories revealed the following significant differences: 0.27 (3.91–3.64) for “HR”, 0.53 (4.00–3.47) for “E”, 0.30 (3.91–3.61) for “G”, and 0.40 (3.93–3.53) for “S”. In particular, the difference in the means of “E” was large. This suggests that the manufacturing sector focuses on environmental initiatives because of the large environmental burden of the sector in its value chain, which may have contributed to the difference from the mean value of the nonmanufacturing sector. On the other hand, the differences in CSR ratings for “HR” and “G” were not as large as those for “E” because differences in efforts between the manufacturing and nonmanufacturing sectors were less apparent.

5. Results and Discussion

5.1. Results

5.1.1. All Companies

Table 3. Results of the panel data analysis for all companies.

CSR Rating		HR			E			G			S
Financial Indicator		PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Linear	Model	FE	FE	FE	FE	RE	RE	FE	FE	RE	FE	FE	RE
	COEF	0.012 (0.261)	−0.854 (−0.308)	−0.309 (−0.835)	0.053 (1.179)	4.232 *** (3.051)	0.345 (1.358)	0.062 (1.470)	0.237 (0.085)	0.675 *** (3.015)	0.089 * (1.713)	−0.448 (−0.638)	0.715 *** (2.747)
	Adjusted R2	0.811	0.292	0.449	0.811	0.008	0.015	0.812	0.292	0.020	0.812	0.292	0.018
Quadratic	Model	FE	-	FE	FE	-	FE	FE	-	RE	FE	-	RE
	COEF_L	0.147 (0.762)	-	−1.573 (−0.986)	0.168 (0.761)	-	1.716 (0.959)	−0.098 (−0.426)	-	−3.386 ** (−2.193)	0.206 (1.066)	-	−1.900 (−1.393)
	COEF_Q	−0.018 (−0.678)	-	0.168 (0.744)	−0.015 (−0.510)	-	−0.211 (−0.869)	0.021 (0.673)	-	0.535 ** (2.566)	−0.016 (−0.561)	-	0.352 * (1.868)
	Adjusted R2	0.811	-	0.449	0.811	-	0.449	0.812	-	0.022	0.812	-	0.020
	Turning point	N.S.	-	N.S.	N.S.	-	N.S.	N.S.	-	3.16	N.S.	-	N.S.
	Mean									3.79

Note 1: The number of observations is 3175 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF_L represents a linear term of the coefficient, and COEF_Q represents a quadratic term of the coefficient. Note 4: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

summarizes the relationship between the independent-variable CSR ratings and the dependent-variable financial indicators on the basis of the results of panel data analysis of all 635 sample companies. In addition to 12 cases for the linear model, analysis was conducted for 8 cases for the quadratic model, in which the addition of a quadratic term was found to be effective on the basis of the results of the nested model test. The results from all the analyses, including those for the control variables, are shown in

Table A1. Results of the panel data analysis (linear model) for all companies.

CSR Rating	HR			E			G			S
Financial Indicator	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Model	FE	FE	FE	FE	RE	RE	FE	FE	RE	FE	FE	RE
COEF	0.012 (0.261)	−0.854 (−0.308)	−0.309 (−0.835)	0.053 (1.179)	4.232 *** (3.051)	0.345 (1.358)	0.062 (1.470)	0.237 (0.085)	0.675 *** (3.015)	0.089 * (1.713)	−0.448 (−0.638)	0.715 *** (2.747)
Ln(Emp)	−0.188 *** (−3.408)	0.936 (0.118)	1.586 ** (2.280)	−0.189 *** (−3.444)	−3.395 *** (−3.377)	0.642 *** (3.807)	−0.188 *** (−3.446)	0.948 (0.119)	0.581 *** (3.666)	−0.186 *** (−3.366)	0.942 (0.119)	0.522 *** (2.900)
CER	−0.001 * (−1.895)	−0.047 (−1.499)	0.009 * (1.874)	−0.001 * (−1.949)	−0.004 *** (−2.374)	0.001 *** (3.171)	−0.001 * (−1.922)	−0.047 (−1.495)	0.001 *** (2.987)	−0.001 * (−1.922)	−0.047 (−1.495)	0.001 *** (2.956)
EM	0.035 ** (2.204)	0.486 (0.687)	0.145 (1.137)	0.035 ** (2.219)	−0.322 (−0.635)	0.097 (0.841)	0.034 ** (2.187)	0.483 (0.686)	0.094 (0.818)	0.035 ** (2.207)	0.485 (0.686)	0.096 (0.842)
Constant	2.943 *** (6.716)	19.203 (0.341)	−3.111 (−0.606)	2.842 *** (6.712)	34.553 *** (5.591)	2.164 * (1.913)	2.810 *** (6.784)	16.071 (0.284)	1.680 (1.471)	2.722 *** (6.463)	18.011 (0.313)	1.990 * (1.801)
Adjusted R2	0.811	0.292	0.449	0.811	0.008	0.015	0.812	0.292	0.020	0.812	0.292	0.018

Note 1: The number of observations is 3175 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF represents the coefficient. Note 4: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

and

Table A2. Results of the panel data analysis (quadratic model) for all companies.

