Sustainable Strategies: Navigating Corporate Social Responsibility and Irresponsibility for Enlightened Self-Interest

Abstract

As firms increasingly engage in both corporate social responsibility (CSR) and irresponsibility (CSIR) activities, this study expands the traditional question “does it pay to do more CSR?” to explore the less-studied question “does it pay to do less CSIR?”. We employ stakeholder theory and expectancy disconfirmation theory to outline three sustainable strategies for firms to financially benefit (enlightened self-interest) from CSR/CSIR activities: proactive strategy (increasing CSR, or doing more good), rectification strategy (reducing CSIR, or doing less bad), and aggressive strategy (increasing CSR while reducing CSIR, or doing more good and less bad). Our research objective is to evaluate the financial viability of different CSR/CSIR strategies. We hypothesize that the rectification strategy will surpass the proactive approach, while anticipating that the aggressive strategy will emerge as the most financially advantageous. Our dataset consists of 12,567 firm-year observations (3422 firms) spanning 1994 to 2013, and we conduct rigorous analyses to evaluate these strategies. The findings reveal that the rectification strategy surpasses the proactive strategy, with the aggressive strategy emerging as the most advantageous. The study contributes theoretically and offers managerial insights into these results.

1. Introduction

The construct of corporate social performance (CSP) encompasses a broad spectrum of behaviors that can either be beneficial, represented by corporate social responsibility (CSR) activities, or detrimental, encapsulated by corporate social irresponsibility (CSIR) actions [1]. CSR embodies virtuous initiatives that extend beyond self-interest to benefit firms’ stakeholders and society [2]. The extant studies on CSR have mainly explored the value-enhancing capabilities of firms’ CSR activities. Although the literature shows inconsistent conclusions regarding the influences of firms’ CSR on their financial performance, many studies show that firms use CSR strategically to align their social goals and corporate objectives to maximize firm value for their shareholders and stakeholders [3,4,5]. Consequently, firms have been encouraged to intensify their CSR activities [6,7].

In contrast, CSIR comprises detrimental actions initiated by firms that primarily benefit select stakeholders while endangering numerous others [8]. Unlike CSR, CSIR results in adverse consequences, triggering scandals [9], impairing firm reputation [10,11], eroding stakeholder trust, and ultimately leading to financial losses [12]. Furthermore, CSIR can inflict catastrophic consequences on businesses, society, and the environment, such as irreparable environmental pollution and degradation [13,14,15]. A typical example is the 2010 Deepwater Horizon explosion, which marked the worst offshore oil spill in U.S. history, leaving enduring scars in the Gulf of Mexico [16,17] a decade later [18].

A prevailing theme in academia underscores the simultaneous engagement of firms in CSR and CSIR activities [19,20]. Firms, as individual entities, exhibit variances in the extent of CSR and CSIR activities they undertake [21]. The tension between CSR and CSIR is particularly evident when firms employ CSR as a form of “image washing” to mitigate the adverse effects of CSIR [22].

Considering the severe repercussions of CSIR on stakeholders and society, scholars emphasize the paramount importance of refraining from engaging in harmful practices, emphasizing that this takes precedence over expanding CSR activities [11,23]. However, for firms already embroiled in CSIR, an important question emerges: can a strategy aimed at reducing CSIR restore stakeholder trust and satisfaction, thereby improving corporate financial performance (CFP)? CFP is an indicator of a firm’s profitability and growth potential [24,25]. This question remains largely unaddressed, despite its significance in providing firms with compelling incentives to rectify their CSIR, thereby mitigating the adverse effects on stakeholders and society. Simultaneously, two related questions warrant attention: What are the other strategic options available, and which strategy proves most advantageous for firms to adopt? Answering these questions can furnish managers with lucid guidance concerning their CSR/CSIR-related strategies.

