The Nonlinear Relationship between Corporate Social Responsibility and Hospitality and Tourism Corporate Financial Performance: Does Governance Matter?

Abstract

This paper is interested in examining the impact of corporate social responsibility and governance on corporate financial performance. We selected a panel of 141 worldwide hospitality and tourism firms spanning the period 2012–2018 to assess the effects (direct and indirect) of corporate social responsibility and governance on corporate financial performance (measured in terms of return on assets, return on equity and Tobin’s Q). Although a few studies examine the moderating effect of certain factors, our study fills this gap by examining the moderating effect of governance practices (governance structure and institutional quality) on the nonlinear relationship between corporate social responsibility and corporate financial performance. The results of the system generalized method of moments suggest the existence of a nonlinear, U-shaped relationship between corporate social responsibility and corporate financial performance (return on equity and Tobin’s Q). This nonlinearity is confirmed for corporate social responsibility and corporate financial performance (measured by return on assets). However, this relationship is inverted-U-shaped. Furthermore, our results also show that lagged corporate social responsibility, governance practices, firm-specific variables and macroeconomic variables affect current corporate financial performance. The predictions of stakeholders and agency theories are validated. Given our results, it is recommended that policy makers trade off the benefits and costs of corporate social responsibility and take appropriate financial strategies, thus enabling value creation for their companies.

1. Introduction

The implementation of appropriate strategies serves to maximize firm value and minimize costs. However, the development of these different strategies requires understanding and responding to the requirements of all the company’s stakeholders (shareholders, employees, consumers, communities, etc.). It is therefore necessary for the company to use corporate social responsibility (CSR) activities to enhance its performance [1,2]. According to [3], CSR is perceived as a multidimensional concept. It constitutes a set of economic, legal, social, environmental, ethical, and corporate governance decisions, thus aiming to satisfy the expectations and interests of the company’s many stakeholders (internal and external) and thus enabling the creation of value for them [3,4]. From this, engaging in CSR activities offers several benefits to the company, such as improved profitability, competitiveness, stability, the sustainability of the company and even a strengthening of its moral identity [5,6,7]. Specifically, hospitality and tourism (H&T) firms often aim for long-term performance. By using CSR, these firms may seek to reduce costs, attract more CSR-conscious customers, or develop new revenue opportunities. In addition, CSR activities can help improve the reputation and brand image of these firms, which can attract a wider customer base and retain existing customers.

The H&T sector Is increasingly influenced by CSR, with initiatives such as the International Tourism Partnership and the publication of annual reports highlighting a growing commitment to ethical practices. However, it is important to consider both the pros and cons of CSR. On the one hand, CSR can enhance the brand image of H&T companies, attract customers who are aware of social and environmental issues, and foster fruitful partnerships. It can also lead to a more efficient use of resources, thereby improving performance. On the other hand, implementing CSR programs involve significant costs, particularly for staff training and compliance with environmental standards. Failure to meet commitments can also damage a company’s reputation. There are also situations where companies fail to meet their commitments, and this has a negative impact on their reputation [8]. For example, cruise lines have recently been criticized for their negative impact on certain marine destinations. In these circumstances, it is crucial to conduct in-depth research to assess the overall impact of CSR on the H&T corporate financial performance (CFP), based on the trade-off between its benefits and costs.

In theory, the relationship between CSR and CFP has its origins in two main theories, namely stakeholder theory [1] and the corporate finance theory (in particular, agency theory [9,10]). Under stakeholder theory, [1] shows that investment in CSR activities generates several benefits to various stakeholders (e.g., improved reputation, productivity, loyalty, etc.). This, in turn, contributes to the creation of value for them, including for shareholders. In contrast, according to corporate finance theory, CSR is seen as a value-destroying strategy as it generates high agency costs [9,10]. In practice and with reference to hospitality literature, there are differing viewpoints on the impact of CSR on CFP. Notably, several studies have shown a positive relationship between CSR and CFP in the H&T industry [5,11,12,13,14,15,16,17,18,19]. Nevertheless, others have found that CSR has a negative impact on CFP [11,20,21,22,23,24,25]. There is also some empirical work in the literature that has proven the existence of a nonlinear relationship between CSR and CFP, confirming both the predictions of stakeholders and corporate finance theories [8,26,27,28].

Based on the corporate finance literature and with reference to agency theory, engagement in CSR activities is not the only pillar through which CFP is affected. There are other pillars, the most common of which are corporate governance practices (GP) that are capable of directly or indirectly affecting CFP [29,30,31,32,33,34]. In addition, see [19,35] on corporate governance. It is also important to note that CFP is not exclusively influenced by CSR and/or governance. Let us assume that all companies react and interact to the environment in which they operate (economic, institutional, legal environment, etc.). Therefore, it makes sense to assess the contribution of the business environment to the optimization of CFP [36].

In this context, the main objective of this paper is to analyze the nonlinear relationship between the CSR and CFP of worldwide H&T firms. Based on the literature in this field, few studies have examined the nonlinear relationship between CSR and CFP [8,28], neglecting the dynamic nature of CFP. Therefore, this study aims to fill this gap by examining the dynamic nature of firm performance using both accounting (ROA, ROE) and market (Tobin’s Q) measures. Based on this relationship, two types of firms can be defined. First, companies that experience a rise (reduction) in CFP as a result of improving their CSR practices, up to a certain point where further improvements by managers could enhance their CSR practices. Secondly, firms that surpass the optimum level can show a CFP increase.

Likewise, it is important to further assess the relationship between CSR and CFP, controlling for other factors that may moderate this relationship. Based on the literature, few studies have examined the moderating role of certain factors on the CSR–CFP nexus. These studies focused on firm-level (firm size [37], CSR awareness [38] and quality management [8]) and economic-level (economic conditions [33], oil prices [39]) conditions, geographical diversification [40] and innovation [41] factors. However, there has been no research concentrating on the moderating effect of governance practices (governance structure and institutional quality). Although some studies suggest that CSR and GP impact CFP, another research gap concerns the potential correlation between CSR and GP, as both approaches share common objectives, including transparency and trust among all stakeholders. In other words, it is necessary to examine the interactions between these two factors, i.e., their moderating effect.

The paper makes two significant contributions to the existing literature. Firstly, it explores the nonlinear relationship between CSR and CFP in worldwide H&T firms. Previous studies have mostly investigated the static model to explore the association between CSR and CFP [8]. To the best of our knowledge, this is the first paper to investigate the dynamic model examining the relationship between CSR and CFP in the H&T industry. Secondly, it analyzes the moderating role of GP on the CSR–CFP link.

This study has some important limitations. Firstly, it is based on international samples, raising questions about the generalizability of the results. Researchers should look further into the matter, distinguishing between developed and developing economies. In addition, the period studied is short, without considering the impact of the COVID-19 pandemic. In addition, the study focuses on the moderating effect of governance practices (governance structure and institutional quality), but other moderating factors such as organizational culture deserve exploration. Finally, the use of the ESG pillars to characterize CSR offers a global view, but an analysis of the individual pillars (environmental, social and governance) may provide clarification on their impact on PSC.

This paper is structured as follows. The first section deals with the context of our research (introduction), followed by a presentation of the literature review. The third section is devoted to the methodology adopted in our research. The analysis and discussion of the results are dealt with in the fourth section. The conclusion, the different stages and the results of our research will be presented in the fifth and last section.

2. Literature Review and Hypothesis Development

2.1. CSR and Corporate Financial Performance

The question of the link between CSR and CFP has been the subject of much research in various disciplines (management, finance, accounting, etc.). Referring to the accounting and finance literature, the study of the association between CSR and CFP is a topic that has been widely discussed in socialist and corporate finance theories, the most common of which are stakeholders and agency theories.

Under stakeholder theory, [1] states that the maximization of corporate value is not restricted to reconciling the link between the firm’s managers and shareholders. It requires considering the participation of other stakeholders (financial and non-financial) in the firm’s activities [42]. This participation takes place through the organization’s commitment to investing in CSR activities. This is likely to appease the needs of all stakeholders (partners, shareholders, lenders, employees, customers, suppliers, communities, associations, NGOs, etc.). It further helps to create value for all stakeholders, including shareholders. To do this, [1] recommends that the organization invest in CSR activities because this is likely to encourage managers to reconcile their relations with their stakeholders to achieve the desired objectives. According to this theorist, engaging in CSR activities is a viable strategy, helping to improve the profitability, stability, growth, and sustainability of the organization. However, stakeholder theory may not be appropriate for small companies [43,44].

Contrary to the socialist view and based on the principle of shareholder wealth maximization, proponents of the agency theory have challenged the idea that engagement in CSR activities improves CFP [9,10]. According to [45], CSR does not maximize company value. In dealing with the manager–shareholder relationship, the manager assumes responsibility only to satisfy the needs of the shareholders. The individual responsibility of the enterprise is then to maximize shareholder value; [9] added that engaging in CSR activities generates agency problems and additional costs for the company. Investment in CSR activities may lead managers to misuse the resources of their companies and not add value to the company. This misuse of resources generates problems of information asymmetry and agency costs, thus deteriorating firm value [46]. Therefore, proponents of agency theory predict a negative correlation between CSR and CFP.

Certainly, improving CFP is not just about engaging in CSR activities. It requires the implementation of a good corporate governance strategy. Under stakeholder theory, [1] indicated that companies with good governance have good and sustainable corporate performance. In addition, agency theorists [10] showed that the application of good governance maximizes firm value. They suggested that the implementation of adequate governance practices and the improvement of financial reporting quality minimizes agency costs and financial costs. To avoid agency costs, the manager engages in the publication of certain information that is useful for the different partners of the firm (shareholders/creditors) and prevents conflicts of interest between them. This strategy, which qualifies as a voluntary disclosure strategy for financial information, is likely to minimize agency costs. Notably, good corporate governance is an effective way to diminish the costs of monitoring managers by shareholders (investors) and lenders (creditors). In addition to reducing agency costs, the implementation of a governance strategy helps to minimize the cost of capital incurred by the company’s transactions, while reducing the level of information asymmetry between shareholders and creditors. This consequently generates a reduction in financial costs and maximization of the firm’s value [47].

