2.1 Organizational culture framework

Organizational culture is one of the most influential concepts in management research and practice, owing to its strong correlation with organizational outcomes. It imparts a sense of identity between its members based on values embodied in organizational rules, standard procedures, and goals (Jones, 2013). Organizational culture develops as a response to challenges of external adaptation and internal integration. A widely accepted definition by Schein (1990) describes organizational culture as a set of basic assumptions that shapes beliefs and rules. These assumptions, validated and refined over time, guide behaviors and attitudes within the organization in socially acceptable ways and are regarded as the key to organizational success. Therefore, the successful implementation of SDGs practices heavily relies on the values and ideological foundations of an organization's culture.

Linnenluecke and Griffiths (2010) expand this concept by linking it to corporate sustainability, suggesting that a sustainability-oriented organizational culture paves the pathway for the adoption of corporate sustainability principles. Using the competing values framework, which categorizes organizational cultures by focus (Jones et al., 2005), the authors argue that organizations driven by a rational goal culture prioritize resource efficiency as a core component of corporate sustainability. This reflects managers’ recognition of the benefits of cost reduction and increased efficiency, potentially to guide responsible consumption and production decisions.

Subsequent investigations have advanced this idea through the green organizational culture, introducing an ideology balancing economic, social, and ecological development (Bertassini et al., 2021; Gürlek & Tuna, 2017). This concept underscores the importance of environmental management, which builds additional environmental capabilities that can provide a competitive advantage. Therefore, integrating SDGs associated with resources and environment can be seen as an outcome of organizational culture and strategic management decisions. Sustainability orientation emerges when an organization adopts an outward-looking culture and proactively integrates sustainable development goals into strategies to enhance operational effectiveness. To support this transition, organizations could implement CSR initiatives to cultivate a green, sustainability-oriented culture across all parts of the value chain (Wickert et al., 2016).

2.2 Dynamic capabilities theory

Rooted in the natural-resource-based framework (NRBV) (Hart, 1995; Hart et al., 2010; Russo & Fouts, 1997), organizational resources and strategic capabilities are key to achieving a sustained competitive advantage, contingent upon how effectively organizations manage their relationships with natural resources and the environment. As natural resources become scarcer, the demand for corporate sustainability and CE implementation increases, motivating firms to minimize resource use while maintaining or increasing output. This strategic management of resources provides an opportunity for firms to craft distinctive CSR profiles that highlight sustainability, differentiating them from competitors (Wickert & Risi, 2019).

While these competitive advantages may address environmental and social challenges arising from CSR decisions, they have been criticized for assuming that resources, skills, and organization structures remain static. The unavailability of resources and environmental issues accelerate the market complexity. Developed from NRBV, dynamic capabilities theory (Teece et al., 1997) emphasizes a firm's organizational and strategic routines to integrate, build, and reconfigure internal and external competencies to thrive in rapidly changing environments and achieve long-term success. This framework further supports the idea that firms can strategically integrate CSR considerations into their CE initiatives, thereby advancing their ability to adapt its resource configuration to changing markets (Battisti et al., 2022; Cavusgil et al., 2007; Leonidou et al., 2023). Dynamic capabilities are, therefore, essential for developing proactive socio-environmental practices and firm performance. Hence, firms should “evolve” within the context of SDGs and CE through CSR initiatives, as failure to maintain a strong CSR commitment may undermine their competitive advantage.

Scholars have increasingly discussed dynamic capabilities from a sustainability or green perspective (Buzzao & Rizzi, 2020; Correggi et al., 2024), which extends the original concept to include specific capabilities that enable firms to systematically sense and seize sustainable development opportunities. In this context, firms must act on these insights to align their strategies with CE objectives. Although existing literature has explored CE from the dynamic capabilities perspective (e.g., Coppola et al., 2023), studies reveal that while CE effectively addresses environmental concerns, it may overlook broader social issues (Baah et al., 2023; Corvellec et al., 2021). Incorporating CSR initiatives can address this oversight by promoting a more comprehensive approach to sustainability, bridging social and environmental dimensions.

3.1 CSR, resource, and environmental sustainability

Addressing natural resource consumption and emissions reduction has become a key priority in corporate sustainability (Erzurumlu et al., 2023; Welch & Southerton, 2019). Guided by the global agenda of the SDGs, the integration of CSR initiatives, corporate governance, and organizational culture is widely recognized as essential for successfully adopting and achieving sustainable development within organizations (Baumgartner, 2014; López et al., 2007). A firm's commitment to CSR has been conceptually linked to its instrumental ability to implement strategic plans that align with enhanced sustainability performance (Engert et al., 2016; Fonseca et al., 2021). Barnett (2007) shows that a firm's stakeholder influence capacity (SIC) exhibits path dependency, highlighting the role of sustained CSR commitment in fostering trust and collaboration with stakeholders, which, in turn, shapes a firm's future outcomes. This connection forms the basis of our study on CSR precedence, which reflects a firm's strategic commitment to CSR through historical performance and adaptation to meet societal expectations.