To answer these critical questions, this study first identifies strategic positioning options regarding firms’ CSP, especially for those concurrently involved in CSR and CSIR. It subsequently compares and ranks these strategies in terms of their effectiveness in enhancing CFP. The study relies on an integrated theoretical framework founded on stakeholder theory and expectancy disconfirmation theory (EDT). The integration of these theories allows for the incorporation of stakeholder expectations as an intermediary factor. This is crucial due to the expectation gap between stakeholder expectations and firms’ self-perceived responsibilities, which significantly influences stakeholder behaviors towards these firms, thereby explaining the resultant effects on CFP. Building on this integrated theoretical framework, the study identifies three strategic positioning options concerning firms’ CSR/CSIR behaviors, namely proactive strategy (increasing CSR or doing more good), rectification strategy (reducing CSIR or doing less bad), and aggressive strategy (increasing CSR while reducing CSIR, doing more good and less bad).

Empirically, this study utilizes a comprehensive and longitudinal dataset comprising 12,547 firm-year observations from 1994 to 2013, covering 3422 firms. A series of unique analyses is conducted through regression models to determine the most effective strategic positioning option for firms seeking to enhance CFP. The results reveal that the rectification strategy surpasses the proactive strategy in effectiveness; thus, doing less bad proves more valuable than doing more good.

The primary theoretical contributions are as follows. Firstly, the study contributes to CSR literature by extending discussions beyond the question of “does it pay to do more CSR” to consider “does it pay to do less CSIR.” Secondly, it theoretically summarizes three strategic positioning options for firms, particularly those simultaneously involved in CSR and CSIR, allowing them to formulate their CSR/CSIR strategies. The study also underscores that CSIR is theoretically distinct from CSR, presenting an asymmetrical construct, which enriches the research on CSIR. Furthermore, the study introduces EDT to the CSP-CFP literature, extending the theory’s scope beyond its initial focus on customer groups to encompass all stakeholders. This extension enables the inclusion of stakeholder expectancy as an intermediary factor, elucidating the underlying mechanisms of the study’s theoretical arguments.

From a practical standpoint, the study underscores the merits of prioritizing the reduction of CSIR when making CSP-related decisions. This is crucial because CSIR can lead to severe consequences for the environment and society, warranting a sense of urgency in controlling and mitigating CSIR, particularly before expanding CSR initiatives [26].

In the upcoming sections, we will commence with a comprehensive literature review prior to delving into the development of theoretical arguments. Subsequently, we will present the methodology employed, elucidate the results obtained, and engage in a discussion that encompasses the implications and limitations of this study.

2. Literature Review

Scholars have delved extensively into the connection between CSR and CFP. While numerous meta-analyses tend to present a positive CSR-CFP relationship [27,28,29,30,31], the literature has offered inconsistent findings based on various theoretical frameworks [32].

First, scholars rooted in Friedman’s trade-off theory [33] found a negative relationship between CSR and CFP [34]. They argue that resources allocated to CSR could be better employed in profit-generating activities. Second, adherents of stakeholder theory [35] suggest that CSR positively relates to CFP [36,37,38] by satisfying stakeholder needs [7] and enhancing the firm’s “goodwill reservoir” [6] and reputation [39]. Some even argue that a virtuous cycle exists between CSR and CFP, as CSR enhances competitive advantages, subsequently boosting CFP and stimulating further CSR [40,41,42,43].

Third, scholars have proposed non-linear CSR-CFP relationships. Bhattacharya and Sen (2004) [6] observed that consumers from East Asia are price-sensitive, often hesitant to pay extra for products or services offered by environmentally responsible firms. Chen et al. (2018) [44], on the other hand, examined publicly traded non-financial service companies in the US. Together, these two studies suggest a U-shaped relationship: both very low (minimized cost on Corporate Social Responsibility) and very high levels of CSR (resulting in high stakeholder satisfaction) correlate with higher Corporate Financial Performance (CFP). Conversely, others [45,46] advocate an inverted U-shaped relationship, indicating that CSR positively relates to CFP up to an optimal level, beyond which further CSR efforts harm CFP [45,47]. More interestingly, a few recent studies [48,49] have found a positive but non-linear CSR-CFP relationship.

Additionally, some scholars argue for no specific CSR-CFP relationship [50]. The reason is that costs and benefits of CSR can offset each other [51], and confounding factors like advertising expenses and environmental and social regulations can disrupt the CSR-CFP link [2,52], making it challenging to discern a consistent CSR-CFP relationship.