According to the hospitality literature, research on the impact of engagement in CSR activities on CFP has been the subject of several empirical studies. These studies have been undertaken in various industries such as the hotel, casino, airline and restaurant industries. The outcomes of these studies have been varied. Firstly, certain studies proposed a direct relationship (either positive or negative) between CSR and a CFP. For example, various researchers have demonstrated the existence of a positive correlation between the CSR and CFP of companies in different sectors, including the H&T industry [19,20,48,49,50].

Referring to the empirical studies related to the H&T sector, some researchers, like [12], proved a positive relationship between CSR and CFP (measured by return on asset (ROA) and average market value) in the hotel and casino industries. In addition, ref. [26] found similar results in the airline industry. The most important conclusion reached by [26] is that hospitality firms’ increased engagement in CSR activities serves to maximize their values. This result was further confirmed by [20] for the case of the United States and by [17] for the case of Portugal. These authors concluded that the implementation of socially responsible strategies by companies, combined with the understanding and satisfaction of various stakeholders, helps them to enjoy greater economic benefits than their competitors.

However, other researchers have found opposing results to those mentioned above in both the H&T sector and in other industries [22,24,39,51]. The second group contains empirical studies that have demonstrated neutrality between CSR and CFP. Empirically, this is because the CSR has no significant influence on CFP [21,22,52]. The third group includes empirical work that has found a non-linear relationship between CSR and CFP in various industries [8,26,27,28]. More specifically, [26] showed the existence of a U-shaped relationship between CSR and CFP (measured by ROE), operating in the restaurant industry. The same result has already been confirmed by [28], where the authors advocated for the need to make a trade-off between the profits and costs of investing in CSR activities. Their study agrees with the predictions of stakeholder theory. The results obtained by [8] further supported the predictions of stakeholder theory, as the authors found a non-linear relationship between CSR and CFP in the technology sector (measured by ROE). These different studies have suggested that the link between CSR and CFP is U-shaped. Nevertheless, although other researchers found a nonlinear relationship between the two CSR–CFP constructs, they have found that this nonlinearity is inverted-U-shaped, supporting the hypotheses proposed by both stakeholder and agency theories [28].

In summary, although the literature on the study of the correlation between CSR and CFP is vast, it remains ambiguous. Indeed, there is no unanimous relationship between the two constructs. In this context, ref. [53] has shown that the divergence of results obtained by researchers could be explained by the frequent use of different measures related to CSR (dimension, scores, etc.), and/or to CFP (including accounting and market measures). In addition, this discrepancy could be attributed to the chosen context (developed or developing countries), the economic environment (GDP growth, inflation rate, etc.) and the specific characteristics of the company (listed or unlisted, SME or large company, its sector of activity, its capital structure, its governance, etc.) and to country-level factors (e.g., the governance practices applied by the countries), or even the estimation method applied by the researchers [19,36].

According to [21,22], there are three reasons for the neutrality between SCR and CFP. First, ignoring time series effects leads to biased results. Second, the failure to consider the mediating role of intangible assets in the design process generates a sample selection bias and poor results. Finally, the lack of control variables that better explain the model to be regressed are the main causes of the non-robustness in the results obtained.

To address the missing information, we will investigate the relationship between CSR and CFP in the worldwide hospitality industry. Our main contribution is using the CFP dynamic model (the generalized method of moments (GMM) system). This approach enables us to factor in the changing nature of CSR investments, tackle the endogeneity issue, and identify any nonlinear connections between the two concepts. We will also consider the influence of governance practices (both structural and institutional quality), as well as how governance practices affect the relationship between CSR and CFP. To enhance our findings, we will include specific variables related to individual firms and macroeconomic factors. This will allow us to gain more robust results.

Indeed, most studies focus on analyzing the linear relationship between CSR and CFP, without considering the trade-off between these two variables. However, CSR generates both costs and benefits that can differ depending on the degree of success achieved regarding CSR [8,28,54].

Theoretically, the existence of the relationship between CSR and CFP can be explained by two main theories. The first is agency theory [9,10], based on the view that too high a level of investment (overinvestment) [55,56] in CSR can lead to a negative relationship with firm performance. This perspective, based on managerial opportunism and agency theory, argues that CSR engagement can represent an excessive use of valuable resources, motivated primarily by opportunistic interests such as image enhancement or reputation protection [54]. As a result, the costs of CSR could outweigh the benefits in terms of financial performance, which could have a negative impact on the firm’s profitability and competitiveness [55,57]. The second corresponds to the stakeholder theory, which is based on the idea that CSR generates greater utility and therefore contributes to increasing CFP [1]. In this context, CSR can be used by companies to resolve conflicts between managers and non-investing stakeholders. This reduces costs. It mitigates potential risks by acting as a form of insurance for the firm’s relational intangibles. By generating moral capital and goodwill among stakeholders, CSR provides insurance protection that reduces the firm’s exposure to risk and thereby maximizes its value [57].

Existing works in the literature highlight that the impact of CSR on CFP can vary (positive or negative) depending on the reaction of different stakeholders. Companies that make significant progress in CSR may be encouraged by stakeholders, according to stakeholder theory. However, under agency theory, companies are at risk of neglecting the efforts of all stakeholders and are at risk of neglect. They focus solely on shareholder wealth maximization. Under these conditions, it is only when CSR reaches a certain level of success that the benefits of CSR in terms of CPF outweigh the costs [8,28,58]. Previous studies have hypothesized that the influence of corporate social responsibility (CSR) on CFP could follow a U-shaped curve [8,26,27]. Furthermore, it has been suggested that this influence could also take the form of an inverted U-shaped curve, based on the principles of agency and stakeholder theories [59].

In summary, the nonlinear relationship between CSR and CFP could be apprehended via agency and stakeholder theories. This means that at relatively low CSR values, CSR has positive but diminishing effects as CSR increases, because as CSR increases, the negative effect starts to dominate more and more. Once a threshold is crossed, the effects in line with agency theory dominate those of stakeholder theory. In this case, managers may be inclined to invest excessively in costly CSR activities to the detriment of shareholders’ interests, which can lead to a decrease in CFP. In this case, we devise our first hypothesis as follows:

H1. 

The relationship between CSR and CFP is nonlinear.

H1(a). 

CSR positively affects CFP due to the dominating effect of the factors from the stakeholder theory over the agency theory.

H1(b). 

CSR negatively affects CFP due to the dominating effect of the factors from the agency theory over the stakeholder theory.

2.2. Do Corporate Governance Indicators Matter?

Certainly, improving CFP is not just about engaging in CSR activities. It requires the implementation of a sound corporate governance strategy. This will allow the firm to increase its profitability and maximize shareholder wealth. Theoretically, the existing link between corporate governance and CFP has its origins in agency theory. Specifically, ref. [10] showed that the implementation of good governance maximizes the firm’s value. They suggested that developing adequate governance practices and enhancing the quality of financial information minimizes agency and financial costs. To avoid agency costs, the manager engages in the disclosure of certain information that is useful for the different partners of the firm (shareholders/creditors) to prevent conflicts of interest between them. This strategy, thus qualifying a strategy of voluntary disclosure of financial information, is likely to minimize agency costs. Therefore, good corporate governance is an effective way to decrease the costs of monitoring managers by shareholders (investors) and lenders (creditors). In addition to reducing agency costs, the implementation of a corporate governance strategy helps to minimize the cost of capital incurred by firm transactions, while reducing the level of information asymmetry between shareholders and creditors. This consequently generates a decrease in financial costs and maximization of the firm’s value [47]. Furthermore, in the context of the stakeholder theory, ref. [1] indicated that companies with good governance have good and sustainable business performance.

Research into the association between corporate governance and CSR has been a major focus in the literature within the framework of stakeholder theory. According to [60,61], good corporate governance helps create value for all stakeholders (including maximizing shareholder value) through investment in CSR activities. According to these authors, maximizing shareholder value is not only achieved through the integration of value for all stakeholders. This strategy requires the implementation of appropriate governance practices. However, proponents of agency theory have shown that engagement in CSR activities urges managers to behave opportunistically and expropriate shareholder wealth. Managerial opportunism could in turn generate increased agency costs and deterioration in firm value [62,63]. On this theme, [64] indicates that investment in CSR activities is time consuming. It takes a long time to contribute to improving CFP, even if that firm applies appropriate governance practices. [64] links this to the demands on managers not to assume their responsibilities over the long term.

Empirically, numerous researchers have shown that the application of appropriate governance practices helps to ensure the objectives set by the company. They have proven that corporate governance maximizes CSR performance [29,65,66]. As an illustration, several studies showed that the board of directors is an important element in maximizing CSR performance [29,31,32,33]. In addition, a few studies have proven that governance structure moderates the impact of CSR on CFP [20,67]. Given these different arguments and with reference to the H&T literature, the topic related to the study of the moderating role of corporate governance on CFP in the H&T sector via CSR is the least exploited. To fill this gap, we will test this effect while formulating our second hypothesis, as follows:

H2. 

The nonlinear relationship between CSR and FP is moderated by CGS.

2.3. Does Country-Level Governance Matter?

In addition to corporate governance, the governance practices applied by a country play a significant role in improving corporate performance. Referring to the financial literature, many researchers have advocated for the implementation of good country-level governance to improve disclosure and accountability. In this context, refs. [68,69,70] showed that the development of good country-level governance allows for the better allocation of resources (or availability of external funding), promotes investment activities (especially in R&D) and thus increases the CFP [69], considering that country-level governance (or institutional factors) allows firms to cope with uncertainties when engaging in investment activities [71,72]. In addition, according to [54], good (poor) institutional quality may act as a fundamental factor influencing the behavior of firms, inducing them to adopt responsible (irresponsible) practices. Recently, refs. [73,74,75] showed that situational and institutional factors are likely to influence CSR initiatives and their impact on CFP. In addition, ref. [76] argued that poor national policies (in particular, corruption) could be the cause of lower levels of CFP. In line with the work of [70,75], we will formulate our third hypothesis as follows:

H3. 

