The moderating role of a corporate life cycle with the impact of economic value-added on corporate social responsibility: Evidence from China's listed companies

Emerging Markets Review 55 (2023) 101021 Available online 16 March 2023 1566-0141/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). The moderating role of a corporate life cycle with the impact of economic value-added on corporate social responsibility: Evidence from China’s listed companies Xiaojuan Wu a , c , Dana Dluhoˇ sov´ a b , Zdenˇ ek Zmeˇ skal b , * a Hebei GEO University, School of Management, Department of Accounting, Shijiazhuang 050031, China b VSB-Technical University of Ostrava, Faculty of Economics, Department of Finance, Sokolska 33, 702 00 Ostrava, Czech Republic c VSB-Technical University of Ostrava, Faculty of Economics, Department of Business Administration, Sokolska 33, 702 00 Ostrava, Czech Republic ARTICLE INFO Original content: “Xiaojuan Wu, Dana Dluhoˇ sov´ a and Zdenˇ ek Zmeˇ skal, Dataset: The moderating role of a corporate life cycle on corporate social responsibility China companies” (Original data) JEL: D22 D91 L25 M14 O1 Keywords: Corporate social responsibility participation index Corporate financial performance Corporate life cycle stage Economic value added Analytic network process Fractional regression ABSTRACT The moderating role of corporate life cycle stages with the impact of relative economic value added (EVA) indicator on corporate social responsibility participation index (CSRPI). Chinese Ashare listed companies are investigated. The CSRPI weights are calculated by Analytic Network Process. Fractional regression with interaction is used. The corporate life cycle stages moderate the relationship between relative EVA measure and CSRPI. Surprisingly, this impact is confirmed for companies at non-mature stages, but not for mature companies. Model type, weights and corporate life cycle robustness were confirmed. The findings have implications for stakeholders in understanding companies’ social behaviour in the Chinese market. 1. Introduction Corporate Social Responsibility (CSR) has become an important issue facing modern companies (Jiao et al., 2021). The phenomena Abbreviations: EVA, economic value added; CSR, corporate social responsibility; CRSPI, corporate social responsibility participation index; CLC, corporate life cycle; CFP, corporate financial performance; ANP, analytic network process; CSMAR, China Stock Market and Accounting Research Databases. ** Corresponding author. E-mail addresses: [email protected], [email protected] (X. Wu), [email protected] (D. Dluhoˇ sov´ a), [email protected] (Z. Zmeˇ skal). Contents lists available at ScienceDirect Emerging Markets Review journal homepage: www.elsevier.com/locate/emr https://doi.org/10.1016/j.ememar.2023.101021 Received 30 June 2022; Received in revised form 23 February 2023; Accepted 10 March 2023 Emerging Markets Review 55 (2023) 101021 2 are deeply analysed and investigated. Jo and Harjoto (2011) pointed out that CSR engagement can positively influence a company’s value. The benefits of CSR are also associated with its impact on company reputation (Fatma et al., 2015; Fombrun, 2005), company innovation (Chkir et al., 2021), consumer loyalty (Park et al., 2017; Islam et al., 2021), and risk reduction (Jo and Na, 2012). These advantages brought about by CSR participation have attracted more and more companies to invest in CSR activities. According to the Global Sustainable Investment Review (2020) released by the Global Sustainable Investment Alliance (GSIA), global sustainable investment has reached US$35.30 trillion in 2020, an increase of 15% in two years (2018–2020). Corporate Financial Performance (CFP), as one of the crucial indicators measuring the results of corporate operations, is also one of the indispensable factors in determining corporate activities. The direction of causality between CFP and CSR has always been a topic that researchers have discussed. But even if the same method (meta-analysis) is used to examine their relationship, different researchers draw different conclusions. For example, Orlitzky et al. (2003) found that CSR seems more relevant to accounting-based CFP indicators than market-based indicators; Wang et al. (2016) found that subsequent CFP is associated with prior CSR, while the reverse direction is not supported. Furthermore, Endrikat et al. (2014) discovered a positive and partially bidirectional relationship between corporate environmental performance (CEP) and CFP, and Hang et al. (2019) suggest that CFP can increase CEP in the short-run (one year). However, the effects disappear in the long run (after more than one year), and in turn, increasing CEP has no short-term impact on a CFP, while a company benefits significantly in the long-term period. In the face of inconsistent study results, some scholars claim that there is still a lot of research to be done before the relationship between CSR and CFP is fully understood (Margolis and Walsh, 2003; Alshehhi et al., 2018). In particular, Margolis and Walsh (2003) have stressed the importance of developing models incorporating omitted variables, testing mediating mechanisms and contextual conditions, and establishing causal links between CFP and CSR in theory. Alshehhi et al. (2018) also emphasise the importance of moderators in understanding this relationship. Some scholars have started to take steps along these lines. Surroca et al. (2010) discovered an indirect relationship between CSR and CFP that relies on the mediating effect of a company’s intangible resources but no direct relationship between them. Wang et al. (2016) examined the moderating effect of environmental context on the link between CSR and CFP based on literature analysis. They observed that companies from advanced economies had stronger CSR-CFP relationships than companies from developing economies. According to a US sample, Cho and Lee (2017) found that CSR