Therefore, the literature presents inconclusive findings regarding the CSR-CFP link, partly due to the inconsistent approaches to measuring CSR. A typical way of measuring CSR is that studies often treat CSIR as a symmetric concept with CSR, combining strengths (considered as CSR) and concerns (considered as CSIR) as the measurement of CSR [53]. In recent times, scholars advocate that CSR and CSIR are distinct yet interdependent concepts with variant causes and consequences, each deserving scholarly attention [20,23,54,55,56]. Consequently, research has begun to examine how CSIR relates to CFP as a standalone construct [20]. Just as in the CSR-CFP studies, research on the CSIR-CFP relationship also yields inconsistent arguments and findings. Some argue for a positive CSIR-CFP relationship, at least in the short term. For instance, Stewart (2016) [57] found that CSIR predicts higher CFP in the subsequent year, as CSIR activities offer firms cost advantages. However, once CSIR is disclosed, these advantages vanish, leading to a negative long-term CSIR-CFP relationship [58,59]. Disclosed CSIR can lead to increased media scrutiny, impaired stakeholder relationships, reputational damage, and financial risks [13,60,61,62].

Nevertheless, some studies [63] question whether stakeholders consistently hold firms accountable for their CSIR activities, suggesting that firms may face no consequences due to stakeholders’ limited attention [64]. Therefore, the extant literature lacks a clear and consistent CSIR-CFP relationship.

Recent research has extended the focus to examine how concurrent CSR and CSIR affect CFP. Some studies suggest a positive effect of CSR on CFP in the presence of CSIR, indicating that CSR’s “angel-halo effect” prevails [15,61]. However, other studies refute this, suggesting that the positive influence of CSR may be weaker than the negative influence of CSIR in a combined scenario [65].

Furthermore, a study by Price and Sun (2017) [12] uncovers that firms with lower engagement in CSR and CSIR display higher CFP compared to firms highly involved in both. Kang et al. (2016) [66] find that the insurance mechanism (i.e., CSR ensures against subsequent CSIR) is not supported, but the penance mechanism (i.e., CSR trails the firm’s prior CSIR) is supported. However, this study also suggests that “the penance mechanism is ineffective in offsetting negative performance effect due to CSIR” (p. 59).

Despite these enriching studies on CSR and CSIR, the question “does it pay to do less CSIR?” (i.e., the impacts of firms’ rectification strategy, doing less bad, on CFP) and how the rectification strategy compares to proactive (doing more good) and aggressive (doing more good and less bad) strategies remain unanswered. Building upon the existing research on the outcomes of firms’ CSR and CSIR, this study aims to contribute to the literature by addressing these questions concerning the financial effects of firms’ varying approaches to doing more good and/or doing less bad.

3. Theoretical Framework and Hypotheses Development

In the subsequent sections, we develop our integrated theoretical framework, drawing from both stakeholder theory and EDT, to examine the potential strategic choices that firms may adopt concerning CSR/CSIR for financial gain. We harness the explanatory potential of theoretical traditions centered around negativity bias to underpin our arguments in the comparative analysis of these strategic alternatives.

3.1. The Influence of Stakeholders’ Expectancy Disconfirmation

Stakeholder theory, as elucidated by Freeman (1984) [35], underscores the significance of businesses forming relationships and generating value for all their stakeholders, encompassing parties such as suppliers, consumers, and employees who are affected by or influence a firm’s decisions and practices. Stakeholder theory offers an apt framework for examining CSP since it views firms as open systems, fostering mutual relationships between the focal firm and its stakeholders [67,68].

Within these mutual relationships in the context of CSP, stakeholders, as articulated by Wood and Jones (1995, p. 243) [68], undertake a four-fold role:

• Establish expectations regarding corporate performance, which may be explicit or implicit, and may or may not be communicated.
• Experience the effects of corporate behaviors, whether or not they are cognizant of the source.
• Evaluate the effects or potential effects of corporate behaviors on their interests or the alignment of corporate behaviors with their expectations.
• Act upon their interests, expectations, experiences, and evaluations concerning the focal firms.