The nonlinear relationship between CSR and CFP is moderated by institutional quality.

3. Methodology

3.1. Sample

In this study, we selected a sample of 141 listed worldwide H&T firms covering the period 2012–2018. CSR data, governance indicators and financial variables for H&T firms were extracted from the Refinitiv Eikon database. This database provides financial data on listed financial and non-financial companies (including H&T industry). According to Refinitiv, 2019, it includes almost 99% of the world’s market capitalization.

The selection of the sample and study period was primarily due to the availability of data on GRI framework adoption, CSR and governance indicators for H&T firms [77,78]. The Refinitiv Eikon database contains the financial data of 1721 H&T firms [77,78]. However, the data related to CSR and financial data were only available for 89 H&T companies in 2012. Moreover, [78] points out that the number of firms that were engaged in CSR disclosure had risen to 172 companies in 2018. Since our study also focuses on corporate governance data, only 141 companies had both CSR data and corporate governance indicators.

Table 1. Number of firms that are engaged in CSR disclosure.

Country	CSR Data Availability
Australia	9
Brazil	1
Canada	6
Chile	1
China	2
France	3
Germany	5
Greece	1
Hong Kong	12
Ireland	2
Italy	1
Japan	16
Malaysia	4
New Zealand	1
Philippines	1
Singapore	3
South Africa	4
South Korea	2
Spain	3
Sweden	1
Switzerland	1
Taiwan	3
Thailand	1
Turkey	1
United Kingdom	21
United States	36
Total	141

summarizes the number of H&T firms that are engaged in CSR disclosure by country.

3.2. Econometric Model

In this study, we will examine the link between CSR and CFP and the moderating role of corporate and country level governance on this relationship. Specifically, we will assess the three hypotheses mentioned above in three steps.

 Step 1.

CSR and CFP

The first step wants to address the nonlinear association between CSR and CFP. To do this, we used the two-step system GMM developed by [79]. The estimated model can be written as follows:

$$\begin{array}{l}CF{P}_{ijt}={\beta }_0+{\beta }_{1}CF{P}_{ijt-1}+{\beta }_{2}CS{R}_{ijt}+{\beta }_{3}CS{R}_{ijt}^2+{\beta }_{4}CG{S}_{ijt}+{\beta }_{5}LE{V}_{ijt}+{\beta }_{6}SIZ{E}_{ijt}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\beta }_{7}GD{P}_{it}+{\beta }_{8}IN{F}_{it}+{\epsilon }_{ijt}\end{array}$$

(1)

where $CF{P}_{ijt}$ is corporate firm performance measurements (ROA, ROE and Tobin’s Q) of country I of firm j measured at period t, $CS{R}_{ijt}$ is the economic, social and governance responsibility (CSR) score of country i of firm j measured at period t, and $CS{R}_{ijt}^2$ is its quadratic term assessing for the nonlinear relationship. $CG{S}_{ijt}$ is the corporate governance score, $LE{V}_{ijt}$ is the leverage, $SIZ{E}_{ijt}$ is the firm size, $GD{P}_{it}$ is the economic growth rate, $IN{F}_{it}$ is inflation rate and ${\epsilon }_{ijt}$ is the error term.

To investigate the topic of CFP, several papers used two-step system GMM [72,80]. There are several advantages to using system GMM. First, by allowing any independent variable to be correlated with the error term, system GMM can mitigate the problems of omitted variables, measurement error, dynamic panel heterogeneity and potential endogeneity [72,80,81]. This endogeneity can occur because of reverse causality. This causality can occur because CSR impacts CFP, but CFP can also impact CSR [77,82,83], because an increase in the level of CFP can promote managers to engage in CSR activities. Second, it is effective when a panel has a smaller period dimension (T) compared to its cross-sectional dimension (N) [72,81].

In our study, the choice of system GMM is justified by several arguments. First, our study focuses on a worldwide sample, which should generate a heterogeneity problem. Therefore, system GMM could solve this problem. Second, the endogeneity problem was statistically detected using the Durbin Wu Husman (DWH) test. Therefore, the use of system GMM is necessary to solve this problem [79]. Finally, as mentioned earlier, our sample is composed of a smaller period dimension (T is equal to 7) than the cross-sectional dimension (N is equal to 141) [72,81]. This encouraged us to use the GMM system to resolve this problem.

In addition, the AR (1) and AR (2) autocorrelation of first and second order cannot be rejected, which implies that autocorrelation is absent. This suggests that the models’ lag structure is satisfactory and requires only one lag for the CFP variable. The accuracy of the dynamic GMM estimation technique depends on using appropriate instruments in the assessment. In this case, lagged values such as t-1 and t-2 for the difference equation and a single lag for the level equation were used as instruments. The reliability of the instruments was assessed using the Hansen J statistic of over-identifying restrictions. This indicated that the instruments used were reliable for the models.

 Step 2.

CSR and CFP: the moderating effect of governance structure

In the second step, we examined the moderating role of governance structure on the nonlinear relationship between CSR and CFP. To do this, we also used the two-step system GMM developed by [79]. The estimated model can be written as follows:

$$\begin{array}{l}CF{P}_{ijt}={\beta }_0+{\beta }_{1}CF{P}_{ijt-1}+{\beta }_{2}CG{S}_{ijt}+{\beta }_{3}CS{R}_{ijt}+{\beta }_{4}CS{R}_{ijt}^2\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\beta }_{5}CS{R}_{ijt}\times CG{S}_{ijt}+{\beta }_{6}CS{R}_{ijt}^2\times CG{S}_{ijt}+{\beta }_{7}LE{V}_{ijt}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\beta }_{8}SIZ{E}_{ijt}+{\beta }_{9}GD{P}_{it}+{\beta }_{10}IN{F}_{it}+{\epsilon }_{ijt}\end{array}$$

(2)

 Step 3.

CSR and CFP: the moderating effect of institutional quality

In the third step, we analyzed the moderating effect of institutional quality on the nonlinear relationship between CSR and CFP. To do this, we also used the system GMM developed by [79]. The estimated model can be written as follows:

$$\begin{array}{l}CF{P}_{ijt}={\beta }_0+{\beta }_{1}CF{P}_{ijt-1}+{\beta }_{2}I{Q}_{ijt}+{\beta }_{3}CS{R}_{ijt}+{\beta }_{4}CS{R}_{ijt}^2\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\beta }_{5}CS{R}_{ijt}\times I{Q}_{it}+{\beta }_{6}CS{R}_{ijt}^2\times I{Q}_{it}+{\beta }_{7}LE{V}_{ijt}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\beta }_{8}SIZ{E}_{ijt}+{\beta }_{9}GD{P}_{it}+{\beta }_{10}IN{F}_{it}+{\epsilon }_{ijt}\end{array}$$

(3)

3.3. Variables

3.3.1. Dependent Variables

According to [53], it is essential to discriminate between accounting and market measures as they correspond to different characteristics of firm value and are influenced differently by CSR activities. Specifically, accounting measures are employed to assess short-term CFP. They are based on indicators of a historical nature that measure past performance; [84] indicated that accounting measures are sensitive to CSR activities. Therefore, they recommended the need to use market measures.

Following the work of [19,84], we adopted accounting and market measures to measure CFP. As for accounting measurement, we chose the two measures frequently used in the literature, namely return on assets (ROA) and return on equity (ROE). The ROA is measured by the ratio of Operating Income/Total Assets. As for ROE, it is calculated by the ratio of Net Income/Stockholders Equity.

In addition, the market measure was considered in this research to determine the firm’s value. This value reflects the long-term performance of the firm. It is measured by Tobin’s Q ratio, which is the ratio of the product of a company’s share price and the number of ordinary shares outstanding (MVE) plus the liquidation value of preference shares outstanding (PS) plus the value of current liabilities minus current assets plus the value of long-term assets and inventories (DEBT) divided by the book value of total assets (TA). This measure has been used by [84].

3.3.2. Independent Variable

Based on prior studies [8,24], we retained the ESG score to quantify the CSR efforts of H&T companies. This score was constructed based on 178 indicators grouped into three main dimensions, namely CSR, covering issues associated with ten main themes. The economic dimension focuses on three main themes: resource use (20 indicators), emissions (22 indicators) and innovation (19 indicators). The environmental dimension is based on four themes, namely workforce (29 indicators), human rights (8 indicators), community (14 indicators) and product responsibility (12 indicators). The governance dimension is based on three main themes: management (34 indicators), shareholders (12 indicators) and CSR strategy (8 indicators). These scores were collected from the Thomson Reuters Asset4 database.

3.3.3. Moderator Variables

Two governance indices were constructed to investigate the moderating role of governance structure and institutional quality. The first index, the Governance Structure Score (GSS), is composed of four dimensions. Indeed, several investigations have analyzed outline measures of governance structure and their relation to CFP [85,86,87,88] to earnings quality [89] and corporate carbon emission disclosure [90]. Depending on the availability of data, we constructed our Governance Structure Score (GSS) based on four internal governance indicators, namely the size of the board, the independence of the board, the duality of the CEO and gender diversification [78].

Principal component analysis (PCA) was then used to combine these indicators. PCA avoids multicollinearity and reduces the measurement error by aggregating the variables associated with each factor into a distinct composite score [90].

The second index consists of Institutional Quality (IQ), which is a composite measure. According to [91], there are six dimensions of country governance, namely voice and accountability (VA), political stability (PS), government quality (GE), regulatory quality (RQ), rule of law (ROL) and control of corruption (COC). Different sets of scores were used to measure specific aspects, such as government effectiveness from 0 to 4, regulatory quality and political stability from 0 to 12, and corruption, accountability and rule of law from 0 to 6. Higher scores represent stronger institutions. Higher scores tend to mean better governance. To combine these indicators, we also used PCA.

In this case, two specific tests were carried out on the two indices to test the suitability of the data for factor analysis: Bartlett’s test of sphericity and the Kaiser–Meyer–Olkin (KMO) test. The results of these two tests are shown in

Table A1. Results of the Bartlett test of sphericity, KMO measure of sampling adequacy and PCA (total variance explained).