is positively associated with CFP with efficient managers. Li et al. (2021) explore whether the religious atmosphere moderates the impact of corporate philanthropy on CFP based on a Chinese sample. Ahmad et al. (2021) discovered that CSR initiatives positively influence CEP through mediator employee pro-environmental behaviour. Most of the existing literature focuses on studying how moderators or mediators influence the result of CSR on CFP. Few studies are concerned with the reverse effect, that is, how moderators or mediators affect the impact of CFP on CSR. According to the “dynamic resource-based view”, companies have different capacities and resource over time (Helfat and Peteraf, 2003), which is the foundation of different strategies the companies adopt (Wernerfelt, 1984) depending on the corporate life cycle (CLC). That being so, companies at different life cycle stages are associated with varying levels of resources, which are the basis for forming CSR strategies and behaviours (Hasan and Habib, 2017). Mature companies have sufficient resources and obvious competitive advantages, so the impact of mature companies’ CFP on CSR engagement level is more substantial than that of companies in other stages of CLC. As the largest developing economy in the world, China has prominent environmental problems in economic development (Wu and Hąbek, 2021). Major pollutants such as sulfur dioxide, industrial soot, wastewater and solid waste emissions are the most concern related to urbanisation development (Hao et al., 2020), and China’s industrial structure is dominated by heavy industry (Liu and Lin, 2019). CFP is one of the important sources of internal resources for companies. Whether CFP plays a role in Chinese companies’ CSR participation and whether this role is consistent in each stage of CLC is a topic worthy of study. Therefore, based on the dynamic resource-based view and slack resource theory, the aim of the paper is to investigate how the moderator of CLC stages affects the impact of CFP on CSR in Chinese listed companies, including CLC, weights and model type robustness. CSR is defined as a participation level expressed by the corporate social responsibility participation index (CSRPI) calculated by the weighted arithmetic mean of dummy criteria with weights computed using the Analytic Network Process (ANP). Because the CSRPI values belong to a discrete, closed interval [0,1] and non-Gaussian distribution, the fractional regression with a moderating effect is applied. Stemming from companies’ ability to create slack resources, CFP is measured by the relative economic value added (EVA) indicator, which, taking into account all capital costs, is more complex and can improve CFP measuring analysis. For comparison, the ROE measure is also investigated. Both measures stem from the slack resource theory. Robustness tests verify the results on the CRSPI weights composition, model type, and CLC categorisation. All A-share listed companies except financial companies in the Chinese stock exchange are taken into account in the analysed sample. Thus, the paper novelty to explore whether and how the CLC stage in the Chinese economy moderates the impact of CFP, measured by relative EVA, on CSRPI, measured by ANP, employing a fractional regression model. The paper is structured as follows: Section 2 includes the theory and existing literature on the effect of CFP on CSR and how the CLC stages moderate this effect, which facilitates the development of the hypotheses for this paper. Section 3 presents the research sample, measurement of variables, research design and fractional regression model formulations. Section 4 subsequently shows the research results and robustness tests. Finally, Section 5 includes a summary of the results and conclusions. 2. Literature review and hypotheses development 2.1. The impact of corporate financial performance on corporate social responsibility CSR usually represents a relatively high area of management discretion (Carroll, 1979, 1991). The implementation of CSR-related X. Wu et al. Emerging Markets Review 55 (2023) 101021 3 projects may be particularly sensitive to the existence of slack resources (McGuire et al., 1988). Waddock and Graves (1997) put forward the slack resource theory claiming that better financial performance may cause available slack (financial and other) resources. It allows companies to invest in CSR-related projects, such as community relations and employee benefits, philanthropic donations or environmental protection. If slack resources are available, allocating these resources to the social domain produces better social performance. Therefore, better financial performance is a more effective predictor of better corporate social performance. Shahzad et al. (2016) further divide the slack resources into financial slack, human resources slack, and innovational slack, and explore how they affect CSR, respectively. Although the empirical evidence for the effect of CFP on CSR is not unique, much extant literature supports the positive relationship. Clarkson et al. (2011) used the four most polluting industries in the USA as samples and found that the company’s previous better financial resources led to improved environmental performance in the subsequent period. Based on Indonesian companies, Swandari and Sadikin (2016) concluded that profitability significantly influences CSR because “companies with high profits have the flexibility fund to implement CSR programs”. Giannarakis (2014) took a sample consisting of 100 companies from the Fortune 500 list for 2011 and discovered that profitability is positively associated with the extent of CSR disclosure. The same result is obtained from the Chinese sample studied, e. g., by Li and Zhang (2010) and Wu et al. (2021). Based on the aforementioned arguments, the following hypothesis is stated. Hypothesis (H1). : CFP positively impacts CSR participation level for all sample companies. 