Empirical research addressing the connection between CSR and SDGs at the firm level remains sparse (López-Concepción et al., 2021; Mio et al., 2020), notably in terms of discussions on specific SDGs in the context of resource and environmental sustainability. The role of CSR precedence in supporting the achievement of these SDGs has yet to be examined. Prior studies have identified various factors that positively influence sustainable practices. For instance, the presence of female directors increases renewable energy consumption (Atif et al., 2021). Changes in carbon disclosure levels are positively associated with firms’ recycling and carbon performance (Busch et al., 2020; Ma et al., 2022; Qian & Schaltegger, 2017). Additionally, perceptions of strong beliefs from top managers lead to stronger sustainability adoption (Gopalakrishna-Remani et al., 2022). However, the potential impact of CSR initiatives has yielded mixed findings. Kudlak (2019) indicates that CSR is capable of reducing CO₂ emissions when driven by self-regulatory measurements. In contrast, Fukuda and Ouchida (2020) illustrate a scenario where CSR activities may lead to increased emissions despite enhancing overall welfare, suggesting the complexities of CSR's environmental benefits.

In response to these insights, our study extends the literature by examining the relationship between CSR precedence and firms’ management of resources, waste, and emissions. Drawing on the organizational culture framework (Schein, 1990) and the corporate sustainability culture (Linnenluecke & Griffiths, 2010), we propose that CSR precedence, as reflected by past CSR performance, signifies a firm's espoused beliefs and core values that influence its resource use and emission strategies. The interdependence between human relations and ecological systems is framed within the broader context of environmental, social, and corporate governance, which are key components of CSR (Baumgartner, 2014).

3.2 CSR and CE

The CE paradigm, which emphasizes restorative and regenerative patterns of resources and environment, has become a pivotal strategy for firms committed to achieving SDGs (Environment European Commission, 2023). CE is regarded as a catalyst for business sustainability, offering economic, social, and environmental benefits by fostering competitive advantages. These advantages drive growth while emphasizing social cohesion, integration, and environmental stewardship. With growing recognition of net-zero initiatives, CE has become increasingly central to investors and other stakeholders, highlighting its role in sustainable business strategies (Esposito et al., 2023; Morea et al., 2022). Furthermore, CE's alignment with SDGs varies by goal, with responsible consumption and production (SDG12) serving as a key focus. Climate action (SDG 13) is also identified for its potential synergy with CE practices (Nazir & Doni, 2024; Schroeder et al., 2018). This underscores the complementary relationship between CE and SDG practices in promoting resource efficiency and environmental sustainability.

Existing literature has emphasized the increasing convergence of CE and CSR across industries (Scarpellini et al., 2020). According to Kumar et al. (2022), enhancing CSR performance is crucial for firms in the food supply chain, as effective CE adoption can significantly reduce food waste and optimize resource utilization. Ma et al. (2022) confirm a link between CE and CSR engagement, finding that a firm's carbon disclosures—a key aspect of CSR commitments—significantly influence their recycling willingness and CE integration. “Investment in CSR” has been acknowledged as a top criterion for firms to enhance their CE performance, which integrates the principles of CE and CSR to develop circular business processes for long-term competitiveness (Ghosh et al., 2023). However, little research has explicitly examined the correlation between CSR precedence and CE. This research aims to fill this gap by verifying the relationships between firms’ CSR precedence and CE practices.

To illustrate the hypotheses above, we construct a conceptual model that visually represents all the relationships examined in the empirical sections below. Figure 1 serves as a roadmap, outlining the key variables, their interconnections, and the theoretical assumptions underpinning the analysis.

4.1 Sample selection

Our data sample is sourced from merging multiple databases: MSCI KLD database for CSR precedence data (widely used in studies, such as Flammer, 2013; Grewal et al., 2021), Thomson Reuters Asset4 database for resource reduction and emission reduction scores (widely used in studies, such as de Villiers et al., 2022; Liang & Renneboog, 2017), ISS Director for board of director information, Compustat for financial data, and Bloomberg for the ESG data.