CSR Rating	HR			E			G			S
Financial Indicator	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Model	FE	-	FE	FE	-	FE	FE	-	RE	FE	-	RE
COEF_L	0.111 (0.784)	-	−1.234 (−1.067)	0.138 (0.845)	-	1.293 (0.982)	−0.056 (−0.332)	-	−2.316 ** (−2.045)	0.174 (1.252)	-	−1.197 (−1.201)
COEF_Q	−0.018 (−0.678)	-	0.168 (0.744)	−0.015 (−0.510)	-	−0.211 (−0.869)	0.021 (0.673)	-	0.535 ** (2.566)	−0.016 (−0.561)	-	0.352 * (1.868)
Ln(Emp)	−0.188 *** (−3.411)	-	1.587 ** (2.278)	−0.190 *** (−3.448)	-	1.584 ** (2.290)	−0.189 *** (−3.441)	-	0.590 *** (3.715)	−0.187 *** (−3.378)	-	0.516 *** (2.856)
CER	−0.001 * (−1.887)	-	0.009 * (1.865)	−0.001 * (−1.958)	-	0.009 * (1.852)	−0.001 * (−1.911)	-	0.001 *** (2.838)	−0.001 * (−1.924)	-	0.001 *** (2.814)
EM	0.035 ** (2.199)	-	0.145 (1.139)	0.035 ** (2.222)	-	0.146 (1.140)	0.034 ** (2.190)	-	0.094 (0.811)	0.035 ** (2.229)	-	0.093 (0.821)
Constant	2.820 *** (5.965)	-	−1.956 (−0.372)	2.738 *** (6.351)	-	−5.751 (−1.107)	2.965 *** (6.313)	-	5.378 *** (3.316)	2.621 *** (6.154)	-	4.344 *** (2.706)
Adjusted R2	0.811	-	0.449	0.811	-	0.449	0.812	-	0.022	0.812	-	0.020
Turning point	N.S.	-	N.S.	N.S.	-	N.S.	N.S.	-	2.16	N.S.	-	N.S.
Mean									2.79

Note 1: The number of observations is 3175 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF_L represents a linear term of the coefficient, and COEF_Q represents a quadratic term of the coefficient. Note 4: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

in Appendix A.

Significant results were observed for the relationships between CSR ratings and the financial indicators: “E” and the PER, “G” and the ROE, and “S” for the PBR and ROE. No significant relationship was found between “HR” and any financial indicator. A significant positive relationship was observed between “E” and PER in the linear model, which means that companies with a high evaluation of “E” tend to have a high PER. For example, a one-unit increase in the “E” rating would be equivalent to an increase of 4.232 points in the PER.

A positive and U-shaped significant relationship was observed for “G” and the ROE in the linear model and the quadratic model, respectively. Since the adjusted R2 of the quadratic model was slightly larger than that of the linear model, if we adopt the interpretation of the quadratic model, companies with a high rating for “G” will have a low ROE until the turning point of the quadratic curve is exceeded, but once the rating exceeds the turning point, companies with a high rating for “G” will have a high ROE. The turning point (3.16) is located between A (score = 3) and AA (score = 4) in the actual CSR ratings, which suggests that “G” efforts lead to a decline in ROE until the rating is AA, and that companies with a rating of AA or higher will experience an increase in the ROE. Looking at the relationship between the mean and the turning point, the mean (3.79) is above the turning point, suggesting that there are many companies for whom the rating of “G” coincides with an improvement in the ROE.

For “S,” a significant positive relationship was observed with the PBR and ROE in the linear model, which means that companies with a high rating for “S” also tend to have a high PBR and ROE. A one-unit increase in the “S” rating is equivalent to an increase of 0.089 points in the PBR and 0.715 points in the ROE.

5.1.2. Manufacturing Sector

Table 4. Results of the panel data analysis for the manufacturing sector.

CSR Rating		HR			E			G			S
Financial Indicator		PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Linear	Model	FE	RE	RE	FE	FE	RE	FE	FE	RE	FE	FE	RE
	COEF	0.026 (0.491)	6.772 *** (3.580)	0.593 * (1.843)	0.053 (0.848)	4.843 (1.258)	0.448 (1.383)	0.065 (1.276)	−1.752 (−0.474)	0.554 * (1.915)	0.118 * (1.809)	−4.010 (−0.966)	1.097 *** (3.101)
	Adjusted R2	0.797	0.012	0.042	0.797	0.283	0.039	0.797	0.283	0.041	0.797	0.283	0.045
Quadratic	Model	-	-	-	FE	-	-	-	-	-	FE	-	-
	COEF_L	-	-	-	0.232 (0.476)	-	-	-	-	-	0.870 *** (2.831)	-	-
	COEF_Q	-	-	-	−0.022 (−0.369)	-	-	-	-	-	−0.096 ** (−2.463)	-	-
	Adjusted R2	-	-	-	0.797	-	-	-	-	-	0.798	-	-
	Turning point	-	-	-	N.S.	-	-	-	-	-	4.53	-	-
	Mean										3.93

Note 1: The number of observations is 1875 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF_L represents a linear term of the coefficient, and COEF_Q represents a quadratic term of the coefficient. Note 4: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

summarizes the relationship between the independent-variable CSR ratings and the dependent-variable financial indicators on the basis of the results of a panel data analysis of the 375 manufacturing companies. In addition to 12 cases for the linear model, analysis was conducted for 2 cases for the quadratic model, in which the addition of a quadratic term was found to be effective on the basis of the results of the nested model test. All analysis results, including control variables, are shown in

Table A3. Results of the panel data analysis (linear model) for the manufacturing sector.