2.2. The moderating role of corporate life cycle stages The theory of CLC is derived from the scientific literature concerning the organisation and dates back several decades (Penrose, 1959). Existing models differ in the number of stages and activities within each stage (Jawahar and McLaughlin, 2001), but some commonalities exist (Lester et al., 2003). The four-stage model used in this paper is consistent with the main literature. So, the life cycle of a typical organisation comprises “four identifiable but overlapping” stages: introduction, growth, maturity, and decline. Next, in light of the dynamic resource-based view, the role of the CLC stage in moderating the relationship between CFP and CSR is analysed. The “dynamic resource-based view” of the company articulates “general patterns and paths in the evolution of organisational capabilities over time” (Helfat and Peteraf, 2003). This resource-based view posits that companies should identify and cultivate valuable, rare, inimitable and irreplaceable resources (Chaharbaghi et al., 1999). So, they are the basis of the corporate strategy (Wernerfelt, 1984) and the internal source of sustainable competitive advantage (Barney, 1991). The dynamic resource-based theory explains corporate capability’s founding, development, and maturity from the sources of heterogeneity in organisational capabilities and how the resources and capabilities of competitive advantage evolve (Helfat and Peteraf, 2003). Companies in different life cycle stages are linked with varying levels of resources that shape their CSR behaviour (Hasan and Habib, 2017). Companies in the introduction stage of the life cycle are usually small in scale, lack a stable customer base, assume “liability of newness” (Stinchcombe, 1965), and have high initial exit risks. In the early life cycle stages, companies highly evaluate market share gains and capital capabilities (Shahzad et al., 2019). They cannot afford investments related to CSR activities, and their future sustainability is uncertain (Park, 2021). Once the market recognises products, sales soar while companies enter the growth stage. However, facing fierce market competition, growth companies have to focus on infrastructure (Park, 2021), research and development (R&D), and advertising to distinguish their products from their competitors (Shahzad et al., 2019) to achieve high-profit margins to support further development. Companies in the decline stages usually have a scarcity of resources and new plans (Shahzad et al., 2019). It is difficult for them to find any further growth opportunities in the market, leading to a decline in their market share, a deterioration in profitability, an increase in debt, and a decrease in liquidity (Miller and Friesen, 1984). These companies pay more extensive attention to survival strategies. If companies with such a weak financial performance invest in CSR, it is likely to threaten shareholder value (Hasan and Habib, 2017). Therefore, the limited capacity and resource base restrict companies from using scarce resources for CSR projects in the stages mentioned above, thus significantly reducing their CSR participation. On the other hand, mature companies have a well-established customer base as well as stable and predictable performance and cash flows (Jiraporn and Withisuphakorn, 2016). In the mature stage, the effectiveness of companies is greatly improved, but the innovation ability is visibly reduced (Miller and Friesen, 1984). Faced with threats from competitors, mature companies can choose a strategy to build an outstanding reputation and public recognition to distinguish themselves from other companies (McWilliams et al., 2002). And more participation in CSR activities is a wise move to achieve this goal (Fombrun, 2005; Minor and Morgan, 2011). Moreover, the expertise and capabilities generated by organisational maturity enable these companies to make more meaningful CSR contributions. Companies with larger operating scales can allocate and use their resources more effectively to provide specialised CSR initiatives without incurring high additional costs (Udayasankar, 2008). Based on the above analysis, a good resource base, super resource integration ability, and strong demand for unique strategies enable mature companies to be able and willing to participate in more CSR-related activities. In light of the above discussion, the following hypothesis is stated. Hypothesis (H2). : The positive effect of CFP on CSR participation is more potent for companies in the mature stage than for companies in other stages of CLC. X. Wu et al. Emerging Markets Review 55 (2023) 101021 4 3. Methodology applied 3.1. Sample data Chinese A-share listed companies are considered the initial sample for 2018 and 2019 years. All the data involved in this paper is obtained from the China Stock Market and Accounting Research Database (CSMAR). The sample is filtered as follows: (1) financial companies are excluded because their accounting information is different from non-financial companies; (2) company observations yearly with missing variables are dropped; (3) the observations with outliers of the independent variable are excluded. However, due to research purposes, the companies labelled as special treatment (ST or ST*) by the China Securities Regulatory Commission (CSRC) are not dropped like e. g. in the paper (Wang and Lu, 2021). This is because such ST companies’ abnormal financial conditions usually indicate the declination stage of the life cycle. After filtration, the final sample comprises 6730 companies (3328 companies in 2018 and 3402 companies in 2019) observed, and their distribution across CLC stages is shown in Table 1. 