Panel A: Tests of Applicability
		Bartlett Test of Sphericity			KMO
		Chi-Square	Degrees of Freedom	p-Value
GSS		357.387	6	0.000	0.641
IQ		8848.078	15	0.000	0.821
Panel B: PCA (Total Variance Explained)
Component		Eigenvalue	Difference	% of Variance	Cumulative Variance %
GSS	BS	1.810	0.927	0.453	0.453
	BI	0.883	0.103	0.221	0.6733
	CEO	0.780	0.253	0.195	0.868
	GD	0.526		0.132	1.000
IQ	VA	4.675	3.901	0.779	0.779
	PS	0.774	0.372	0.129	0.908
	GE	0.402	0.330	0.067	0.975
	RQ	0.072	0.017	0.012	0.987
	RL	0.054	0.033	0.009	0.996
	CC	0.022		0.004	1.000

. The p-values of the Bartlett’s sphericity test were all less than 0.05 (0.00), and the KMO values were 0.641 and 0.821, respectively, indicating that the data were suitable for PCA. To facilitate the analysis, we normalized these indices on a scale of 0 to 1 using the softmax normalization technique.

3.3.4. Control Variables

In order to have more robustness, we have introduced control variables in our models while referring to both the hospitality and the corporate finance literature [8,81,92,93,94]. First, we used firm financial size, which is determined by the logarithm of total assets, to capture the ability of firms to provide collateral. Leverage is used to control firms’ ability to access external finance. According to agency theory, leverage has a positive effect on CFP [10]. Indeed, leverage can reduce unprofitable investment projects. In this case, leverage could have a disciplinary effect by reducing the agency problem between stakeholders. It is therefore an appropriate measure of external financial constraints. This variable is measured by the ratio of total debt to total assets (LEV).

Finally, to control the macroeconomic conditions (economic stability), we used the inflation rate (consumer prices index (annual %)) and GDP growth (gross domestic product growth rate (at constant 2015 prices)) [36].

Table 2. Definition of variables.

Variables	Definitions	Acronyms	Source
Dependent variables
Return on assets	Ratio of Operating Income/Total Assets	ROA	Refinitiv Eikon
Return on equity	Ratio of Net Income/Stockholders Equity	ROE	Refinitiv Eikon
Tobin’s Q	Ratio of the product of a firm’s stock price and the number of outstanding common shares (MVE) plus the liquidating value of outstanding preferred stock (PS) plus the value of short-term liabilities minus short-term assets plus the value of long-term assets and inventories (DEBT) divided by the book value of total assets (TA)	Q	Refinitiv Eikon
Independent variable
Corporate social responsibility	Economic, Social and Governance Responsibility score	CSR	Refinitiv Eikon (Asset4)
Moderator variables
Corporate governance index	Composite variable using PCA	GSS	Refinitiv Eikon and authors’ own calculations
Board size	Number of directors on the board	BS	Refinitiv Eikon
Board independent	Percentage of non-executive directors of all directors on the board	BI	Refinitiv Eikon
CEO duality	Dummy variable, take one if a firm’s CEO chairs the board simultaneously, zero otherwise	CEO	Refinitiv Eikon
Gender diversification	Percentage of female directors of all directors on the board	GD	Refinitiv Eikon
Institutional quality	Composite variable using PCA	IQ	[91] and authors’ own calculations
Control variables
Leverage	Total debt over total assets	LEV	Refinitiv Eikon
Firm size	Natural logarithm of firm’s total assets	SIZE	Refinitiv Eikon
Inflation rate	Consumer prices index (annual %)	INF	WDI, Word Bank
GDP growth	GDP growth rate (annual %) (gross domestic product growth rate (at constant 2015 prices))	GDP	WDI, Word Bank

summarizes the definitions of all variables used in our study.

4. Results and Discussions

4.1. Diagnostic Tests

To test cross-sectional dependence, we used three cross-section dependence tests (Lagrange Multiplier (LM) test [95], the cross-section dependence (CD) test [96] and the Bias-Adjusted Cross-Sectional Dependence Lagrange Multiplier (CD LM) test [97]) whose null hypothesis was non-cross-sectional dependence (correlation) when weighted. The results of all tests indicated that the null hypothesis of no cross dependence was accepted for all three models, meaning that there was no significant cross dependence between cross-sectional units (

Table 3. Cross-sectional dependence tests.

Tests	Model (1)		Model (2)		Model (3)
	Statistic	Prob.	Statistic	Prob.	Statistic	Prob.
Breusch–Pagan LM	14.133	0.832	11.230	0.557	14.243	0.653
Pesaran scaled LM	−1.751	0.189	0.873	0.243	−0.473	0.256
Pesaran CD	0.083	0.721	−1.356	0.147	1.073	0.093

).

To test the stationarity of all the variables used in our balanced panel, we used a variety of unit root tests, namely [98,99], and Fisher-type (Augmented Dickey–Fuller) [100] tests. The null hypotheses of these tests indicated that all the panels contained a unit root.

Table 4. Unit root tests.

Variables	LLC	IPS	ADF
	Level	Level	Level
ROE	−22.919 ***	−11.152 ***	303.298 ***
	(0.000)	(0.000)	(0.000)
ROA	−15.7750 ***	−11.410 ***	325.558 ***
	(0.000)	(0.000)	(0.000)
Q	-9.627 ***	−10.966 ***	496.541 ***
	(0.000)	(0.000)	(0.000)
CSR	−33.185 ***	−8.889 ***	516.350 ***
	(0.000)	(0.000)	(0.000)
CGS	−14.4409 ***	−5.035 ***	437.858 ***
	(0.000)	(0.000)	(0.000)
IQ	−426.302 ***	−29.829 ***	565.424 ***
	(0.000)	(0.000)	(0.000)
LEV	−18.563 ***	−3.463 ***	385.351 ***
	(0.000)	(0.000)	(0.000)
SIZE	−217.403 ***	−55.982 ***	360.210 ***
	(0.000)	(0.000)	(0.000)
GDP	−69.951 ***	−15.051 ***	664.722 ***
	(0.000)	(0.000)	(0.000)
INF	−27.201 ***	−5.0457 ***	355.577 ***
	(0.000)	(0.000)	(0.000)

Notes: *** present the significance at the 1% levels. The values in parentheses are p-values. LLC is Levin, Lin and Chu test, IPS is Im, Pesaran and Shin W-stat test, ADF is augmented Dickey–Fuller (Fisher Chi-square) test.

presents the results of these tests, showing that all variables were stationary at the level, i.e., I (0).

4.2. Descriptive Statistics and Correlation Test

Table 5. Descriptive statistics.

Variable	Obs.	Mean	Std. Dev.	Min	Max	Skewness	Kurtosis	Normality Test
ROE (log)	987	2.815	−1.347	4.448	5.139	−0.126	7.344	0.000
ROA (log)	987	2.160	2.311	4.1412	3.852	−0.481	4.044	0.000
Q (log)	987	0.626	0.425	−4.605	2.690	−1.344	10.013	0.000
CSR (log)	987	3.873	3.223	1.552	4.551	−0.621	2.012	0.000
GSS (log)	987	3.9771	3.452	0.191	4.570	−0.827	2.340	0.000
IQ (log)	987	4.375	2.855	1.546	4.600	1.525	4.134	0.000
LEV (log)	987	4.127	3.386	1.714	5.564	−1.866	10.336	0.000
SIZE (log)	987	3.076	0.358	2.871	3.198	−1.079	2.857	0.000
GDP growth (log)	987	0.940	0.818	2.212	3.269	−0.790	6.455	0.000
INF (log)	987	0.571	0.519	0.558	2.802	8.163	71.459	0.000

provides a summary of the descriptive statistics of all the variables used in this study from 2012 to 2018. The table presents the mean, standard deviation and number of observations for each variable. In the period of 2012–2018, H&T firms had an average ROA of 2.160, ROE of 2.815 and Tobin’s Q of 0.626. In terms of social performance, the H&T firms studied had an average score of 3.873. Regarding control variables, all control variables had positive mean values through the 2012–2018 period.

Table 6. Correlation matrix.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
(1) ROE	1.000
(2) ROA	0.717 *	1.000
(3) Q	−0.057	−0.176 *	1.000
(4) CSR	0.094 *	0.062	−0.019	1.000
(5) GSS	0.168 *	0.258 *	−0.414 *	0.684 *	1.000
(6) IQ	−0.073 *	−0.030	0.002	−0.418 *	−0.294 *	1.000
(7) LEV	0.045	−0.064 *	0.257 *	0.453 *	0.244 *	−0.351 *	1.000
(8) SIZE	0.012	−0.170 *	0.061	0.272 *	0.179 *	−0.157 *	0.362 *	1.000
(9) GDP	−0.034	0.010	0.059	−0.334 *	−0.258 *	0.448 *	−0.290 *	−0.153 *	1.000
(10) INF	−0.047	−0.081 *	0.347 *	0.055	−0.260 *	−0.104 *	0.034	−0.014	−0.025	1.000

Note: this table presents the Pearson’s correlation matrix of dependent, independent and control variables; * denotes significance at the 5% level.

shows the correlation matrix results between the endogenous and exogenous variables and between exogenous variables in this study. The Pearson correlation test shows that there was no issue of multi-collinearity among the variables, as all the estimated coefficients had values that were below 0.70.

Table 7. VIF correlation.

	Model (1)	Model (2)	Model (3)
CSR	2.61	2.61	2.61
CGS	2.30	2.30	2.30
SIZE	1.45	1.45	1.45
IQ	1.44	1.44	1.44
GDP	1.31	1.31	1.31
INF	1.22	1.22	1.22
LEV	1.17	1.17	1.17
Mean VIF	1.64	1.64	1.64

summarizes the variance inflation factor (VIF) correlation results, confirming that there was no multicollinearity between the variables (VIF is less than 10).