CSR Rating	HR			E			G			S
Financial Indicator	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Model	FE	RE	RE	FE	FE	RE	FE	FE	RE	FE	FE	RE
COEF	0.026 (0.491)	6.772 *** (3.580)	0.593 * (1.843)	0.053 (0.848)	4.843 (1.258)	0.448 (1.383)	0.065 (1.276)	−1.752 (−0.474)	0.554 * (1.915)	0.118 * (1.809)	−4.010 (−0.966)	1.097 *** (3.101)
Ln(Emp)	−0.218 ** (−2.454)	−4.003 *** (−2.647)	1.096 *** (5.089)	−0.219 ** (−2.452)	5.079 (0.400)	1.130 *** (5.151)	−0.217 ** (−2.443)	5.082 (0.404)	1.109 *** (5.261)	−0.220 ** (−2.446)	5.172 (0.410)	0.896 *** (3.101)
CER	−0.002 ** (−2.284)	−0.040 * (−1.723)	0.009 *** (3.206)	−0.002 ** (−2.285)	−0.065 (−1.092)	0.010 *** (3.309)	−0.002 ** (−2.258)	−0.067 (−1.139)	0.010 *** (3.265)	−0.002 ** (−2.312)	−0.067 (−1.130)	0.009 *** (2.950)
EM	0.043 ** (2.191)	−0.014 (−0.024)	0.075 (0.531)	0.044 ** (2.202)	0.236 (0.317)	0.071 (0.504)	0.043 ** (2.182)	0.219 (0.293)	0.072 (0.511)	0.044 ** (2.190)	0.203 (0.274)	0.072 (0.528)
Constant	3.170 *** (4.425)	32.161 *** (3.224)	−2.492 * (−1.670)	3.090 *** (4.310)	−27.307 ((−0.296)	−2.361 (−1.584)	3.049 *** (4.382)	−7.602 (−0.083)	−2.480 * ((−1.668)	2.912 *** (4.362)	−1.549 ((−0.017)	−2.524 * (−1.740)
Adjusted R2	0.797	0.012	0.042	0.797	0.283	0.039	0.797	0.283	0.041	0.797	0.283	0.045

Note 1: The number of observations is 1875 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF represents the coefficient. Note 4: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

and

Table A4. Results of the panel data analysis (quadratic model) for the manufacturing sector.

CSR Rating	HR			E			G			S
Financial Indicator	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Model	-	-	-	FE	-	-	-	-	-	FE	-	-
COEF_L	-	-	-	0.187 (0.510)	-	-	-	-	-	0.678 *** (2.928)	-	-
COEF_Q	-	-	-	−0.022 (−0.369)	-	-	-	-	-	−0.096 ** (−2.463)	-	-
Ln(Emp)	-	-	-	−0.218 ** (−2.460)	-	-	-	-	-	−0.219 ** (−2.436)	-	-
CER	-	-	-	−0.002 ** (−2.296)	-	-	-	-	-	−0.002 ** (−2.319)	-	-
EM	-	-	-	0.044 ** (2.219)	-	-	-	-	-	0.045 ** (2.259)	-	-
Constant	-	-	-	2.900 *** (3.606)	-	-	-	-	-	2.153 *** (3.049)	-	-
Adjusted R2	-	-	-	0.797	-	-	-	-	-	0.798	-	-
Turning point	-	-	-	N.S.	-	-	-	-	-	3.53	-	-
Mean										2.93

Note 1: The number of observations is 1875 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF_L represents a linear term of the coefficient, and COEF_Q represents a quadratic term of the coefficient. Note 4: *** and ** indicate significance at the 1% and 5% levels, respectively.

in Appendix A.

Significant results were found for the following relationships between the CSR ratings and financial indicators: “HR” for the PER and ROE, “G” and the ROE, and “S” for the PBR and ROE. No significant relationship was observed between “E” and any of the financial indicators. For “HR,” a significant positive relationship was found with the PER and ROE in the linear model, which means that companies with a high rating for “HR” tend to have a high PER and ROE. A one-unit increase in the “HR” rating is equivalent to an increase of 6.772 points in the PER and 0.593 points in the ROE. For “G,” a significant positive relationship was found with the ROE in the linear model, which means that companies with a high rating for “G” tend to have a high PER. A one-unit increase in the “G” rating is equivalent to an increase of 0.554 points in the ROE. For “S,” a significant positive relationship was observed with the PBR in the linear model, and an inverted U-shaped significant relationship was found in the quadratic model. Since the adjusted R2 of the quadratic model is slightly larger than that of the linear model, if the interpretation of the quadratic model is adopted, companies with a high rating for “S” will have a high PBR until the turning point of the quadratic curve is exceeded, but once the rating exceeds the turning point, the PBR will decrease. This turning point (4.53) is between AA (score = 4) and AAA (score = 5) in the actual CSR ratings, suggesting that efforts to increase the “S” rating to AAA would not contribute to improving the PBR but would instead lead to a decline in the PBR. Looking at the relationship between the mean and the turning point, the mean (3.93) is below the turning point, suggesting that there are many companies with ratings of AA or below where the “S” rating and PBR improvement coincide. In addition, a significant positive relationship was found between “S” and the ROE in the linear model, which means that companies with high ratings for “S” also have a high ROE. A one-unit increase in the “S” rating is equivalent to an increase of 1.097 points in the ROE.

5.1.3. Nonmanufacturing Sector

Table 5. Results of the panel data analysis for the nonmanufacturing sector.