3.2. Dependent variable – corporate social responsibility participation index The CSR content disclosed in the CSR report and annual statements collected by CSMAR constitutes the basis for judging the degree of CSR participation of the sample companies. Drawing on the existing literature (Sial et al., 2018; Wang and Lu, 2021), the dummy variables for each criterion according to whether the company discloses relevant information are used, whereas number one means disclosure and zero non-disclosure. Therefore, the corporate social responsibility participation index (CSRPI) is defined and used. Some studies (Gulzar et al., 2019; Sial et al., 2018) adopt the simple arithmetic mean method to calculate the CSR index. In the method, all criteria are considered to have the same importance, so they are given the same weight. However, the relations among criteria are often complex and non-linear; therefore, the Analytic Network Process (ANP) method is applied, see e. g., Huang et al. (2005), Tsai et al. (2010), Okan et al. (2015), Yuan and Wu (2010), and Poplawska et al. (2015). Because of the different importance of these criteria in practice, the calculation method of the CSRPI to better measure the level of participation in CSR of the sample companies is computed by the weighted arithmetic mean and with weights determined by the ANP method. So, the compensation effect of criteria is supposed, CSRPIit =∑ j wj·xitj (1) here w j is the criterion weight, x itj is the criterion dummy value. CSMAR classifies the contents of CSR disclosure into eleven criteria, i.e., protection of rights and interests of stakeholders (shareholders, creditors, employees, suppliers, customers), environment and sustainable development, public relations and social welfare, CSR system construction and improvement, working conditions, CSR defect disclosure, and CSR report certification. The eleven criteria are categorised into three groups based on their content–stakeholder protection, environmental and social responsibility, and the overall evaluation of CSR participation. Specifically, the financial stakeholders (shareholders and creditors) and business stakeholders (employees, customers, and suppliers) (Girerd-Potin et al., 2014) are combined into the stakeholder protection group. The environmental and social responsibility group includes environment and sustainable development, public relations, and social welfare. The remaining four criteria belong to the overall evaluation of the CSR participation group. See Table 2 for details of the CSR criteria and their groups. The network relationship between a goal, groups and criteria was proposed, as shown in Fig. 1. The relations among groups are non-linear, with backward and cyclical ones. Thus, the ANP method is used to compute criteria weights. The first step of the ANP is creating local weights related to given criteria or goals. Weights are calculated through pairwise criteria comparisons, making a Saaty matrix (Saaty, 1980, 1996). The next step is the construction of the initial supermatrix W by inserting local weights into it. Furthermore, the weighted supermatrix W is derived by transforming the initial supermatrix column sums to the unity, ensuring the sum of these probabilities of all states is equal to 1. The last step for a supermatrix without cyclicity is to raise the weighted supermatrix to limit powers. Formally, W∝=lim j→∝Wj, where W∝ is a limiting supermatrix, Wj is a weighted supermatrix powered jtimes, and Wj=Π i=[1;j]Wi. The final supermatrix is W=W∝. The input supermatrix with local weights calculated through the Saaty matrix is presented in Appendix Fig. A1. The procedure described led to the calculation of the limiting supermatrix, see Appendix Fig. A2, where in the first column are global weights of criteria. The weight results are also shown in the last column of Table 2. It shows the difference in the importance of criteria. It is worth Table 1 Sample distribution across CLC. CLC 2018 2019 Total Introduction 333 232 565 Growth 900 861 1761 Mature 1740 2017 3757 Decline 355 292 647 Total 3328 3402 6730 X. Wu et al. Emerging Markets Review 55 (2023) 101021 5 emphasising that in order to highlight that the environmental criterion is the most important of all CSR criteria in the current Chinese context, its weights are set to be the highest among all criteria. Finally, the CSRPI of companies is obtained by calculating the weighted average method due to (1) and is in the range [0, 1]. 3.3. Independent variable – relative EVA measure The paper applies the relative Economic Value Added (EVA) as the independent variable to measure CFP. EVA combines the familiar concept of residual income and the principle of modern corporate finance. In addition, the capital should create value for shareholders (Dierks and Patel, 1997). Many papers regard the topic, see e. g., (Dluhoˇ sov´ a, 2004; Ehrbar, 1998; Grant, 1997; Rappaport, 1986; Stewart, 1991; Young et al., 2000). It is important for the owners that the spread be as large as possible and positive. Investment in the company brings more than an opportunity cost in this case. The value of relative EVA is independent of capital or equity level, and the relative company’s performance is measured. Given that all capital costs are considered in relative EVA, the measure is better than CFP measures like ROE (Return on Equity), ROA (Return on Assets), and ROS (Return on Sales) in measuring the company’s ability to create slack resources. The relative EVA value depicted as EVA r comes from the CSMAR database and is calculated in this way: EVA r ≡EVA/C =NOPAT/C −WACC, where NOPAT =NP +(int +R & D −NI) ⋅ (1 −t), WACC =REE A+RD·(1−t)D A, C =E +D − (int +CL +CP), and C is the total capital, NOPAT is net operating profit after tax, WACC is the weighted cost of capital, NP is net profit, int is the interest, R & D is an adjustment for R&D expenses, NI is non-operating income, t is the income tax rate, R E is the cost of equity, R D is the cost of debt, D is the value of debt, E is the equity value, A is the asset value, CL is current liabilities, CP is construction in progress. 