The initial sample, collected from the MSCI KLD database, covers about 2400 publicly traded US firms listed in the MSCI USA Investible Market Index (IMI) from 2006 to 2017. The United States has a long-standing tradition of robust regulatory frameworks, exemplified by the Securities Acts of 1933 and 1934, the National Environmental Policy Act of 1969, the Sarbanes-Oxley Act of 2002, and the Dodd-Frank Act of 2010. These policies promote transparency and environmental sustainability, and foster a supportive environment for CSR, ensuring the reliability and representativeness of our data. The selected period of 2007 to 2017 offers a 10-year window to examine long-term trends and shifts in CSR precedence. To address the unbalanced nature of the MSCI KLD dataset, we choose 2007 as the starting point to ensure consistency in data quality and coverage, with 2017 marking the endpoint due to data availability.

The data merging sequence is as follows: KLD data, Thomson Reuters Asset4 data, ISS Director data for the board of directors, Compustat, and Bloomberg ESG data. We exclude firms associated with typical “sinful” industries, including alcohol, gambling or gaming, tobacco, firearms, military, and nuclear⁴ (Hong & Kacperczyk, 2009; Kim et al., 2012), which are associated with practices that challenge social norms and cause concerns in economic, social, and governance aspects. Firms in these industries have limited discretion to build positive CSR precedence due to inherent conflicts between their core business models and CSR principles, significant negative externalities, and the absence of public trust (Guo et al., 2022; Oh et al., 2016). These limitations undermine their ability to adopt credible CSR practices, potentially introducing biases that are less relevant to our study's focus and results. After removing the missing variables, the main sample consists of 8047 firm-year observations across 1072 firms. In the subsample of energy, water, and total waste disclosure, we observe 2711 firm-year observations across 287 firms for energy use, 2023 firm-year observations across 208 firms for water use, and 1972 firm-year observations across 205 firms for total waste from 2007 to 2017. We believe the data sourced are robust and relevant to our research objectives.

4.2 Dependent variables

Our primary dependent variables are crafted to gauge a firm's alignment with responsible consumption and production (SDG 12) across multiple dimensions, emphasizing resource efficiency and environmental sustainability. Actions reducing emissions and waste also support the objectives of climate action (SDG 13). Given the lack of consensus on firm-level SDG representatives, we first use two scores from the Asset4 database to measure their resource and environmental performance: the resource reduction score and emission reduction score (Khaled et al., 2021; Shaukat et al., 2016). While a few exceptions, such as Umar and Chunwe (2019), have examined the connection between CSR and resource consumption or waste management, research in this area remains limited. In this study, we also incorporate energy consumption, water use, and waste production, which directly reflects how firms reduce negative social and environmental impacts (Schönherr et al., 2017).

Resource reduction (RESRED) encapsulates a company's capacity to reduce the use of natural resources and to find more eco-efficient solutions by improving supply chain management. A higher resource reduction score indicates more efficient use of natural resources. Concurrently, emission reduction (EMISRED) quantifies a company's efforts toward enhancing environmental sustainability in the production and operational processes. It reflects a company's capacity to reduce air emissions, wastes, water discharges, and spills and collaborate with environmental organizations to mitigate environmental impacts on the local and broader communities, aligning with the objectives of SDG 13 (Climate Action). A higher emission reduction score indicates more effective performance in reducing emissions. Both variables are collected from the Asset4 database (Refinitiv symbols TRESGENRRS and TRESEGERS, respectively).

To account for industry-specific variations in resource and emission reduction metrics, we employ the Fama-French 48 industries approach (Fama & French, 1997) to compute the industry-adjusted resource reduction (ADJRESRED) and emission reduction (ADJEMISRED), where the industry-adjusted measurements are calculated using scores divided by their industry average. Changes in the industry-adjusted variables are represented as CHGADJRESRED and CHGADJEMISRED, which account for the potential stickiness of each firm's scores over time.

In the first stage of the Heckman correction model, we incorporate three dummy variables to mitigate potential selection biases. These variables stand for whether firms have disclosed their energy consumption (DENERGY), water use (DWATER), or total waste (DWASTE). To address data skewness related to size differences (magnitudes) when reporting resource consumption, we employ natural logarithms for our consumption and waste datasets. The industry adjustment, consistent with our earlier approach, is applied to these logarithmic values, yielding ADJENERGY, ADJWATER, and ADJWASTE. Annual variations in these logarithmic values are represented as CHGENERGY, CHGWATER, and CHGWASTE.