4.3. Nonlinear Relationship between CSR and CFP

We present the following outcomes obtained by the system GMM method, aiming to investigate the nonlinear correlation between CSR and CFP in worldwide H&T firms. The findings obtained are presented in

Table 8. CSR and CFP: main results.

Variables	(1)	(2)	(3)
	ROE	ROA	Q
ROEt−1	0.978 ***
	(0.011)
ROAt−1		0.787 ***
		(0.011)
Qt−1			0.509 ***
			(0.053)
CSR	−0.421 ***	0.393 ***	−0.272 ***
	(0.091)	(0.114)	(0.046)
CSR2	0.079 ***	−0.053 ***	0.053 ***
	(0.015)	(0.017)	(0.007)
GSS	−0.076 ***	−0.080 ***	0.028 ***
	(0.019)	(0.009)	(0.009)
LEV	−0.020 ***	−0.020 ***	0.020 ***
	(0.001)	(0.001)	(0.001)
Size	−0.081 ***	0.036 ***	0.407 ***
	(0.007)	(0.007)	(0.054)
GDP	−0.056 ***	−0.061 ***	−0.019 ***
	(0.016)	(0.010)	(0.006)
Inflation	0.171	0.207 ***	−0.153
	(0.002)	(0.001)	(0.002)
Constant	1.689 *	−0.911 **	−1.002 ***
	(1.011)	(0.597)	(0.089)
Observations	846	846	846
Number of firms	141	141	141
Number of instruments	137	134	134
AR (1) (p-value)	0.014	0.002	0.000
AR (2) (p-value)	0.076	0.089	0.105
Hansen test (p-value)	0.567	0.496	0.327
Endogeneity tests	0.000	0.000	0.000
Country-specific effects	YES	YES	YES

***, ** and * denote statistical significance at the 1%, 5% and 10% levels, respectively.

. Based on the results attained by the endogeneity DWH test, it is obvious to reject the null assumption of exogeneity. In this case, we have chosen the system GMM method to remedy this problem. Moreover, the results of the autocorrelation test prove the conformity of the null assumption of the first and second order (AR (1) and AR (2)) serial correlation. Thus, we can notice that the outcomes of the J-Hansen test confirmed the validity of the null assumption of instrumental variables.

From Table 8, we note that the CSR, CGS and control variables had effects (negative and/or positive) on CFP. Notably, the coefficients related to the variable $CF{P}_{ijt-1}$ were positive and statistically significant in all three models at the 1% level (see columns 1, 2 and 3). This outcome suggests that the CFP is dynamic in nature, regardless of the indicator employed to measure it. The positive link between current CFP and lagged CFP reflected the presence of low adjustment costs. Specifically, this outcome was justified by the fact that H&T firms bore lower capital costs (cost of debt and cost of equity). They used their financing resources appropriately, so that their financing costs did not exceed their operating returns. It follows from this that H&T firms used their external financing resources efficiently. This is reflected in the presence of good control quality and the effectiveness of the financial strength of these firms.

Let us now turn to the results obtained on the CSR variable. In the three models, it seems that the coefficients associated with the variable $CS{R}_{ijt}$ were statistically significant at the 1% level, but they had opposite signs. The same result was found, moreover, for $CS{R}_{ijt}^2$, where its coefficients were all significant, but they affected CFP in different measures. Based on the different measures of CFP, it should be noted that the non-linear relationship between CSR and CFP needs to be tested to verify the shape of each model associated with each CFP measure. For that purpose, we used the test for the existence of a U-shaped relationship (inverse U) [101]. The purpose of this test is to analyze the lower and upper bounds of the correlation, as well as the extreme point. The results are shown in

Table 9. Test for the U-shaped curve.

	Model (1)		Model (2)		Model (3)
Group	Lower Bound	Upper Bound	Lower Bound	Upper Bound	Lower Bound	Upper Bound
Interval	1.551	4.551	1.551	4.551	1.551	4.551
Slope	−0.176 ***	0.298 ***	0.227 ***	−0.093 ***	−0.106 ***	0.212 ***
	(−3.815)	(6.284)	(3.717)	−2.031)	(−4.212)	(11.608)
Overall test
t-value	5.95 0.000 2.665		2.03 0.021 3.680		4.21 0.000 2.554
p-value
Extreme (inflexion) point

Note: t-values are in parentheses. *** present significance at 1% levels. The U-shaped (inverted U-shaped) curve was tested using the utest command in STATA 17.

. Indeed, the existence of a non-linear relationship is also evident in the results relating to the CSR–ROE link. This shows that the relationship between CSR and ROE became U-shaped, going from positive to negative. However, the results also show that there was an inverted U-shaped relationship between the CSR and the ROA. Finally, the results obtained for Tobin’s Q were similar to those obtained for ROE. This also suggests the presence of a U-shaped relationship.

Turning now to the control variables, the leverage variable (LEV) had a negative and significant coefficient (see columns (1), (2) and (3)). In the ROE model, the leverage had a negative influence on ROE. This result means that the H&T firms’ debt level may be too high, which can raise its interest expenses and financial exposure, potentially diminishing the returns for shareholders. It could also signal that the firm’s assets are not producing enough profits to cover the costs of paying its debt. In the ROA model, the leverage had a negative influence on ROA. This is because the increase in the capital structure generates a reduction in ROA. In this case, long-term debt is an inadequate source of supporting H&T firms to invest and therefore improve their performance. It can lead them to invest in unprofitable capital projects. The perverse impact of the capital structure on CFP could be attributed to the existence of high financing costs that can expose these firms to financial distress (agency costs and bankruptcy costs). To face this kind of risk, H&T firms need to finance their investment projects through their own funds to enhance their performance. In the Q model, leverage also negatively impacted Tobin’s Q. This result shows that capital structure is an efficient means of financing, thus providing value maximization of H&T firms. Consistent with signaling theory, a sign of financial risk or an indication that the firm’s assets are overvalued can, in turn, lead to a lower Tobin’s Q ratio. In other words, it suggests that the market believes the firm’s assets are overpriced, which can be a concern for investors and may result in a lower valuation for the firm.

In addition, the coefficients on firm size are all negative signs (columns (1), (2) and (3)). The inverse correlation between size and CFP is justified by the fact that small H&T firms come to terms with external financial constraints (access to credit), thus preventing them from improving their performance.

Furthermore, the GDP has a negative impact on CFP (see columns (1) and (2)). This result suggests that the benefits of economic growth are not shared equally among H&T firms or that some firms are not able to cope with the challenges and opportunities created by economic growth.

Finally, the coefficient on inflation had a positive sign at the 10% level (column (2)). The positive correlation between inflation and ROA could be attributed to the net income of H&T firms increasing faster than their total assets when the general price level of goods and services rises. This implies that H&T firms can maintain or improve their profitability and efficiency in the use of their assets. However, this result may not reflect the true economic value of H&T firms’ assets and income, as inflation distorts historical cost accounting and reduces the purchasing power of money [55]. However, inflation negatively affects the firm value. This could be because inflation reduces the purchasing power of consumers, who may spend less on travel and leisure activities. It could also be because inflation increases input and operating costs for H&T firms, which could reduce their profitability and cash flow.

4.4. Moderating Role of Governance Structure on CSR–CFP Nexus

The results of the moderating effect of GSS on the CSR–CFP nexus are provided in

Table 10. Moderating effect of governance structure.

Variables	(1)	(2)	(3)
	ROE	ROA	Q
ROEt−1	1.020 ***
	(0.013)
ROAt−1		0.973 ***
		(0.007)
Qt−1			0.920 ***
			(0.006)
CSR	−2.673 ***	−2.760 ***	0.383 ***
	(0.409)	(0.256)	(0.092)
CSR2	0.449 ***	0.440 ***	−0.047 ***
	(0.066)	(0.046)	(0.017)
GSS	−0.536 **	−1.749 ***	0.842 ***
	(0.270)	(0.133)	(0.060)
CSR X GSS	0.313 **	0.985 ***	−0.474 ***
	(0.154)	(0.085)	(0.036)
CSR2 X GSS	−0.055 **	−0.143 ***	0.066 ***
	(0.023)	(0.013)	(0.005)
Leverage	−0.200 ***	−0.200 ***	0.200 ***
	(0.011)	(0.012)	(0.002)
Size	−0.012	−0.045 ***	0.019 ***
	(0.010)	(0.005)	(0.004)
GDP	−0.066 ***	−0.043 ***	−0.025 ***
	(0.013)	(0.015)	(0.009)
Inflation	−0.008	0.009 *	−0.005 **
	(0.008)	(0.005)	(0.002)
Constant	13.412 ***	6.749 ***	0.565 ***
	(3.154)	(2.333)	(0.069)
Observations	846	846	846
Number of firms	141	141	141
Number of instruments	124	131	137
AR (1) (p-value)	0.018	0.003	0.000
AR (2) (p-value)	0.071	0.047	0.214
Hansen test (p-value)	0.317	0.562	0.878
Country specific effects	YES	YES	YES

***, ** and * denote statistical significance at the1%, 5% and 10% levels, respectively.

. In terms of the direct effect of GSS on CFP, the GSS had a positive effect on ROE and Tobin’s Q (columns 1 and 3). However, it had a negative effect on ROA (column 2). A 10.0% increase in GSS led to a 2.78% and 0.23% increase in ROE and Q. However, a 10.0% increase in GSS led to a 9.32% decrease in ROA. A negative impact of GSS on ROA can result from a variety of economic factors. Poor governance can lead to the misallocation of resources, inadequate management incentives, questionable accounting practices, lack of transparency and legal risks. All these elements reduce the profitability of the firm’s assets, resulting in a lower ROA. A strong GSS is essential to ensure efficient resource management and better ROA [19].

We now turn to the analysis of the moderating effect of GSS on the CSR–CFP relationship. We observed a change in curve shape after the introduction of the GSS variable. As a result, GSS moderates the relationship between CSR and CFP.

4.5. Moderating Role of Institutional Quality on CSR–CFP Nexus

Table 11. Moderating effect of institutional quality.