3.4. Moderating variable – corporate life cycle In the existing CLC literature, various measures are used to represent the CLC stages, for example, firm age (Jiraporn and Withisuphakorn, 2016), growth (ur Rehman et al., 2016; Lee and Choi, 2018), size (Porter, 2004), and the retained earnings to total assets ratio (DeAngelo et al., 2006). These methods potentially treat CLC as a sequential pattern. However, some studies argue that CLC does Table 2 roups and criteria of CSRPI decompositions including weights. Symbol Groups and criteria Weights A Stakeholders protection A1 The protection of shareholders’ rights and interests 0.1011 A2 The protection of creditors’ rights and interests 0.0348 A3 The protection of employees’ rights and interests 0.1011 A4 The protection of suppliers’ rights and interests 0.0569 A5 The protection of the rights and interests of customers and consumers 0.1763 B Environmental and social responsibility B1 The environment and sustainable development 0.2833 B2 Public relations and social welfare undertakings 0.0708 C The overall evaluation of CSR participation C1 Implementation of CSR system construction and optimisation 0.0818 C2 Safety production content 0.0350 C3 The company’s deficiencies 0.0350 C4 Verification of CSR report by the third agency 0.0241 Fig. 1. Criteria ANP decomposition of CSRPI. X. Wu et al. Emerging Markets Review 55 (2023) 101021 6 not follow a sequential pattern (Lester et al., 2003; Miller and Friesen, 1984). Therefore, following previous research (Park, 2021; Shahzad et al., 2019; Zhao and Xiao, 2019), the CLC proxies of Dickinson (2011) are adopted to capture the dynamic nature of the CLC. It is assumed that cash flow reflects differences in company profitability, growth, and risk. Hence it is suggested to use cash flow from operation (ONCF), investment (INCF) and financing (FNCF) to assign companies into different life cycle stages. A significant advantage of this division method is that it does not imply a strict order across CLC stages. Instead, it allows companies to move back and forth dynamically between CLC stages (Drobetz et al., 2015). As shown in Table 3, the eight patterns generated from the possible combination of the sign (positive or negative) of three cash flows are divided into five stages by Dickinson (2011). Moreover, Dickinson (2011) also mentions that “the literature is silent regarding cash flows for shake-out firms. Consequently, shake-out firms are classified by default if the cash flow patterns do not fall into one of the other theoretically defined stages.” Other authors, e. g., Jawahar and McLaughlin (2001) and Gupta and Chin (1994), propose four CLC stages (excluding the shake-out stage). According to the authors’ analysis and experience it is proposed in the paper an original division of the three situations of the shakeout stage in Dickinson (2011) into mature or decline stages. Specifically, among the three situations of the shake-out stage, for pattern 4 and 5 shown in Table 3: ONCF >0 indicates that company operation activity is running normally; INCF >0 indicates that the company may dispose of some assets; FNCF >0 indicates the company’s investment demand; FNCF <0 indicates the company is returning investors as the mature company does, all this information shows the company is closer to the mature stage. For pattern 6, ONCF <0 means that the company’s operational activities are abnormal and serious business problems occur, which signals that the company is likely to enter the decline stage. The original authors’ division proposal for matching the relationship between the CLC stage and cash flow symbols of the sample companies is shown in Table 3, row E. The final applied authors model division of CLC by the dummy variable MA characterised by the mature (patterns 3–5) and non-mature (patterns 1–2, 6–8) companies’ stages is presented in Table 3, row F. It stems from stages distribution, see Table 1, and different companies’ behaviour in the mature and non-mature stages. Here, if the company is mature, MA is equal to 1; otherwise, it equals 0. 3.5. Control variables description Control variables considered are based on the early research results to control the potential factors that may affect the level of CSR participation. For instance, research shows that the larger the scale and the lower the financial leverage, the better the CSR (Cordeiro et al., 2018; Li and Zhang, 2010). Therefore, company size (SIZE) measured by the natural log of total assets and financial leverage (LEV) measured by the ratio of total debt to total assets are control variables included analogically in Alsaifi et al. (2021). The level of equity ownership concentration also affects companies’ CSR engagement (Faller and Knyphausen-Aufseß, 2018; Dam and Scholtens, 2013), so equity ownership concentration (EOC) equalling the shareholding ratio of the largest shareholder is included. Furthermore, cash and R&D are important resources for business operation and development and strongly influence CSR (Habib and Huang, 2019; Surroca et al., 2010). And R&D expense rate (R&D) is measured by the ratio of R&D expense to total assets, as in Clarkson et al. (2011). The cash holding rate (CH), measured by the ratio of the sum of the cash balance and marketable securities to total assets, is used similarly to Hsu (2018). In addition, older companies invest significantly more in CSR, especially in diversity and environmental issues (Jiraporn and Withisuphakorn, 2016). Company age (AGE) is controlled and measured by the natural logarithm of company age (Madden et al., 2020). Finally, year and industry categorial variables in the regression analyses to control the impacts of time and industry are included (Khan et al., 2021). 