These stakeholder roles require further clarification, particularly concerning how stakeholders develop expectations, assess a firm’s CSP, and subsequently influence the firm’s CFP. Stakeholders play a crucial role in shaping a firm’s actions and outcomes, impacting decisions that nurture positive relationships and contribute to long-term success. Therefore, understanding stakeholder expectations and experiences is vital for comprehending decision-making processes and performance factors. This ensures that research findings are not only relevant but also applicable to real-world contexts, offering actionable insights for practitioners. To elucidate these connections, we propose that an integrated framework, combining stakeholder theory and EDT, provides a comprehensive explanation of the CSP-CFP link.

EDT, with a focus on consumer behavior, posits that consumers formulate expectations against which they evaluate a firm’s perceived product or service quality [69,70]. This comparison results in expectancy disconfirmation when a gap exists between expectations and perceived product or service performance [71]. Positive disconfirmation occurs when products or services surpass consumer expectations, engendering satisfaction and supportive behaviors, whereas negative disconfirmation arises when products or services fall short of expectations, eliciting dissatisfaction and unsupportive actions [70,71]. While stakeholders engage with focal firms for varied purposes, they essentially follow analogous processes and assume similar roles [68]. Thus, based on the integrated theoretical framework of stakeholder theory and EDT, it is logical to extend EDT’s theoretical insights from consumer behaviors to the broader stakeholder group.

Existing literature underscores that stakeholders shape their expectations by referencing a baseline, referred to as a reference point, which serves as a basis for comparison [72,73,74]. EDT posits that consumers form their perceptions and expectations of a firm’s product or service quality based on available information sources, such as advertising, personal experiences, and word of mouth [71,75,76]. Stakeholders may rely on an internal reference point (i.e., the firm’s past performance) [77] or an external reference point (i.e., the performance of the firm’s industry peers) [78,79]. Following the correspondence theory posited by Ajzen and Fishbein (1980) [80], contrasting a firm’s past and present performance is considered more effective in evaluating the firm’s overall performance than comparing it to an industry standard. Thus, drawing on Ajzen and Fishbein (1980) [80], we infer that stakeholders may designate the firm’s CSP in the previous year as their reference point when forming their expectancy disconfirmation.

According to EDT, stakeholders will experience positive expectancy disconfirmation when the firm undertakes a positive move, such as increasing CSR (i.e., adopting a proactive strategy) or decreasing CSIR (i.e., implementing a rectification strategy), based on the firm’s CSP in the previous year. This positive expectancy disconfirmation is linked to stakeholder satisfaction, ultimately resulting in more favorable stakeholder decisions and behaviors toward the focal firm [40,81,82]. Hence, we formulate the following hypotheses:

H1a. 

Firms’ proactive strategy of increasing CSR (i.e., doing more good) is positively related to CFP.

H1b. 

Firms’ rectification strategy of decreasing CSIR (i.e., doing less bad) is positively related to CFP.

3.2. The Influence of Accumulative Stakeholders’ Expectancy Disconfirmation

Some researchers suggest that repeated interpersonal interactions evolve over time [83], and previous experiences, such as the experience of satisfaction, have a lasting impact [84]. Consequently, individual emotions like satisfaction accrue [85,86]. For example, Bolton (1998) [87] notes that substantial satisfaction stemming from numerous past interactions can accumulate and offset recent unsatisfactory experiences, particularly when customers encounter negative experiences that they deem unusual.

In line with this perspective, we contend that stakeholders accumulate positive experiences. They amass satisfaction through firms’ concurrent proactive and rectification strategies, driven by positive expectancy disconfirmation. We term the strategy of firms that pursue both doing more good and doing less bad simultaneously as an “aggressive strategy.” This strategy may require more corporate resources than opting for either a proactive or rectification strategy to maximize favorable outcomes.

The accumulated satisfaction stemming from an aggressive strategy should surpass that arising from a proactive or rectification strategy. Consequently, stakeholders who encounter a firm’s aggressive strategy will subsequently exhibit more favorable behaviors than those who only experience a firm’s proactive or rectification strategy. Building on this discussion, we present the second set of hypotheses:

H2a. 