Variables	(1)	(2)	(3)
	ROE	ROA	Q
ROEt−1	1.002 ***
	(0.009)
ROAt−1		0.970 ***
		(0.003)
Qt−1			0.977 ***
			(0.002)
CSR	0.865 **	−1.285 ***	−1.462 ***
	(0.393)	(0.324)	(0.176)
CSR2	−0.117 **	0.177 ***	0.230 ***
	(0.058)	(0.044)	(0.027)
IQ	3.597 ***	−3.720 ***	−2.506 ***
	(0.952)	(0.781)	(0.351)
CSR X IQ	−2.431 ***	1.855 ***	1.519 ***
	(0.580)	(0.437)	(0.228)
CSR2 X IQ	0.354 ***	−0.257 ***	−0.241 ***
	(0.085)	(0.060)	(0.036)
LEV	−0.214 ***	−0.212 ***	0.241 ***
	(0.017)	(0.013)	(0.011)
Size	−0.047 ***	−0.009 ***	0.009 ***
	(0.005)	(0.002)	(0.003)
GDP	−0.025 ***	−0.031 ***	−0.017 ***
	(0.007)	(0.001)	(0.003)
Inflation	−0.112 **	0.014	0.011 ***
	(0.001)	(0.011)	(0.001)
Constant	−0.643	2.927 ***	2.455 ***
	(0.620)	(0.576)	(0.271)
Observations	846	846	846
Number of firms	141	141	141
Number of instruments	130	138	140
AR (1) (p-value)	0.017	0.003	0.000
AR (2) (p-value)	0.061	0.080	0.193
Hansen test (p-value)	0.807	0.796	0.630
Country specific effects	YES	YES	YES

***, ** denote statistical significance at the1%, 5% levels, respectively.

reports the results of the moderating effect of IQ on the CSR–CFP relationship. In terms of the direct effect of IQ on CFP, the IQ had a positive effect on ROE and ROA (columns 1 and 2). However, it had a negative effect on Tobin’s Q (column 3). A positive impact of IQ on ROE and ROA means that H&T firms operating in countries with high IQ tend to have higher profits and efficiency than H&T firms operating in countries with low IQ. This could be explained by the fact that high institutional quality fosters an environment conducive to doing business, reduces transaction costs and risks, enhances investor confidence and protection, and promotes innovation and competitiveness, and vice versa.

We turn now to the investigation of the moderating effect of IQ on the CSR–CFP relationship. We also noticed a change in curve shape after the introduction of the IQ variable, except for the Q model. Therefore, IQ moderates the relationship between CSR and CFP.

4.6. Robustness Check

We performed a test by examining the impact of each of the governance structure indicators to verify the robustness of our results. These results confirmed the expected signs (

Table 12. Robustness checks: decomposition of GSS.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	ROE				ROA				Q
	BS	BI	CEO	GD	BS	BI	CEO	GD	BS	BI	CEO	GD
L.PER	1.017 ***	1.015 ***	1.002 ***	0.697 ***	0.962 ***	0.972 ***	0.965 ***	0.765 ***	0.888 ***	0.788 ***	0.818 ***	0.916 ***
	(0.012)	(0.014)	(0.016)	(0.026)	(0.006)	(0.013)	(0.009)	(0.029)	(0.007)	(0.012)	(0.010)	(0.004)
CSR	−1.536 ***	−1.690 ***	−1.123 ***	−0.888 ***	−0.989 ***	−0.863 ***	−0.767 ***	−1.148 ***	0.816 ***	−0.322 **	−0.337 ***	−0.175 *
	(0.288)	(0.172)	(0.185)	(0.324)	(0.104)	(0.114)	(0.088)	(0.244)	(0.125)	(0.132)	(0.114)	(0.102)
CSR2	0.179 ***	0.263 ***	0.178 ***	0.1375 **	0.126 ***	0.145 ***	0.132 ***	0.194 ***	−0.132***	0.067 ***	0.064 ***	0.036 **
	(0.042)	(0.026)	(0.029)	(0.054)	(0.014)	(0.017)	(0.013)	(0.042)	(0.019)	(0.019)	(0.017)	(0.017)
IQ	−0.061 ***	−0.1650	−0.022	0.108	−0.037 ***	−0.414 ***	0.172 ***	−0.173 **	0.786 **	0.257 ***	0.191 ***	0.132 ***
	(0.008)	(0.158)	(0.151)	(0.116)	(0.005)	(0.137)	(0.058)	(0.088)	(0.003)	(0.064)	(0.054)	(0.041)
CSR X IQ	0.176 ***	0.014 **	−0.956 *	−0.011 ***	0.128 ***	0.018 ***	−0.440 *	0.898 ***	−0.630 ***	−0.016 ***	−0.012 ***	−0.864 ***
	(0.036)	(0.006)	(0.056)	(0.004)	(0.017)	(0.005)	(0.002)	(0.003)	(0.001)	(0.002)	(0.002)	(0.001)
CSR2 X IQ	−0.105 ***	−0.141 ***	0.120 ***	0.131 ***	−0.123 ***	−0.019 ***	0.323 ***	−0.111 ***	0.174 ***	0.170 ***	0.147 ***	0.941 ***
	(0.001)	(0.060)	(0.050)	(0.005)	(0.001)	(0.005)	(0.002)	(0.004)	(0.008)	(0.003)	(0.002)	(0.001)
Leverage	−0.102	−0200 ***	−0.201 ***	−0.012	−0.021 ***	−0.022 ***	−0.112 ***	−0.112 ***	0.054 ***	0.012	0.013 ***	0.017 ***
	(0.101)	(0.052)	(0.001)	(0.011)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.117)	(0.001)	(0.001)
Size	−0.018 *	−0.036 ***	−0.037 ***	−0.053 ***	−0.035 ***	−0.040 ***	−0.053 ***	−0.066 ***	0.018 ***	0.055 ***	0.048 ***	0.021 ***
	(0.010)	(0.010)	(0.008)	(0.014)	(0.003)	(0.007)	(0.005)	(0.012)	(0.004)	(0.007)	(0.006)	(0.004)
GDP	−0.072 ***	−0.068 ***	−0.037	−0.048 **	−0.057 ***	−0.048 ***	−0.057 ***	−0.025	−0.043 ***	−0.043 ***	−0.050 ***	−0.010
	(0.010)	(0.018)	(0.024)	(0.020)	(0.016)	(0.018)	(0.020)	(0.016)	(0.009)	(0.009)	(0.009)	(0.007)
Inflation	−0.010 *	−0.002	0.008	0.003	0.001	0.006	0.007	−0.007	0.020 ***	0.016 ***	0.004	0.010 ***
	(0.006)	(0.006)	(0.008)	(0.009)	(0.006)	(0.006)	(0.006)	(0.006)	(0.003)	(0.003)	(0.003)	(0.003)
Constant	15.009 **	5.548	−6.591	1.839 ***	2.255 ***	−4.652	7.147	10.333	0.501	1.847 ***	−2.026	1.133 ***
	(6.826)	(6.878)	(9.055)	(0.585)	(0.861)	(6.592)	(6.550)	(6.608)	(0.755)	(0.377)	(3.291)	(0.304)
Observations	846	846	846	846	846	846	846	846	846	846	846	846
Number of firms	141	141	141	141	141	141	141	141	141	141	141	141
Number of instruments	124	124	124	126	132	132	131	133	133	133	137	138
AR (1) (p-value)	0.020	0.018	0.018	0.017	0.003	0.003	0.003	0.002	0.001	0.000	0.000	0.000
AR (2) (p-value)	0.063	0.069	0.054	0.052	0.067	0.081	0.063	0.068	0.299	0.243	0.218	0.228
Hansen test (p-value)	0.540	0.348	0.304	0.828	0.371	0.344	0.428	0.979	0.309	0.529	0.442	0.373
Country specific effects	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES

***, ** and * denote statistical significance at the 1%, 5% and 10% levels, respectively.

).

Next, we performed an additional robustness test by examining the impact of each of the IQ on CFP. These results confirmed the expected signs (