3.6. Fractional regression models formulation In many strategy and management cases, the dependent variable of interest is a proportion or a fraction, i.e. it ranges in the interval [0, 1]. Papke and Wooldridge (1996) introduced fractional regression applied in economics, and Wooldridge (2010) added technical discussion. By method comparison, Villadsen and Wulff (2021) demonstrated that fractional regression is a best-practice technique for many outcomes in the form of fractions, proportions, or percentages that are of interest to management and strategy researchers. In the paper, first, the dependent variable CSRPI ranges in the closed [0, 1] interval with the discrete non-Gaussian distribution. So the fractional response function regression is suitable for the intended model estimation (Baum, 2008). Second, the paper aims to investigate the impact of CFP on CSR (H1) and the moderating effects of the CLC stages on the association between CFP and CSR (H2). Therefore, Model 1, (2) which serves to test H1, is proposed, and the moderating and interaction effect of CLC hypothesised by H2 is verified by Model 2, (3), Table 3 Company life cycle stages and divisions. Row Division /Patterns 1 2 3 4 5 6 7 8 A Dickinson’s Introduction Growth Mature Shake-Out Decline B ONCF − + + + + − − − C INCF − − − + + − + + D FNCF + + − + − − + − E Authors’ Introduction Growth Mature Decline F Authors’ model Non-mature Mature Non-mature X. Wu et al. Emerging Markets Review 55 (2023) 101021 7 CSRPIit =β0+β1·EVArit+ +β2·SIZEit +β3·LEVit +β4·EOCit +β5·CHit +β6·R&Dit +β7·AGEit +β8·Industryit +β9·Yearit + ε it (2) CSRPIit =β0+β1·EVArit +β2·MAit +β3·EVArit ·MAit+ +β4·SIZEit +β5·LEVit +β6·EOCit +β7·CHit +β8·R&Dit +β9·AGEit +β10 ·Industryit +β11 ·Yearit + ε it (3) Here i is a company; t is a year; CSRPI is participation index of company CSR; EVAr is relative EVA and represents CFP; MA is a dummy variable referring to the company mature stage; EVAr ⋅ MA is the interaction effect between relative EVA and the CLC stages. The rest of the company-level control variables: SIZE is asset size, LEV is leverage, EOC is equity ownership concentration, CH is cash holding rate, R & D is research and development expenditure rate, AGE is company age, Industry is the industry of the company, and Year is the period of sample; ε is random residual deviation. The definition of each variable and how they are measured are explained in Table 4. The model computations were carried out using STATA software with a fractional regression module. 4. Results and discussion 4.1. Descriptive statistics and correlation matrix Table 5 summarises the statistics of all variables used in the analysis. For the 6730 observations, the mean value of CSRPI is 0.643, reflecting the sample companies’ average CSR participation level, and the mean value of EVAr is −0.011, which means the average CFP of sample companies is not desirable. Moreover, the statistical results show that 55.8% of the study sample companies are mature. The CSRPI ranges closed [0, 1] interval distribution with the discrete non-Gaussian distribution. The significance of the SK test (Skewness/ Kurtosis) for normality verification in Stata is 0.0000, so the null hypothesis of the normality distribution for CSRPI is rejected. Table 6 highlights the results of the correlation between the models’ variables. Because both ROE and EVAr are CFP measures, their correlation coefficient value is 0.782. But they do not appear in a model simultaneously; therefore, they do not affect multicollinearity. On the other hand, ROE and relative EVA are statistically different measures. The coefficient values of all other independent variables are less than 0.60, indicating that there is no issue of multicollinearity and that the estimated variables are sufficiently independent. In addition, a positive correlation between independent variables EVAr and CSRPI exists, which provides preliminary evidence that CFP is positively associated with CSR participation level. And mature companies show a positive correlation with CSRPI. 4.2. Regression relationship of EVAr and CSRPI - Model 1 Table 7 reports the results obtained by fractional regressing CSRPI on EVAr and control variables in Model 1 to study the first hypothesis for evaluating the impact of EVAr on CSRPI. The coefficient of EVAr is positive and statistically significant (β =0.282, p < 0.01), indicating that, on the whole, the increase in EVAr has a positive impact on the active participation of Chinese listed companies in CSR activities. This result is consistent with previous studies (Sial et al., 2018; Naciti, 2019; Zahid et al., 2019). Hence, in Hypothesis H1, a positive association between EVAr and CSRPI is confirmed. 4.3. The moderating effect estimation of CLC stages - Model 2 Table 7 presents results regarding the second hypothesis, hypothesising that mature companies may significantly strengthen the relationship between EVAr and CSRPI. To test this proposition, the continuous variable EVAr was used in the interaction term with the mean centre treatment according to the previous literature (Guo and Shen, 2019; Chen et al., 2012) and was then included in the analysis to alleviate the problem of multicollinearity and facilitate the interpretation of the main effect. Next, the moderating effect of CLC stages is examined by the link between EVAr and CSRPI. Model 2 tests the moderating effects of CLC stages. As shown in the results of Model 2, MA is connected significantly positively with CSRPI (β =0.058, p <0.05). This result is consistent with previous research results (Hasan and Habib, 2017; Jiraporn and Withisuphakorn, 2016). It means mature companies are more involved in CSR activities than non-mature companies. The results of Model 2 also show that the interaction of the EVAr⋅MA is connected significantly negatively Table 4 Variables description. Symbol Criteria Calculation formula CSRPI corporate social responsibility participation index a weighted average of criteria and weights EVAr relative EVA the ratio of EVA to total capital MA company mature stage for a mature company 1, otherwise 0 SIZE company size the natural log of total assets of a company LEV financial leverage the ratio of total debt to total assets EOC equity ownership concentration