CSR Rating		HR			E			G			S
Financial Indicator		PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Linear	Model	FE	FE	RE	RE	RE	RE	FE	RE	RE	FE	RE	RE
	COEF	−0.004 (−0.053)	−8.638 * (−1.907)	0.228 (0.528)	0.067 (1.282)	4.170 * (1.937)	0.402 (0.953)	0.050 (0.714)	7.027 *** (2.968)	0.802 (2.233)	0.046 (0.571)	9.383 *** (3.665)	0.295 (0.790)
	Adjusted R2	0.836	0.308	0.004	−0.003	0.009	0.005	0.836	0.018	0.010	0.836	0.026	−0.002
Quadratic	Model	-	-	RE	RE	-	RE	RE	-	RE	-	-	-
	COEF_L	-	-	−2.918 * (−1.904)	−0.040 (−0.184)	-	−3.934 * (−1.711)	−0.610 * (−1.799)	-	−5.699 ** (−2.550)	-	-	-
	COEF_Q	-	-	0.429 * (1.895)	0.015 (0.465)	-	0.602 * (1.870)	0.097 ** (1.973)	-	0.873 *** (2.807)	-	-	-
	Adjusted R2	-	-	0.007	−0.002	-	0.012	0.020	-	0.017	-	-	-
	Turning point	-	-	3.40	N.S.	-	3.27	3.14	-	3.26	-	-	-
	Mean			3.64			3.47	3.61		3.61

Note 1: The number of observations is 1300 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

summarizes the relationship between the independent-variable CSR ratings and the dependent-variable financial indicators on the basis of the results of the panel data analysis of the 260 nonmanufacturing companies. In addition to 12 cases for the linear model, further analysis was conducted for 5 cases for the quadratic model, in which the addition of a quadratic term was found to be effective on the basis of the results of the nested model test. All analysis results, including the control variables, are shown in

Table A5. Results of the panel data analysis (linear model) for the nonmanufacturing sector.

CSR Rating	HR			E			G			S
Financial Indicator	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Model	FE	FE	RE	RE	RE	RE	FE	RE	RE	FE	RE	RE
COEF	−0.004 (−0.053)	−8.638 * (−1.907)	0.228 (0.528)	0.067 (1.282)	4.170 * (1.937)	0.402 (0.953)	0.050 (0.714)	7.027 *** (2.968)	0.802 (2.233)	0.046 (0.571)	9.383 *** (3.665)	0.295 (0.790)
Ln(Emp)	−0.205 *** (−2.665)	−4.958 (−0.539)	0.365 (1.491)	−0.101 * (−1.932)	−3.975 *** (−3.461)	0.309 (1.230)	−0.206 *** (−2.695)	−4.215 *** (−3.792)	0.263 (1.103)	−0.203 ** (−2.589)	−5.528 *** (3.665)	0.336 (1.251)
CER	−0.000 (−0.962)	−0.030 (−1.101)	0.001 ** (2.352)	−0.000 (−1.463)	−0.003 *** (−3.165)	0.001 ** (2.172)	−0.000 (−1.007)	−0.004 *** (−3.373)	0.000 * (1.663)	−0.000 (−0.975)	−0.005 *** (−3.824)	0.001 ** (2.143)
EM	0.022 (0.841)	0.893 (0.623)	0.099 (0.488)	0.024 (0.912)	−0.533 (−0.574)	0.104 (0.517)	0.021 (0.825)	−0.631 (0.712)	0.096 (0.484)	0.022 (0.841)	−0.647 (0.718)	0.101 (0.504)
Constant	3.027 *** (5.398)	78.141 (1.224)	4.999 *** (2.807)	2.125 *** (5.711)	39.298 *** (5.082)	4.971 *** (2.836)	2.892 *** (5.453)	33.144 (4.096)	4.208 ** (2.333)	2.879 *** (4.890)	36.650 *** (4.891)	5.047 *** (2.954)
Adjusted R2	0.836	0.308	0.004	−0.003	0.009	0.005	0.836	0.018	0.010	0.836	0.026	−0.002

Note 1: The number of observations is 1300 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF represents the coefficient. Note 4: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

and

Table A6. Results of the panel data analysis (quadratic model) for the nonmanufacturing sector.

CSR rating	HR			E			G			S
Financial Indicator	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Model	-	-	RE	RE	-	RE	RE	-	RE	-	-	-
COEF_L	-	-	−2.059 * (−1.882)	-0.010 (-0.062)	-	−2.731 (−1.635)	−0.416 * (−1.714)	-	−3.953 ** (−2.436)	-	-	-
COEF_Q	-	-	0.429 * (1.895)	0.015 (0.465)	-	0.602 * (1.870)	0.097 ** (1.973)	-	0.873 *** (2.807)	-	-	-
Ln(Emp)	-	-	0.347 (1.424)	−0.100 * (−1.923)	-	0.323 (1.293)	−0.102 ** (−1.994)	-	0.284 (1.186)	-	-	-
CER	-	-	0.001 ** (1.965)	−0.000 (−1.500)	-	0.001 * (1.821)	−0.000 (−1.643)	-	0.000 (1.494)	-	-	-
EM	-	-	0.090 (1.199)	0.024 (0.916)	-	0.104 (0.514)	0.023 (0.893)	-	0.082 (0.410)	-	-	-
Constant	-	-	7.866 *** (3.784)	2.206 *** (5.560)	-	8.481 (3.392)	2.652 *** (5.567)	-	9.870 *** (4.224)	-	-	-
Adjusted R2	-	-	0.007	−0.002	-	0.012	0.020	-	0.017	-	-	-
Turning point	-	-	2.40	N.S.	-	N.S.	2.14	-	2.26	-	-	-
Mean			2.64				2.61		2.61

Note 1: The number of observations is 1300 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: COEF_L represents a linear term of the coefficient, and COEF_Q represents a quadratic term of the coefficient. Note 4: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

in Appendix A.