Table 13. Robustness checks: decomposition of IQ.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)	(17)	(18)
	ROE						ROA						Q
	VA	PS	GE	RQ	RL	CC	VA	PS	GE	RQ	RL	CC	VA	PS	GE	RQ	RL	CC
PERt−1	0.956 ***	0.968 ***	0.956 ***	0.997 ***	0.982 ***	0.978 ***	0.974 ***	0.925 ***	0.965 ***	0.949 ***	0.973 ***	0.972 ***	0.986 ***	0.987 ***	0.992 ***	0.939 ***	0.933 ***	0.988 ***
	(0.017)	(0.009)	(0.015)	(0.008)	(0.007)	(0.005)	(0.005)	(0.007)	(0.008)	(0.009)	(0.006)	(0.005)	(0.001)	(0.001)	(0.003)	(0.003)	(0.003)	(0.003)
CSR	−2.334 ***	−1.507 ***	−1.319 ***	−1.096 ***	−1.102 ***	−1.085 ***	−0.395 ***	−0.660 ***	−0.852 ***	−0.659 ***	−0.432 ***	−0.551 ***	−0.234 ***	−0.180 ***	−0.400 ***	0.429 ***	0.258 **	−0.330 ***
	(0.317)	(0.205)	(0.287)	(0.209)	(0.185)	(0.170)	(0.081)	(0.143)	(0.150)	(0.190)	(0.128)	(0.105)	(0.042)	(0.049)	(0.106)	(0.058)	(0.117)	(0.079)
CSR2	0.384 ***	0.242 ***	0.197 ***	0.195 ***	0.178 ***	0.175 ***	0.069 ***	0.109 ***	0.092 ***	0.071 **	0.040 *	0.058 ***	0.046 ***	0.025 ***	0.036 *	−0.091 ***	−0.060 **	0.038 ***
	(0.047)	(0.036)	(0.045)	(0.033)	(0.031)	(0.027)	(0.013)	(0.022)	(0.024)	(0.035)	(0.023)	(0.017)	(0.006)	(0.007)	(0.018)	(0.010)	(0.024)	(0.014)
IQ	−2.334 ***	−1.507 ***	−1.319 ***	−1.096 ***	−1.101 ***	−1.085 ***	−0.395 ***	−0.660 ***	−0.852 ***	−0.659 ***	−0.432 ***	−0.551 ***	−0.234 ***	−0.180 ***	−0.400 ***	0.429 ***	0.258 **	−0.330 ***
	(0.317)	(0.205)	(0.287)	(0.209)	(0.185)	(0.170)	(0.080)	(0.143)	(0.150)	(0.190)	(0.128)	(0.105)	(0.042)	(0.049)	(0.106)	(0.058)	(0.117)	(0.078)
CSR X IQ	0.384 ***	0.242 ***	0.197 ***	0.195 ***	0.178 ***	0.175 ***	0.069 ***	0.108 ***	0.092 ***	0.071 **	0.040 *	0.058 ***	0.046 ***	0.025 ***	0.036 *	−0.094 ***	−0.060 **	0.038 ***
	(0.047)	(0.036)	(0.045)	(0.033)	(0.032)	(0.026)	(0.013)	(0.022)	(0.024)	(0.035)	(0.023)	(0.017)	(0.006)	(0.007)	(0.018)	(0.001)	(0.024)	(0.013)
CSR2 X IQ	−0.471	−0.090 **	0.222	0.039	−0.137 ***	−0.134 ***	−0.267 ***	−0.374 *	−0.531 ***	−0.459 ***	−0.436 ***	−0.421 ***	−0.010	0.04592	−0.238 ***	−0.883 ***	0.595 ***	−0.152 ***
	(0.357)	(0.040)	(0.403)	(0.052)	(0.035)	(0.040)	(0.030)	(0.202)	(0.079)	(0.134)	(0.072)	(0.038)	(0.022)	(0.03196)	(0.033)	(0.176)	(0.151)	(0.019)
Leverage	−0.311 ***	−0.217 ***	−0.227 ***	−0.207 ***	−0.298 ***	−0.241 ***	−0.223 ***	−0.125 ***	−0.233 ***	−0.274 ***	−0.253 ***	−0.227 ***	0.125 ***	0.121 ***	0.119 ***	0.117 ***	0.123 ***	0.119 ***
	(0.017)	(0.011)	(0.014)	(0.012)	(0.078)	(0.023)	(0.019)	(0.013)	(0.016)	(0.019)	(0.011)	(0.018)	(0.011)	(0.012)	(0.014)	(0.013)	(0.011)	(0.011)
Size	−0.064 ***	−0.048 ***	−0.041 ***	−0.058 ***	−0.042 ***	−0.043 ***	−0.027 ***	−0.043 ***	−0.034 ***	−0.040 ***	−0.028 ***	−0.041 ***	−0.011 ***	−0.003	0.000	0.017 ***	0.013 ***	0.003
	(0.011)	(0.005)	(0.009)	(0.006)	(0.004)	(0.004)	(0.004)	(0.006)	(0.006)	(0.008)	(0.006)	(0.004)	(0.002)	(0.003)	(0.003)	(0.003)	(0.002)	(0.002)
GDP	−0.040 ***	−0.015 ***	−0.050 ***	−0.033 ***	−0.015 ***	−0.012 ***	−0.033 ***	−0.022	−0.019 ***	−0.069 ***	−0.022 ***	−0.016 ***	−0.017 ***	0.003	−0.003 **	−0.007	−0.008	−0.006 ***
	(0.015)	(0.004)	(0.018)	(0.003)	(0.005)	(0.004)	(0.004)	(0.017)	(0.005)	(0.011)	(0.004)	(0.003)	(0.002)	(0.002)	(0.002)	(0.009)	(0.010)	(0.002)
Inflation	0.050	−0.101 ***	0.061	−0.078 ***	−0.013 ***	−0.017 ***	−0.024	0.081 *	0.023	0.072	0.043	0.082	0.417 ***	0.409 ***	0.392 ***	−0.411 ***	0.397	0.395 ***
	(0.041)	(0.001)	(0.005)	(0.004)	(0.007)	(0.001)	(0.031)	(0.047)	(0.242)	(0.611)	(0.234)	(0.147)	(0.007)	(0.004)	(0.003)	(0.023)	(0.304)	(0.007)
Constant	−2.656	3.006 ***	−3.553	2.227 ***	2.313 ***	2.259 ***	1.125 ***	−6.900	2.581 ***	2.111 ***	1.668 ***	1.926 ***	0.467 ***	0.305 ***	0.992 ***	3.977 **	−1.116	0.685 ***
	(5.203)	(0.275)	(4.923)	(0.324)	(0.286)	(0.276)	(0.126)	(5.066)	(0.288)	(0.533)	(0.210)	(0.182)	(0.083)	(0.085)	(0.148)	(1.941)	(1.326)	(0.110)
Observations	846	846	846	846	846	846	846	846	846	846	846	846	846	846	846	846	846	846
Number of firms	141	141	141	141	141	141	141	141	141	141	141	141	141	141	141	141	141	141
Number of instruments	125	126	125	126	128	128	137	136	132	131	132	132	137	137	137	139	139	137
AR (1) (p-value)	0.016	0.017	0.017	0.017	0.016	0.016	0.003	0.003	0.003	0.003	0.003	0.003	0.000	0.000	0.001	0.000	0.000	0.001
AR (2) (p-value)	0.068	0.061	0.062	0.063	0.061	0.061	0.086	0.087	0.082	0.071	0.082	0.081	0.195	0.160	0.163	0.056	0.124	0.175
Hansen test (p-value)	0.249	0.731	0.241	0.829	0.611	0.628	0.613	0.448	0.608	0.275	0.660	0.635	0.526	0.688	0.648	0.791	0.779	0.626
Country specific effects	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES

***, ** and * denote statistical significance at the 1%, 5% and 10% levels, respectively.

).

4.7. Discussion

The aim of this research was to provide a theoretical clarification and empirical evaluation of the impact of CSR on CFP to address the inconsistent findings in existing research. The literature indicates a relationship between CSR and CFP [8,28]. However, these studies ignore the dynamic nature of CFP. In addition, other studies have investigated the impact of governance practices on CFP [60,61]. However, since CSR and governance practices are two distinct methodologies that share the same goals of transparency and trust among all stakeholders, tensions between them may emerge. We examined the nonlinear association between CSR and CFP, as well as the moderating impact of GP on the CSR–CFP nexus. For this purpose, we employed the system GMM using data covering the period 2012–2018. Our findings offer empirical evidence supporting H1. According to this hypothesis, CSR has a nonlinear effect on CFP. In addition, these results provide empirical evidence in support of H2 and H3, that GP (GS and IQ) moderates the CSR–CFP nexus.

Moreover, the result revealed that the relationship between CSR and CFP has two shapes of curves (U-shaped and inverted U-shaped) depending on the measurement of CFP. As for the ROE model, we observed that the curve had two parts (Figure 1): in the left part, ROE decreased as CSR increased, while in the right part, ROE increased exponentially as CSR increased beyond an extreme point. This means that higher CSR levels have different effects on ROE in H&T firms depending on the position on the curve. The coefficients of CSR switched from negative to positive, showing that higher CSR levels led to lower ROE for H&T firms. There is an optimal CSR level of 2.665, which is the point where higher CSR levels increase ROE for H&T firms. Indeed, the negative effect of CSR on ROE could be assigned to the presence of excessive costs and the manager–shareholder relationship. Specifically, involvement in CSR activities seems to be very costly for managers of H&T firms at the beginning. To avoid these costs, managers first engage in timely investment projects, thus supporting the creation of shareholder value. This strategy allows the interests between managers and shareholders to be reconciled and for the agency costs to be minimized. This outcome proves the validity of the agency theory and the acceptance of hypothesis H1(b). However, above this extreme point (2.665), an increasing CSR improved ROE. Specifically, once managers have satisfied the expectations of their shareholders, they are able to invest in CSR activities and create value for their stakeholders. They also help them to enjoy high economic benefits. The positive link between CSR and ROE was confirmed by [19]. This finding proves the predictions of stakeholder theory; hence, hypothesis H1(a) is accepted. In summary, our results align with those found by [8,26], where they confirmed the existence of a U-shaped relationship between CSR and ROE.

Turning to the ROA model, however, the coefficients of CSR switched from positive to negative, showing that higher CSR levels lead to higher ROA for H&T firms. There is an optimal CSR level of 3.680, which is the point where higher CSR levels reduce ROA for H&T firms (see Figure 2). This outcome indicates that the relationship between the CSR and ROA has become inverted-U-shaped. Specifically, the positive effect of CSR on ROA suggests that H&T firms become involved in CSR activities to enhance their ROA. As a result, these firms adopt socially responsible strategies, thus supporting and meeting the expectations of their stakeholders. This leads them to further leverage the competitive (economic) advantage they hold and to use their invested funds efficiently. This strategy then allows them to increase their performance. This result is like that found by [20]. This outcome is reliable according to the predictions of stakeholder theory, and hence hypothesis H1(a) is accepted. However, the negative effect of CSR on ROA reveals that H&T firms that invest heavily in CSR activities face valuable losses (bankruptcy risks). They are confronted with taking on excessive transaction costs, resulting in deteriorating performance. Specifically, engagement in CSR activities hinders H&T firms from using their assets in an efficient manner. This is likely to prevent the firms’ managers from serving the economic interests of their stakeholders (shareholders/investors). The failure to satisfy these different stakeholders is too costly for hotel companies. It generates high costs that are not compensated by other stakeholders (customers, consumers, communities, etc.). This outcome confirms the predictions of the agency theory from which the H1(b) hypothesis was accepted. These results are like those found by [28].