the ratio of the largest shareholder ownership CH cash holding rate the ratio is the sum of cash balance and marketable securities devided by total assets R&D R&D expense rate the ratio of R&D expense to total assets AGE company age indicator the natural logarithm of company age Industry The company’s industry categorical variable with two-digit codes for each industry Year years of analysis the 2018 sample is 2018, and the 2019 sample is 2019 X. Wu et al. Emerging Markets Review 55 (2023) 101021 8 with CSRPI (β = − 0.413, p <0.05), which means that the effect of EVAr on CSRPI for mature companies is not strengthened but weakened. This result is contrary to Hypothesis H2. Because the parameters estimated in the fractional regression model only provide the sign of the marginal effect of the covariates on the outcome, the magnitude is difficult to interpret. Therefore, the magnitude by calculating the average marginal effect is investigated. From the marginal effect of Model 2 in Table 7, for mature companies, the average marginal effect of EVAr on CSRPI is positive but not significant (p >0.1). In contrast, for non-mature companies, the average marginal effect of EVAr on CSRPI is positive and Table 5 Descriptive statistics. Variables Mean SD Min P5 P50 P95 Max CSRPI 0.643 0.252 0.000 0.071 0.753 0.941 0.976 EVAr −0.011 0.160 −3.071 −0.258 0.005 0.137 1.574 MA 0.558 0.497 0.000 0.000 1.000 1.000 1.000 SIZE 22.301 1.355 17.545 20.446 22.142 24.764 28.636 LEV 0.432 0.206 0.008 0.119 0.420 0.788 1.758 EOC 33.140 14.477 3.003 12.797 30.761 59.962 89.093 CH 0.182 0.129 0.001 0.037 0.148 0.451 0.926 R&D 0.019 0.023 0.000 0.000 0.015 0.056 0.702 AGE 2.227 0.801 0.693 0.693 2.303 3.258 3.401 ROE 0.024 0.328 −7.091 −0.344 0.067 0.218 5.013 Table 6 Correlation matrix. Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (1) CSRPI 1.000 (2) EVAr 0.120* 1.000 (3) MA 0.054* 0.104* 1.000 (4) SIZE 0.199* 0.134* 0.013 1.000 (5) LEV −0.020 −0.272* −0.159* 0.449* 1.000 (6) EOC 0.033* 0.172* 0.092* 0.193* 0.009 1.000 (7) CH −0.031* 0.172* 0.143* −0.201* −0.357* 0.069* 1.000 (8) R&D 0.038* 0.023 0.037* −0.236* −0.193* −0.108* 0.174* 1.000 (9) AGE 0.032* −0.134* 0.054* 0.428* 0.292* −0.076* −0.141* −0.288* 1.000 (10) ROE 0.100* 0.782* 0.064* 0.091* −0.293* 0.118* 0.141* 0.011 −0.097* 1.000 * Significance at the 5% level. Table 7 Results of Models 1 and 2 executed through fractional logit regression. Variables Model 1 Model 2 CSRPI Margins CSRPI Margins (logit) (logit) Non-mature Mature EVAr 0.282 0.063 0.437 0.098 0.005 (2.61)** (2.61)** (3.01)** (3.02)** (0.17) MA 0.058 (2.12)* EVAr⋅MA −0.413 (−2.14)* SIZE 0.245 0.055 0.245 0.055 0.054 (18.21)** (18.55)** (18.24)** (18.55)** (18.51)** LEV −0.570 −0.127 −0.563 −0.127 −0.125 (−6.55)** (−6.57)** (−6.42)** (−6.45)** (−6.42)** EOC −0.001 0.000 −0.001 0.000 0.000 (−1.35) (−1.35) (−1.43) (−1.43) (−1.43) CH −0.226 −0.051 −0.230 −0.052 −0.051 (−1.95) (−1.95) (−1.98)* (−1.98)* (−1.98)* R&D 3.086 0.690 3.058 0.688 0.679 (3.79)** (3.8)** (3.77)** (3.78)** (3.78)** AGE −0.027 −0.006 −0.030 −0.007 −0.007 (−1.32) (−1.32) (−1.46) (−1.46) (−1.46) Industry/ Year incorporated incorporated incorporated incorporated incorporated Constant −4.563 −4.578 (−15.64)** (−15.67)** * and ** denote significance at 5% and 1%, respectively. X. Wu et al. Emerging Markets Review 55 (2023) 101021 9 significant (p <0.01). These results imply that for mature companies, the effect of EVAr on CSRPI is weakened, while for non-mature companies, the effect of EVAr on CSRPI is enhanced. Moreover, for non-mature companies, the average marginal effect of EVAr on CSRPI is 0.098, which is much higher than the overall average marginal effect of EVAr on CSRPI in Model 1 (0.063). These results more accurately demonstrate that non-mature companies are more involved in CSRPI activities with the increase of EVAr. In addition, Hoetker (2007) emphasised that a graphical presentation allows readers to have the most complete understanding of the interaction effect. The phenomenon found in the study is illustrated intuitively in Fig. 2. For instance, Allison (1999) and Williams (2009, 2010) have mentioned that there may be problems using interaction terms for non-linear models, such as logit and probit. In particular, when the residual variability between groups is different, that is, when there is heteroscedasticity, parameter estimation may be biased and inefficient. In other words, once heteroscedasticity is present, the interaction term based on homoscedasticity in the original model is no longer significant. Fractional proportional models suffer shortcomings just like the logit and probit models, as emphasised by Williams (2015). In order to solve this potential problem affecting results, a heteroskedastic probit model where residual variance can differ and the same data in Model 2 are used. The result reports that heteroscedasticity in the sample does not exist (p >0.1 for ln (sigma)), which indicates that Model 2 is specified correctly and the results are reasonable and acceptable. This phenomenon in non-mature companies could be explained by CSR involvement’s reputational and strategic values (Hasan and Habib, 2017). This argument suggests that early-stage companies are equally likely to invest in CSR activities. The survival and development of young enterprises depend primarily on the support of stakeholders in terms of resources (Jawahar and McLaughlin, 2001). And CSR engagement is considered an effective tool for getting such support. Regarding the availability of resources as a precursor to CSR investment, Udayasankar (2008) argues that even resource-constrained companies may benefit from CSR activities because CSR participation allows restricted companies to obtain exclusive access to critical resources. Moreover, although CSR costs are high, the marginal benefit of CSR investments may be greater for younger companies than their mature counterparts (Hasan and Habib, 2017). Therefore, the