Significant results were observed for the following relationships between the CSR ratings and financial indicators: “HR” for the PER and ROE, “E” and the PER, “G” for the PBR and ROE, and “S” and the PER. A significant negative relationship was observed for “HR” with the PER in the linear model, which means that companies with a high rating for “HR” tend to have a low PER. A one-unit increase in the “HR” rating is equivalent to a decrease of 8.638 points in the PER. Only the U-shaped quadratic model revealed a significant relationship between “HR” and the ROE. Since the turning point of the quadratic curve is 3.40, this suggests that “HR” efforts lead to a decrease in the ROE until the rating is AA, and that companies with a rating of AA or higher exhibit an increase in the ROE. Since the mean (3.64) is above the turning point, this result suggests that there are many companies where the high rating for “HR” and the improvement of ROE is consistent.

A significant positive relationship was found for “E” with the PER in the linear model, which means that companies with a high rating for “E” tend to have a high PER. A one-unit increase in the “E” rating is equivalent to an increase of 4.170 points in the PER. Only the U-shaped quadratic model revealed a significant relationship between “E” and the ROE. Since the turning point of the quadratic curve is 3.27, this suggests that “E” efforts lead to a decrease in the ROE until the rating is AA and that companies with a rating of AA or higher exhibit an increase in the ROE. Since the mean (3.47) is above the turning point, the result suggests that there are many companies where the high rating for “HR” and the improvement of ROE is consistent.

Only the U-shaped quadratic model revealed a significant relationship with the PBR and ROE for “G.” The turning point for the PBR was 3.14, and that for ROE was 3.26, suggesting that “G” efforts lead to a decrease in the PBR and ROE until the rating is AA and lead to improvements in the PBR and ROE for companies rated AA or higher. As the mean (3.61) is above the turning point, this result suggests that there are many companies for whom the evaluation of “G” coincides with improvements in the ROE. A significant positive relationship was observed between “G” and the PER in the linear model, which means that companies with a high rating for “G” tend to have a high PER as well. A one-unit increase in the “G” rating is equivalent to an increase of 7.027 points in the PER. A significant positive relationship was also noted between “S” and the PER in the first-power model, which means that companies with a high rating for “S” tend to have a high PER as well. A one-unit increase in the “S” rating is equivalent to an increase of 9.383 points in the PER.

5.1.4. Summary

Figure 2 summarizes the hypotheses and the significant relationships between financial indicators and CSR ratings on the basis of the results of the panel data analysis described above. This analysis was conducted for all companies, the manufacturing sector, and the nonmanufacturing sector, combining four types of CSR rating categories with three types of financial indicators, resulting in a total of 36 cases.

A significant relationship was observed in 17 of these 36 cases. Of these, 10 cases showed a significant positive relationship only in the linear model, and 2 cases showed a positive relationship in both the linear model and the quadratic model. On the other hand, one case showed a significant negative relationship in the linear model, and three cases showed a significant relationship only in the quadratic model. As a significant positive relationship was observed between the CSR ratings and financial indicators in 12 cases, or one-third of the total cases, Hypothesis 1, “Companies with advanced ESG activities have greater financial performance,” is supported for some combinations. For the 15 cases in which a quadratic model was adopted after the addition of the quadratic terms was checked, a significant relationship was shown in 6 cases. Among these, in 4 cases, the linear model was not significant, and only the quadratic model was applicable. This supports Hypothesis 2, “There is a nonlinear relationship between ESG activities and financial performance,” suggesting that the adoption of the quadratic model was effective. Furthermore, the presence or absence of a relationship with the financial indicators varies depending on the CSR rating category, and the relationship between the CSR ratings and financial indicators differs across all companies and between manufacturing and nonmanufacturing sectors. These findings support both Hypothesis 3, “The relationship between ESG activities and financial performance varies by ESG category”, and Hypothesis 4, “The relationship between ESG activities and financial performance varies by industry sector”.

5.2. Discussion

5.2.1. Relationship Between ESG Activities and Financial Performance

On the basis of the abovementioned results, we look more closely at the relationship between ESG activities and financial performance. First, we consider the relationship between the financial indicators and the PBR and CSR ratings. The PBR is the value of a stock price divided by net assets per share, and if the PBR exceeds 1, it means that the company’s value is evaluated higher than its net assets. Yanagi [50] regarded the portion of the PBR that exceeds 1 as market added value, which reflects nonfinancial value. In the analysis of all companies, the only CSR category that showed a significant positive relationship with the PBR was “S”. In other words, social activities are recognized and evaluated as market added value. Since the items that make up Toyo Keizai’s CSR rating for “S” include communication with stakeholders and information disclosure, community activities, sustainable procurement, and compliance with the SDGs, these activities contribute to increasing corporate value. In contrast, activities in the categories of “HR”, “E”, and “G”, which did not show a significant correlation with the PBR, do not lead to increased corporate value.

In the case of categorizing companies into manufacturing and nonmanufacturing sectors, the relationship between the PBR and “S” was significant for the manufacturing sector, as for all companies, but no significant relationship was shown for the nonmanufacturing sector. This difference may be due to the difference in the impact of sustainable procurement, one of the rating items for “S”. Compared with the nonmanufacturing sector, manufacturing industries have built global supply chains for raw materials, parts, etc., and therefore, it is thought that the implementation of procurement that takes sustainability into consideration is more likely to be evaluated from the perspective of risk management, etc. In addition, the inverted U-shaped model is also significant in the manufacturing sector, and for AAA-rated companies that exceed the turning point, “S” activities may have a negative relationship with the PBR. This suggests that there are concerns that sustainability procurement in the manufacturing sector will lead to diminishing returns.