Finally, as for the Tobin’s Q model, the results were like the ROE model results, showing that the association between the CSR and ROA has become U-shaped (Figure 3). Specifically, the coefficients of CSR switched from negative to positive. There was then an optimal CSR threshold, estimated at 2.554, below which the commitment of H&T firms in CSR activities does not allow them to maximize their value. However, above this optimal point, the engagement of these firms in CSR activities helps them maximize their value. The negative association between CSR and Tobin’s Q indicates that managers of H&T firms do not rely on CSR activities to enhance shareholder wealth. They use other means of disclosure (e.g., debt and/or equity holdings) to create new investment opportunities and maximize the value of their firms. However, above this threshold of 2.554, managers invest heavily in CSR activities to see the expectations of all their stakeholders, including the creation of shareholder value. These results are in contrast with those found by [19]. They encourage the predictions of agency theory and stakeholder theory, from which both hypotheses H1(b) and H1(a) are accepted.

Generally, the nonlinear relationship between CSR and H&T CFP could be qualified to the planning horizon (i.e., time) and stakeholder behaviors. Specifically, investment in CSR activities requires significant funding. It is a long-term investment that is financed by long-term funding sources. It follows from this that investment in CSR activities requires a lot of time to finance. Similarly, stakeholders behave rationally when they know the actual value of companies’ CSR practices. That is why they take more time to respond reasonably and therefore achieve their objectives. In this context, [19] showed that the optimal level of CSR, set by managers, leads to a change in the behavior of stakeholders. Based on a certain optimal level of CSR, stakeholders govern themselves in a responsible way towards their companies. This change is expected to alter the CSR–CFP nexus from negative to positive, thus revealing its U-shape. This result aligns with those found by [8,26,27,67].

Furthermore, the findings show that the GP (GSS and IQ) moderates the CSR–CFP nexus. As for GSS effect, we observed that the shape curves were changed. Regarding the ROE model, the relationship between CSR and ROE remained U-shaped (see Figure 4). When H&T firms first start adopting CSR practices, they may face higher initial costs, such as when investing in sustainable initiatives, complying with ethical standards, and implementing social policies. These costs can temporarily reduce ROE. However, as these CSR practices bear fruit, they can enhance the firm’s reputation, strengthen customer loyalty, attract socially responsible investors, reduce regulatory risks, and improve operational efficiency. This can lead to an increase in ROE over the long term. GSS plays a crucial role in how these effects manifest themselves. A strong GSS can promote effective CSR implementation and help mitigate upfront costs, while poor governance can hamper these efforts.

As for ROA model, the association between CSR and ROA became U-shaped (see Figure 5). When H&T firms first start investing in CSR practices, they may incur high initial costs. For example, they may need to invest in more environmentally friendly technologies, improve working conditions or contribute to social projects. These costs can temporarily reduce ROA, as they affect profitability relative to assets. However, as these CSR practices begin to bear fruit, they can have long-term benefits. For example, they can improve the firm’s reputation, attract more loyal customers, reduce regulatory risks, and increase operational efficiency. These benefits can contribute to an increase in ROA over the medium and long term. The firm’s GSS plays a moderating role in influencing how CSR actions are implemented. Good governance can enable the effective management of the initial costs and future benefits of CSR, while poor governance can hinder these efforts.

As for Tobin’s Q model, the relationship between CSR and Q developed an inverted U-shape (see Figure 6). Once H&T firms start implementing CSR initiatives, they can incur significant initial costs. For instance, investments in environmentally friendly technologies, improvements in working conditions or contributions to community projects can temporarily reduce firm value. Over time, however, CSR efforts can enhance corporate reputation, attract sustainability-conscious travelers, reduce regulatory risks, and strengthen customer loyalty. These long-term benefits can increase the firm’s value. The GSS in the tourism industry plays a crucial moderating role, optimizing the management of the initial costs of CSR and maximizing long-term benefits, by promoting better decision-making and a more harmonious relationship with stakeholders. Inadequate governance, on the other hand, can hamper these efforts and prolong the period of high initial costs.

Regarding the moderating effect of IQ, we found that shape curves were also modified. As in the ROE model, the relationship between CSR and ROE developed an inverted U-shape (see, Figure 7). There are several possible explanations for this result. Firstly, when institutional quality is low, H&T firms have little incentive to adopt CSR practices, as they face an unstable, uncertain and corrupt environment. As a result, they may prioritize short-term profit over sustainable development. Secondly, when institutional quality increases, H&T firms are more encouraged to adopt CSR practices, as they benefit from a more favorable, transparent and regulated environment. In this way, they can take advantage of the opportunities offered by the growing sustainable H&T industry, which is responding to growing social demand. Finally, when institutional quality reaches an optimal level, tourism companies can find the right balance between economic and societal requirements. They can then adopt CSR practices that maximize tourism ROE, while respecting ethical and environmental standards.

In addition, the result of the ROA model shows that the relationship between CSR and ROA is U-shaped (see Figure 8). This result can be explained by the fact that in countries with high IQ, companies are subject to strict CSR regulations. As a result, H&T firms tend to adopt stronger CSR practices and invest more in responsible initiatives, which can improve their ROA. However, in countries where IQ is low, CSR regulations and monitoring are often less stringent. As a result, H&T firms may have less incentive to invest in CSR, which may have a limited impact on their ROA.

However, the results of Tobin’s Q model show that the relationship has not been changed and remains U-shaped (Figure 9). This indicates that institutional quality has not influenced the relationship between CSR and Q (firm value). This result could be explained by the fact that in the H&T sector, firms may have a certain autonomy to implement CSR initiatives, irrespective of institutional quality. More specifically, H&T firms uniformly adopt strong and consistent CSR practices, regardless of the institutional context in which they operate. For example, if a company is strongly motivated by reputational considerations, or if its customers and investors demand high standards of social responsibility, it may be less influenced by the institutional quality of its operating environment.

This study has several limitations that could be considered in future research. Firstly, it is a question of sample and period. This study focuses on an international sample. In this case, it is important for researchers to subdivide the sample into developed and developing economies. In addition, the period used is short and characterized by a degree of financial stability. It is therefore important for the researchers to extend the study period to take account of the COVID-19 period. Secondly, the study looked at the moderating effect of governance practices (GSS and IQ). As a result, it is important to examine the moderating effect of other factors, namely culture, and financial and non-financial risks.

5. Conclusions

The purpose of the study is to examine the nonlinear relationship between CSR and CFP, as well as the moderating impact of governance practices (GSS and IQ) on this relationship. The paper selected a sample of 141 worldwide H&T firms covering the period 2010–2018. This paper used two types of CFP measure, namely the accounting measure (ROE and ROA) and the market measures (firm value (Tobin’s Q)). For this purpose, system GMM was performed to consider endogeneity and heterogeneity problems to achieve the objective of this study. To control for the impact of other factors that may influence CFP, four control variables (leverage, financial size, inflation, GDP) were used.

The findings clearly show that the relationship between CSR and CFP is curvilinear. However, the shape of the relationship depends essentially on the measurement of CFP. Specifically, for the accounting measures (ROE and ROA), we observed that the curve changed its shape (U-shaped and inverted U-shaped, respectively). These results could be attributed to stakeholder behavior. For the market measure (Tobin’s Q), the relationship between CSR and CFP was U-shaped, indicating that stakeholders are risk-takers.

Furthermore, the results show that governance practices (GSS and IQ) moderate the relationship between CSR and CFP. This is reflected in the extent of governance in the proper conduct management of CSR practices and in ensuring the sustainability of H&T firms.

5.1. Implications

Our results have important implications for managers and policy makers, in addition to enriching the recent literature on CFP determination in the H&T sector. Specifically, in the short term, managers of worldwide H&T firms are advised to engage in CSR activities related to internal stakeholders (shareholders, employees, etc.), spending funds on improving working conditions for employees. This is likely to create value for the stakeholders within the firms and therefore improve their operational efficiency. Similarly, they must also rely on internal financing (self-financing) to fund their investment projects, including investment in CSR activities. In fact, optimizing CSR policies, implementing the right governance practices and making the right financing decisions help H&T firms minimize their costs and improve their economic and financial returns. These different strategies safeguard them against operational and financial risks. In the long run, H&T companies should engage in social and environmental CSR activities, thus helping to promote their reputation. To do so, these firms can leverage external financing (especially long-term debt) to maximize the market value of their businesses and create value for their external stakeholders. Managers are also advised to consider changing economic conditions (macroeconomic stability, economic growth, etc.) before making investment decisions.

Furthermore, policymakers should promote the importance of effective governance mechanisms within H&T companies. This includes encouraging transparent and accountable management practices, as well as the appointment of independent directors who can help oversee CSR initiatives. In addition, they should advise H&T companies to strike a balance in their CSR activities. Although CSR is important, excessive, or untargeted spending on CSR does not necessarily lead to improved business performance. Therefore, guidelines on the optimal level of investment in CSR need to be defined. Finally, they should improve the overall quality of institutions, including legal and regulatory frameworks, in the H&T sector. This can be achieved by enhancing the rule of law, reducing corruption, and ensuring effective contract enforcement. Strong institutions provide a conducive environment for CSR initiatives and can positively impact CFP.

5.2. Limitations and Future Research

This study has several limitations that open opportunities for future research. Firstly, it should be noted that this study is based on international samples, which raises questions about the generalizability of the results. In this context, it would make sense for researchers to carry out more in-depth analyses, distinguishing between developed and developing economies. In addition, the study period was relatively short and was marked by a degree of financial stability. Therefore, researchers would benefit from extending the study period to incorporate the impacts of the COVID-19 pandemic, which may have had significant effects on the relationship between CSR and CFP. Secondly, this study focused on the moderating impact of governance practices (GSS and IQ) on CSR–CFP. For a more complete understanding of this complex relationship, researchers also explore other potential moderating factors, such as organizational culture or financial and non-financial risks. This would provide a better understanding of the subtle interactions that influence the way CSR affects CFP, while considering the diversity of contexts and factors at work. In sum, these suggestions offer promising avenues for future studies aimed at deepening our understanding of these complex dynamics. Furthermore, the ESG score is used to characterize corporate social responsibility, which is the overall measure of CSR. Researchers can use the other pillars (environmental, social and governance) to specify the effect of each pillar on CFP. Finally, the cross-section dependence test is an important diagnostic for panel data analysis [102,103,104]. In this case, it is recommended that researchers consider the results of cross-sectional dependence problems in future research.