Aggressive strategy (i.e., doing more good and doing less bad) is more effective than proactive strategy (i.e., doing more good) in enhancing CFP.

H2b. 

Aggressive strategy (i.e., doing more good and doing less bad) is more effective than rectification strategy (i.e., doing less bad) in enhancing CFP.

3.3. The Influence of Stakeholders’ Negativity Bias

Research in social psychology consistently demonstrates a pronounced bias in attention allocation toward negative information [88,89]. In other words, individuals tend to devote more attention to negative information than to positive information [90,91]. Negative information naturally commands individuals’ attention [92], a phenomenon known as the “negativity bias” in attention allocation [93]. The negativity bias is considered to be one of the most fundamental and pervasive psychological principles [94] (p. 362).

When we apply this negativity bias to the context of CSR/CSIR, it becomes evident that CSR and CSIR attract asymmetric stakeholder attention, with CSIR receiving more attention. This can be attributed to the fact that, in general, firms’ CSR activities generate positive information, whereas CSIR activities convey negative information [21,95]. Furthermore, according to EDT, increasing CSR and decreasing CSIR lead to stakeholder satisfaction. However, in light of stakeholders’ negativity bias, it is reasonable to infer that asymmetric attention allocation results in varying levels of satisfaction regarding firms’ efforts in CSR and CSIR. Subsequently, these differences in satisfaction translate into distinct favorable decisions and behaviors on the part of stakeholders toward firms. Building on this discussion, we propose the third hypothesis.

H3. 

Ceteris paribus, rectification strategy (i.e., reducing CSIR, doing less bad) is more effective in enhancing CFP than proactive strategy (i.e., increasing CSR, doing more good).

4. Methodology

4.1. Data Collection

The data for this study were collected from two primary sources: first, the MSCI ESG STATS (formerly known as Kinder Lydenberg Domini, KLD) for information on firms’ CSR and CSIR initiatives. The KLD database includes publicly listed U.S. firms in the S&P 500 index and the Domini 400 Social Index, making it a widely used and recognized source in the literature [96,97,98,99,100]; second, the Standard & Poor’s (S&P) Compustat database for financial data.

4.2. CSR and CSIR Indicators

The KLD data source provides a composite score based on ESG (environmental, social, and governance) categories. These categories encompass seven dimensions, namely environment, community, human rights, employee relations, diversity, products, and governance. Within each dimension, there are “strength” and “concern” indicators, resulting in approximately 70 indicators (sub-categories) across the seven dimensions (please refer to Appendix A, where we have listed detailed indicators for each dimension, encompassing both strength and concern indicators). These “strength” or “concern” indicators are binary variables, where a value of 1 indicates that a firm meets the criteria for the particular indicator, while a value of 0 indicates otherwise. We calculate a firm’s strength ratings (SR) and concern ratings (CR) separately across the seven dimensions on an annual basis, serving as indicators of the firm’s CSR and CSIR efforts.

4.3. Financial Data

The financial data used in this study were sourced from the Standard & Poor’s (S&P) Compustat database, which provides comprehensive financial information for publicly listed firms. From this data, we computed several financial metrics, including Tobin’s Q, market-to-book ratio, cash holdings, leverage, and return on assets (ROA).

4.4. Data Cleaning and Sample

In this research, we leverage KLD data to investigate sustainable strategies across various types of companies. The data’s credibility is rooted in its extensive coverage, spanning multiple industries and geographical regions within the United States, which ensures a representative sample reflective of U.S. corporate practices. This comprehensive dataset includes detailed environmental, social, and governance (ESG) metrics, which allow for nuanced analysis of company behaviors and strategies related to sustainability. The depth and breadth of the KLD data not only affirm their relevance and applicability to CSR research but also strengthen the validity of our sample. Therefore, the use of KLD data in our study is both reasonable and representative, providing a solid foundation for assessing the efficacy and diversity of sustainable strategies in the corporate sector.