Next, let us consider the relationship between the PER and CSR ratings. The PER is the stock price divided by net income per share and indicates expectations for the future prospects and growth of a company. In the analysis of all companies, the only CSR category that showed a significant positive relationship with the PER was “E.” The relationship between a company’s environmental activities and corporate value/competitiveness has been widely discussed since Porter [51] proposed in 1991 that environmental regulations induce innovation and strengthen corporate competitiveness [52]. Recent studies have reported that the relationships are positive [46], negative [41], and neutral [42,43,44], and the results are not consistent. However, the results of this study support the view that a company’s environmental activities lead to its growth potential. In contrast, activities in the “HR,” “G,” and “S” categories are regarded as not leading to growth potential.

A comparison of the manufacturing and nonmanufacturing sectors revealed that the relationship between the PER and “E” was not significant for the manufacturing sector but was significant and positive for the nonmanufacturing sector for all companies. Aydogmus et al. [44] noted that environmental initiatives require costs such as capital investment, so a significant relationship with corporate value (Tobin’s q) cannot be seen. Manufacturing companies with production facilities tend to incur more costs for environmental measures than nonmanufacturing companies; thus, it is possible that no significant trend was found in the manufacturing sector in this study. Furthermore, while no significant relationship was found between the PER and “HR” for all companies, a positive relationship was observed in the manufacturing sector and a negative relationship was observed in the nonmanufacturing sector, which is thought to offset the results. On the basis of a literature survey from 1990 to 2005, Becker and Huselid [53] noted that human resource management efforts are unlikely to be directly related to financial performance. This suggests the possibility of conducting a more in-depth study by classifying industries in more detail. The reasons why opposite results were found for the relationship between the PER and “HR” in the manufacturing and nonmanufacturing sectors will be a topic of future research. Additionally, in the nonmanufacturing sector, “G” and “S” also had positive relationships with the PER, and, as a result, a significant relationship was found between all the CSR rating categories and the PER. As shown in Table 2 above, the mean CSR rating for the nonmanufacturing sector is lower than that for the manufacturing sector, and it can be said that ESG activities are lagging behind; thus, it is possible that expectations for future growth due to ESG activities are more likely to emerge in nonmanufacturing industries. However, the details will be investigated as a research topic in future studies.

Finally, let us consider the relationship between the ROE and CSR ratings. The ROE is the value obtained by dividing profits by equity capital and indicates management efficiency. In the analysis of all companies, the CSR categories that showed a significant positive relationship with the ROE were “G” and “S.” This means that companies that are evaluated for their “corporate governance” and “sociality” activities also show good management efficiency. A significant positive relationship between the ROE and corporate governance was also reported in a study by Wang [45] targeting Chinese companies. In a study targeting German companies, Velte [42] reported a significant relationship between corporate governance and ROA, which indicates management efficiency, including total assets. Moreover, activities for “HR” and “E,” which are not significantly related, are costly and take time for the effects of investment to appear, so they are not thought to lead to management efficiency [44,53]. Regarding the relationship between the ROE and “G”, as well as a positive significant relationship in the linear model, a U-shaped significant relationship was also shown in the quadratic model. This may reflect the fact that a significant positive relationship was observed in the linear model in the manufacturing sector, and a U-shaped significant relationship was found in the quadratic model in the nonmanufacturing sector. The U-shaped relationship observed in the nonmanufacturing sector suggests that for companies with a low rating for “G”, the costs required for “G” lead to a deterioration in management efficiency until the turning point is reached.

Regarding the relationship between the ROE and “S”, a significant positive result was shown in the linear model in the manufacturing sector, as with all companies, but no significant relationship was found in the nonmanufacturing sector. This is the same result as the relationship between the PBR and “S” mentioned above, and it is possible that sustainable procurement, which is thought to be a major difference between manufacturing and nonmanufacturing companies in the evaluation of “S”, influences it. In other words, it is suggested that the implementation of sustainable procurement leads to management efficiency. In addition, regarding the relationship between the ROE and “E”, a U-shaped significant relationship was found in the quadratic model in the nonmanufacturing sector. This suggests that for nonmanufacturing companies with a low rating for “G”, the costs required for “G” lead to a deterioration in management efficiency until the turning point is reached. Furthermore, regarding the relationship between the ROE and “HR”, which did not show significant results for all companies, a positive relationship was shown in the linear model in the manufacturing industry, and a U-shaped relationship was shown in the quadratic model in the nonmanufacturing sector; this will be a future research topic, including the fact that the effects were offset overall.

5.2.2. Research Limitations and Further Topics

With respect to the limitations and challenges of this research, let us first summarize the indicators used in the analysis. The CSR ratings of Toyo Keizai, Inc., which were used as independent variables, were useful in terms of ensuring the comprehensiveness of the CSR information of Japanese companies. However, because the five-level ratings were converted into discrete numerical values, i.e., 5, 4, 3, 2, and 1, the resolution of the analysis may have decreased. In this study, we limited the financial performance indicators to the PBR and its components, the PER and ROE, which are emphasized by the Japanese stock market. However, in order to delve deeper into the relationship between each of the financial indicators clarified here and ESG activities and to analyze and discuss the relationship between corporate value and ESG from multiple angles, we would also like to consider using indicators such as the ROA [29,35,37,39,40,41,42,44,48] and Tobin’s q [30,31,32,39,42,43,44,46], which have been used as indicators of corporate value in previous research. In addition, because this study focuses on Japanese companies, when the analysis results are applied to other countries or regions, regional characteristics must be carefully considered. Doğan et al. [46] suggested that the importance of corporate sustainability performance factors varies from region to region. To utilize the results of this research globally, we would like to consider regional characteristics, such as the differences in the business environments in which Japanese and overseas companies operate, and use these characteristics to develop appropriate strategies and policies.