non-mature companies are willing to participate in CSR projects when their resources are available (for example, EVAr is better). In other words, for non-mature companies, the effect of EVAr on CSRPI is obviously strengthened. The entry of a company into the mature stage indicates its success in the fierce market competition. At this moment, they do not depend on the support from stakeholders for survival because they can develop by themselves and have the ability to “fulfil all the responsibilities they have toward each primary stakeholder group” (Jawahar and McLaughlin, 2001). One of the results confirmed that mature companies have a higher CSR participation level than non-mature companies. But mature companies in this study seem to be more inclined to adopt a maintenance strategy in terms of the range of CSR participation because their CSR participation (CSRPI) level is no longer affected by EVAr. However, previous studies show that mature companies may be more willing to participate in certain CSR activities. Such mature companies are more responsible for diversity and environmental awareness but weaker in human rights and product safety (Jiraporn and Withisuphakorn, 2016). Since the research only deals with the impact of EVAr on the extent of company participation in CSR activities, it was found that mature companies weaken this relationship. The impact of EVAr on the depth of CSR activities can be one of the future research directions. Fig. 2. The relationship between EVAr and CSRPI in different CLC stages. X. Wu et al. Emerging Markets Review 55 (2023) 101021 16 Shahzad, F., Lu, J., Fareed, Z., 2019. Does firm life cycle impact corporate risk taking and performance? J. Multinatl. Financ. Manag. 51, 23–44. https://doi.org/ 10.1016/j.mulfin.2019.05.001. Sial, M.S., Zheng, C., Khuong, N.V., Khan, T., Usman, M., 2018. Does firm performance influence corporate social responsibility reporting of Chinese listed companies? Sustainability (Switzerland) 10 (7). https://doi.org/10.3390/su10072217. Stewart, G.B., 1991. The Quest for Value. HarperCollins, New York. Stinchcombe, A.L., 1965. Social Structure and Organisation. Rand McNally, Chicago. Surroca, J., Trib´ o, J.A., Waddock, S., 2010. Corporate responsibility and financial performance: the role of intangible resources. Strateg. Manag. J. 31 (5), 463–490. https://doi.org/10.1002/smj.820. Swandari, F., Sadikin, A., 2016. The effect of ownership structure, profitability, leverage, and firm size on corporate social responsibility (CSR). Binus Business Rev. 7 (3), 315. https://doi.org/10.21512/bbr.v7i3.1792. Tsai, W.H., Hsu, J.L., Chen, C.H., Lin, W.R., Chen, S.P., 2010. An integrated approach for selecting corporate social responsibility programs and costs evaluation in the international tourist hotel. Int. J. Hosp. Manag. 29 (3), 385–396. https://doi.org/10.1016/j.ijhm.2009.12.001. Udayasankar, K., 2008. Corporate social responsibility and firm size. J. Bus. Ethics 83 (2), 167–175. https://doi.org/10.1007/s10551-007-9609-8. Villadsen, A.R., Wulff, J.N., 2021. Are you 110% sure? Modeling of fractions and proportions in strategy and management research. Strateg. Organ. 19 (2), 312–337. https://doi.org/10.1177/1476127019854966. Waddock, S.A., Graves, S.B., 1997. The corporate social performance–financial performance link. Strateg. Manag. J. 18 (4), 303–319. https://doi.org/10.1108/ IJCHM-04-2017-0195. Wang, J., Lu, J., 2021. Religion and corporate tax compliance: evidence from Chinese Taoism and Buddhism. Eur. Bus. Rev. 11 (2), 327–347. https://doi.org/ 10.1007/s40821-020-00153-x. Wang, Q., Dou, J., Jia, S., 2016. A Meta-analytic review of corporate social responsibility and corporate financial performance: the moderating effect of contextual factors. Bus. Soc. 55 (8), 1083–1121. https://doi.org/10.1177/0007650315584317. Wernerfelt, B., 1984. A resource-based view of the firm. Strateg. Manag. J. 5 (2), 171–180. https://doi.org/10.1002/smj.4250050207. Williams, R., 2009. Using heterogeneous choice models to compare logit and probit coefficients across groups. Sociol. Methods Res. 37 (4), 531–559. https://doi.org/ 10.1177/0049124109335735. Williams, R., 2010. Fitting heterogeneous choice models with oglm. Stata J. 10 (4), 540–567. https://doi.org/10.1177/1536867x1101000402. Williams, R., 2015. Combining heterogenous choice and fractional response models to analyse interaction effects when the dependent variable is a proportion. In: European Survey Research Association 2015 Meetings, July (14). Wooldridge, J.M., 2010. Econometric Analysis of Cross Section and Panel Data. MIT press, Cambridge, Massachusetts. Wu, X., Hąbek, P., 2021. Trends in corporate social responsibility reporting. The case of Chinese listed companies. Sustainability 13 (15), 8640. https://doi.org/ 10.3390/su13158640. Wu, X., Dluhoˇ sov´ a, D., Zmeˇ skal, Z., 2021. Corporate social responsibility and profitability: the moderating role of firm type in Chinese appliance listed companies. Energies 14, 227. https://doi.org/10.3390/en14010227. Young, S.D., O’byrne, S.F., Young, D.S., Young, S., Stephen, O’Byrne, 2000. EVA and Value-Based Management. McGraw-Hill Professional Publishing. Yuan, C.M., Wu, F.P., 2010. A performance evaluation model on corporate social responsibility based on ANP. In: The 2nd International Conference on Information Science and Engineering. IEEE, pp. 2534–2538. https://doi.org/10.1109/ICISE.2010.5688965. Zahid, M., Rahman, H.U., Muneer, S., Butt, B.Z., Isah-Chikaji, A., Memon, M.A., 2019. Nexus between government initiatives, integrated strategies, internal factors and corporate sustainability practices in Malaysia. J. Clean. Prod. 241, 118329 https://doi.org/10.1016/j.jclepro.2019.118329. Zhao, T., Xiao, X., 2019. The impact of corporate social responsibility on financial constraints: does the life cycle stage of a firm matter? Int. Rev. Econ. Financ. 63 (September 2018), 76–93. https://doi.org/10.1016/j.iref.2018.08.010. X. Wu et al.