Aligned with the core principles of CE, which aims to maximize resource efficiency and minimize waste generation, we introduce CE variables in line with the spirit of recycling and regeneration (Hailemariam & Erdiaw-Kwasie, 2022; Panchal et al., 2021). These variables—RENEWR, RECWATR, and RECWASTR,—respectively, quantify key aspects of a firm's CE practices. Specifically, RENEWR measures the proportion of renewable energy used in total energy consumption. RECWATR captures the proportion of recycled water to total water use. RECWASTR assesses the percentage of waste that is recycled out of the total waste generated. They offer comprehensive and quantitative insights into a firm's commitment to sustainable resource management. Consistently, we apply industry adjustments to these logarithmic values, denoting the variables as ADJRENEWR, ADJRECWATR, and ADJRECWASTR.

4.3 Independent variables

We follow the widely used MSCI KLD database to construct a firm's CSR precedence based on CSR scores, which serve as our two main independent variables. The MSCI KLD database offers dual scores to describe a firm's CSR performance: CSR strength score and CSR concern score (MSCI ESG RESEARCH, 2019). Based on the measurements of KLD dimensions (Kumar et al., 2016; Price & Sun, 2017), the CSR strength score is constructed on positive environmental, social, and governance indicators in which the firm demonstrates excellence, such as clean technology, biodiversity, and community engagement. In contrast, the CSR concern score reflects negative indicators related to public and regulatory concerns, highlighting actions that might elicit external concerns and involvement due to perceived irresponsibility, such as toxic emissions, community impacts, and child labor.

Regarding potential endogeneity concerns, we exclude scores under environmental criteria for the two main variables since they could be directly related to our dependent variables (resource reduction, emission reduction, energy use, water use, and total waste). Our two CSR measures then represent firms’ corporate governance and social responsibility-related CSR performance. Thus, we utilize the total CSR strength scores without any CSR concerns (STRNOCON) to represent firms with a positive (high) CSR precedence. We believe firms with positive CSR precedence will proactively engage in positive CSR and have no reason to use positive CSR to offset/greenwash their negative CSR (CSR concerns) since they are not associated with any CSR concerns.

Our second dependent variable represents firms with a negative (low) CSR precedence, measured by a firm's total CSR concern scores without any strength scores (CONNOSTR). These firms have yet to rectify their irresponsible actions, such as bolstering positive CSR engagement or mitigating CSR concerns. We employ a 1-year lag for the two primary independent variables to capture the firm's previous (past) CSR performance. Additionally, we apply a 2-year lag as a robustness check.

4.4 Control variables

We first control for a firm's transparency level, potentially influenced by ESG reporting disclosures and the board's commitment to CSR monitoring, by including the ESG disclosure score (ESGDISCL) (Liao et al., 2015). A dummy variable (CSRCOMM) is constructed to represent whether the firm has a CSR committee, which determines whether this firm is more likely to report its energy use, water use, and total waste in the first stage of our Heckman correction model.

Second, we control the linkage between senior executives’ payment and sustainability (CSR) targets, referred to as DLINK. This linkage might motivate the adoption of sustainable resource consumption and emission reduction at the managerial and decision-making levels. Prior studies suggest that gender diversity enhances a firm's adaptive capacity in CSR, noting that women might be more attuned to CSR nuances (Cabeza-García et al., 2018; Kuzey et al., 2022). Reflecting this, we include variables such as the presence of a female CEO (WCEO) and the percentage of female board members (PCTWBOD) to capture the impact of gender diversity.

Existing literature has also highlighted the important role of board characteristics on CSR performance (Bolourian et al., 2023; Liao et al., 2018). Our analysis introduces a series of control variables to capture board dynamics. First, we include the percentage of independent board members (PCTINDEP) to control for board independence and monitoring effectiveness (Kassinis & Vafeas, 2002). Second, board size (BODSIZE) controls the group dynamic, information effectiveness, and stakeholder diversity (Chang et al., 2017; Kock et al., 2012). Following Jain and Zaman (2020), the average percentage of board meeting attendance (BODATTEND) represents the board member engagements and monitoring activities. Lastly, CEO duality indicates whether the chair of the board also holds the position of the corporate CEO.

Furthermore, to represent firms’ potential growth, we first consider the market value over the book value of firms (MKTBOOK). Regarding the correlation between a firm's financial and environmental performance (Castilho & Barakat, 2022), we include the leverage (LEV), defined as total debt divided by total assets, and the firms’ total assets (SIZE) as a measure of firm size. Innovation also influences a firm's environmental performance (Hao & He, 2022), so we also include firms’ research and development expenses to total net sales ratio (RNDR) to control for this factor. A detailed variable description list is available in Table 1.