In this study, by classifying companies into manufacturing and nonmanufacturing sectors and adopting a quadratic model, a different relationship between ESG and financial performance was shown in comparison to when a linear model was simply applied to all companies. This shows the effectiveness of the framework of this study, but also leads to the need for further analysis. The classification of the manufacturing and nonmanufacturing sectors highlights the importance of considering industry characteristics. Looking at the average PBR and PER by industry published monthly by the Japan Exchange Group [54], there is more than a 10-fold difference between the industry with the maximum value and the industry with the minimum value. To examine industry characteristics in detail, it is necessary to conduct further analysis using more-detailed industry classifications or extracting the major manufacturing and nonmanufacturing industries, which will be a research topic for the future.

A significant relationship was found in the quadratic model, with one inverted U-shaped case in the manufacturing sector and four U-shaped cases in the nonmanufacturing sector. In Japan, the manufacturing sector is leading the way in SDG initiatives [47]. Table 2 shows that the average CSR ratings for the manufacturing sector are higher than those for the nonmanufacturing sector, suggesting that the manufacturing sector is more advanced in terms of their ESG activities. For example, it is possible to consider that in the manufacturing sector, as initiatives are underway, further initiatives will lead to diminishing returns due to overinvestment, whereas in the nonmanufacturing sector, cases have emerged in which delays in initiatives will lead to declines in financial performance at the time of initial investment. By conducting further analysis using the detailed industry classifications mentioned above, it may be possible to examine industry characteristics and the application of the quadratic model in more detail. As an extension of the squared model, Chau et al. [55] analyzed the nonlinear relationship between ESG and corporate value by modeling a cubic response function. We would like to utilize such analytical methods in future research.

While this study analyzed the relationship between ESG activities and financial performance one year later on the basis of previous studies [42,44], depending on the ESG category, it may take time for the results of the efforts to appear. For example, Yanagi’s [50] study assumed a lag effect—that is, how many years later ESG performance will be reflected in the improvement in the PBR. Rojo-Suárez and Alonso-Conde [56] suggested that better ESG performance results in lower value creation at longer horizons, mainly because substitution effects channel market value via higher long-term discount rates. Depending on the time lag, there may be either a positive or negative impact, and it is important to consider the time lag when looking at the relationship between ESG activities and financial performance. For reference, on the basis of the current five-year dataset, we also conducted an additional experiment using a three-year dataset with a two-year lag (see

Table A7. Results of the panel data analysis for all companies [2 years lag].

CSR Rating		HR			E			G			S
Financial Data		PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Linear	Model	FE	RE	FE	FE	RE	RE	FE	RE	RE	FE	RE	RE
	COEF	−0.033 (−0.520)	7.064 *** (4.070)	−0.166 (−0.309)	0.005 (0.112)	3.083 * (1.941)	0.493 (1.464)	0.036 (0.757)	5.809 *** (3.992)	0.695 ** (2.082)	−0.018 (−0.250)	8.845 *** (5.387)	0.813 ** (2.489)
	Adjusted R2	0.886	0.017	0.613	0.886	0.006	0.019	0.886	0.013	0.023	0.886	0.024	0.025
Quad -ratic	Model	FE	-	-	FE	-	RE	FE	-	RE	FE	-	-
	COEF_L	0.071 (0.418)	-	-	−0.042 (−0.251)	-	−1.668 (−0.994)	0.125 (0.698)	-	−2.579 (−1.635)	0.196 (1.251)	-	-
	COEF_Q	−0.018 (−0.552)	-	-	0.008 (0.266)	-	0.392 (1.293)	−0.016 (−0.531)	-	0.585 ** (2.027)	−0.040 (−1.111)	-	-
	Adjusted R2	0.886	-	-	0.886	-	0.021	0.886	-	0.026	0.886	-	-
	Turning point	-	-	-	-	-	-	-	-	-	-	-	-
	Mean

Note 1: The number of observations is 1905 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

,

Table A8. Results of the panel data analysis for the manufacturing sector [2 years lag].

CSR Rating		HR			E			G			S
Financial Data		PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Linear	Model	FE	RE	FE	FE	FE	FE	FE	RE	FE	FE	RE	FE
	COEF	−0.008 (−0.115)	4.885 ** (2.213)	0.297 (0.412)	−0.038 (−0.581)	−11.524 * (−1.723)	−0.023 (−0.023)	0.014 (0.240)	5.487 *** (3.075)	−0.161 (−0.207)	0.083 (1.058)	9.875 *** (4.343)	−0.359 (−0.310)
	Adjusted R2	0.877	0.012	0.602	0.877	0.435	0.602	0.877	0.013	0.602	0.877	0.027	0.602
Quad -ratic	Model	-	-	-	FE	-	-	-	-	-	FE	-	-
	COEF_L	-	-	-	−0.329 (−0.826)	-	-	-	-	-	0.442 * (1.714)	-	-
	COEF_Q	-	-	-	0.049 (0.720)	-	-	-	-	-	−0.061 (−1.472	-	-
	Adjusted R2	-	-	-	0.877	-	-	-	-	-	0.877	-	-
	Turning point	-	-	-	-	-	-	-	-	-	-	-	-
	Mean

Note 1: The number of observations is 1125 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the italic number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

and

Table A9. Results of the panel data analysis for the nonmanufacturing sector [2 years lag].

CSR Rating		HR			E			G			S
Financial Data		PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE	PBR	PER	ROE
Linear	Model	FE	FE	RE	RE	FE	RE	RE	FE	RE	FE	FE	RE
	COEF	−0.074 (−0.582)	18.25 ** (2.536)	0.564 (1.095)	0.078 (1.510)	6.469 (1.335)	0.915 (1.628)	0.134 ** (2.257)	4.657 (0.850)	0.790 (1.384)	−0.125 (−1.015)	−3.898 (−0.798)	0.755 (1.582)
	Adjusted R2	0.904	0.487	0.008	−0.002	0.479	0.008	0.016	0.479	0.009	0.904	0.478	0.009
Quad -ratic	Model	-	-	-	RE	-	RE	-	-	RE	-	-	-
	COEF_L	-	-	-	−0.257 (−1.499)	-	−1.379 (−0.636)	-	-	−1.936 (−0.706)	-	-	-
	COEF_Q	-	-	-	0.066 * (1.921)	-	0.442 (1.029)	-	-	0.497 (0.957)	-	-	-
	Adjusted R2	-	-	-	0.006	-	0.011	-	-	0.014	-	-	-
	Turning point	-	-	-	-	-	-	-	-	-	-	-	-
	Mean

Note 1: The number of observations is 780 in all cases. Note 2: FE represents the fixed effects model, and RE represents the random effects model. The number in the upper row is the coefficient, and the number in parentheses in the lower row is the t value for the FE or the z value for the RE. Note 3: ** and * indicate significance at the 5% and 10% levels, respectively.

in Appendix A). While a significant positive relationship was found between the PER and all CSR rating categories for all companies, differences from the original analysis were observed, such as the disappearance of the relationship between “S” and the PBR. This suggests that ESG initiatives can have a positive effect on expectations for corporate growth, depending on the time lag. As a future challenge, however, we would need to analyze the impact of the time lag more precisely by preparing a longer-term dataset.

6. Conclusions

The purpose of this study was to determine the impact of the nonfinancial performance of ESG activities on financial performance. The results of our panel data analysis of 635 Japanese companies revealed that the combinations of ESG indicators that showed that significant relationships varied depending on the ESG field, but that companies with more advanced ESG activities had better financial performance; moreover, there were also cases where a nonlinear relationship existed. The analysis results also differed between the manufacturing and nonmanufacturing sectors, suggesting the importance of considering industry characteristics. Analysis of all the companies revealed that efforts in the “E” category led to an improvement in the PER, which indicates a company’s growth potential; efforts in the “G” category led to an improvement in the ROE, which indicates management efficiency; and efforts in the “S” category led to an improvement in the PBR, which indicates market added value. More detailed analysis of these factors is needed to understand that the financial indicators showing significant relationships vary depending on the ESG category, and that these relationships also vary depending on the industry; however, the insights gained from this study are likely to provide useful suggestions for companies to develop strategies for sustainable management and for policymakers to support corporate efforts.

For companies to effectively implement ESG activities, it is important to clarify the objectives and expected results of each ESG category. ESG activities that lead to greater financial performance make internal decision-making easier, but if the focus is on financial performance, it is necessary to be careful about overinvestment. On the basis of their findings that ESG investments by companies have an inverted U-shaped relationship with corporate value, Bagh et al. [57] suggested that firms should embrace well-balanced, enduring ESG strategies to steer clear of excessive investment and declining returns.

However, ESG activities that do not lead to better financial performance cannot be neglected. Given the increasing demand for ESG disclosure, all ESG activities will continue to be important. Furthermore, the “HR” category, which did not have a significant relationship with the financial indicators in the analysis of all the companies, comprises activities that form the basis of ESG activities, and it is generally expected that enhancing human capital will improve a company’s long-term competitiveness. Japan’s Ministry of Economy, Trade and Industry (METI) [58] released a report that states that in order to sustainably increase corporate value, it is important to have a human resource strategy that is compatible with management strategies, such as securing and developing human resources and building organizations that generate innovation and added value.

In order to proceed with ESG activities that do not lead to improved financial performance, it is advisable to clarify the objectives and impacts of these activities by understanding nonfinancial performance. For example, to understand social impacts other than financial performance, it would be useful to construct a logic model, as proposed by Teubler [59], to examine how ESG activities create social impacts through outputs and outcomes or to attempt to convert social impacts into a monetary value, as in the impact-weighted accounting proposed by Rouen and Serafeim [60]. In addition, when considering policy measures to encourage corporate ESG activities, it is important to design well-defined and detailed regulations and incentives that consider the relationships between ESG categories and industry characteristics and financial performance.

The demand for ESG is one of the important changes that affects the business environment of companies. Teece [61] stated that in a situation of increased uncertainty in the business environment, it is important to utilize “dynamic capability,” a higher-order transformational capability, to integrate, build, and reallocate resources inside and outside of the company. While companies must understand the results of related initiatives, corporate transformation is required to improve both financial and nonfinancial performance through the optimal allocation of resources. Furthermore, effective corporate ESG activities help promote the development of a sustainable society, as exemplified by the SDGs. We hope that our findings regarding the different impacts of ESG activities according to ESG category and industry sector, with the consideration of nonlinear relationships, will assist firms in the development of sustainable management strategies and support policymakers in the development of measures to promote corporate ESG activities.