4.5 Empirical model

RCP_i.t represents firms’ performance on energy consumption, water use, emission reduction, and total waste in the current year. We estimate the first equation using the ordinary least squares as the baseline approach and incorporate the fixed effects of industry and year with two-way clustering of the standard errors (Petersen, 2008).

This equation incorporates a dummy variable (CSRCOMM_i.t), which indicates whether the firm has a CSR committee monitoring its CSR performance. It serves as an instrumental variable that captures the firms’ willingness to report and disclose their natural resources consumption and total waste. Another instrumental variable incorporated is the firm's ESG disclosure score, suggesting the level of ESG reporting transparency (ESGDISCL_i.t). The second stage regression is modeled identically to Equation (1). Our simultaneous estimation of the first and third equations forms the Heckman two-step regression.

CEP_i.t represents a firm's CE practices, which are measured by the renewable energy consumption relative to total energy consumption (RENEWR_i,t), recycled water use relative to total water use (RECWATR_i,t), and recycled waste relative to total waste (RECWASTR_i,t).

5.1 Descriptive statistics

Table 2 presents summary statistics of the 8047 observations. Firms exhibit average scores of 49.85 and 48.94 in resource reduction (RESRED) and emission reduction (EMISRED), respectively. Approximately 34% of our sample reported data on energy consumption (DENERGY), while 25% provided information on water use (DWATER) and total waste generation (DWASTE). Firms that reported their resource consumption status have average natural logarithm values of 6.88, 7.9, and 3.76 for energy consumption, water use, and total waste, respectively. These logs correspond to 973,000 MWh of energy consumption, 2.7 million cubic meters of water use, and 43,000 metric tons of total waste per year.

Compared to firms with negative CSR precedence (CONNOSTR), we find more firms exhibit positive CSR precedence (STRNOCON), with an average value of 0.65. According to indicators related to CSR report, 12% of firms in our sample have a CSR committee (CSRCOMM) and the average ESG disclosure score (ESGDISCL) is 16.84. Additionally, only 2% of firms implement a compensation structure that links senior executives’ pay to corporate sustainable performance (DLINK). Besides, 4% of firms have women CEOs (WCEO), and the average percentage of women on the board (PCTWBOD) is 15.93%. The average percentages of independent board members (PCTINDEP), board size (BODSIZE), and board meeting attendance (BODATTEND) are 82.39%, 11 board members, and 79.21% board attendance. Approximately 49% of our sample firms have the CEO acting as the chairperson of the board (CEOCHAIR). For financial metrics, the average market value of the firm to its book value (MKTBOOK) of our sample firms is 1.36, and the average natural log of total assets of 8.96, corresponding to around $7.8 billion in total assets.

Table 3 shows the sample distribution across different industries formed by the Fama-French 48 industries classification (Fama & French, 1997). Approximately, firms from financial industries (banks, insurance, real estate, and other financial services) and those from manufacturing industries (SIC codes between 2000 and 3999) each account for a third of the observations in our sample. For robustness, we re-examine the results by analyzing different time scales, comparing manufacturing firms versus non-manufacturing firms, excluding financial firms and industries with a limited number of firms. Detailed explanations of these robustness checks are provided in the corresponding section.

5.2 Regression results

5.3 Robustness tests

We deepen our understanding of our findings through several robustness tests. First, using the generalized method of moment (GMM) model, we estimate the dynamic panel data regression (Arellano & Bond, 1991) to address potential serial correlation, reverse causality, endogeneity, and unobserved heterogeneity (Wintoki et al., 2012). The GMM results also align with the significant influence of CSR precedence as reported in our results. In addition, we present the regression results by comparing 1-year lagged CSR precedence and 2-year lagged CSR precedence. As shown in Table 7, the results are consistent with our main findings presented in Table 4.

Our research further explores potential industry-specific factors by analyzing both manufacturing and non-manufacturing sectors. Considering that manufacturing firms typically consume more resources and produce more waste, we reassess our first regression by comparing subsamples of manufacturing firms and non-manufacturing firms. The results affirm the substantial impacts of CSR precedence on both manufacturing and non-manufacturing firms regarding their resource and emission reduction scores, highlighting that our results are not driven by firms from the manufacturing industries in our sample. Similarly, to account for stricter regulatory requirements faced by financial and utilities firms, which constitute approximately one-third of our sample, we exclude these firms and find that the results remain consistent with our main findings.⁵ Additionally, we exclude industries with a small number of firms (e.g., soft drinks, coal, and aircraft) to test whether our findings are influenced by these outliers. The unabated results confirm the consistency of our conclusions.