After merging the data from KLD and S&P Compustat, and conducting necessary data cleaning, we removed 4202 observations with missing financial and CSR/CSIR data. The resulting dataset comprises 12,567 firm-year observations, representing 3422 unique firms across different sectors. KLD data has been collected since 1991. The observations that meet the group conditions are outlined in ‘

Table 1. The distribution of treatment and control groups.

Strategic Position	Condition	N
Proactive strategy (Treatment 1)	ΔSRit,t−1 = SRit-SRit-1 > 0 and ΔCRit,t−1 = CRit − CRit−1 = 0	1629
Rectification strategy (Treatment 2)	ΔSRit,t−1 = SRit-SRit-1 = 0 and ΔCRit,t−1 = CRit − CRit−1 < 0	2211
Aggressive strategy (Treatment 3)	ΔSRt,t−1 = SRit-SRit-1 > 0 and ΔCRit,t−1 = CRit − CRit−1 < 0	1302
No change in CSR/CSIR (Control Group)	ΔSRit,t−1 = SRit-SRit-1 = 0 and ΔCRit,t−1 = CRit − CRit−1 = 0	7425
Total		12,567

Note: SR stands for strength ratings and CR stands for concern ratings.

. The distribution of treatment and control groups, such as Proactive strategy (Treatment 1), Rectification strategy (Treatment 2), Aggressive strategy (Treatment 3), and No change in CSR/CSIR (Control Group), began to appear in the dataset from 1994 onwards. All continuous variables were winsorized at the 1st and 99th percentiles, except for CSR ratings. It should be highlighted that data beyond 2013 were not included due to significant changes in the sub-categories of the KLD database, leading to a substantial number of missing values. This aligns with the approach taken by previous literature that used data up to 2013 [101]. Consequently, the data cover the period from 1994 to 2013.

4.5. Variables

4.5.1. Dependent Variable—Change in Tobin’s Q

Tobin’s Q is a widely accepted measure of a firm’s growth potential and long-term profitability [42]. It is considered a forward-looking indicator of a firm’s future financial performance [102]. Since the effects of CSP are typically manifested in long-term CFP, we use Tobin’s Q as a proxy for CFP to investigate the CSP–CFP relationship [42,103]. Specifically, we employ the first difference in Tobin’s Q as the dependent variable (∆Tobin’s Q_it = Tobin’s Q_it − Tobin’s Q_it−1).

4.5.2. Independent Variable—Change in Strategic Positioning with CSR Initiatives

We establish our primary independent variable using the CSR/CSIR ratings from the previous year (t − 1) as the reference point. Firms are categorized into three strategic groups, proactive strategy, rectification strategy, or aggressive strategy, based on changes in strength and concern ratings. If a firm’s strength rating (SR) in year t increases compared to the rating in the previous year (t − 1), while the concern rating (CR) remains unchanged, we classify the firm as employing a proactive strategy (ΔSR_it,t−1 = SR_it − SR_it−1 > 0, ΔCR_it,it−1 = CR_it − CR_it−1 = 0). This group of firms constitutes Treatment 1, indicating an increase in CSR or doing more good. Conversely, if a firm’s concern rating (CR) in year t decreases compared to the rating in the previous year (t − 1), while the strength rating (SR) remains constant, we classify the firm as adopting a rectification strategy (ΔSR_it,t−1 = SR_it − SR_it−1 = 0, ΔCR_it,it−1 = CR_it − CR_it−1 < 0). This group of companies is identified as Treatment 2, reflecting a decrease in CSIR or doing less bad.

For firms taking an aggressive strategy, both the SR and CR change. If the SR in year t increases compared to the previous year (t − 1), and the CR in year t decreases compared to the previous year (t − 1), the firm is considered to be employing an aggressive strategy (ΔSR_it,t−1 = SR_it − SR_it−1 > 0, ΔCR_it,t−1 = CR_it − CR_it−1 < 0). This group of companies falls under Treatment 3, indicating they are simultaneously doing more good and doing less bad. Firms that do not experience any change in strength and concern ratings are designated as the control group (ΔSR_it,t−1 = SR_it − SR_it−1 = 0, ΔCR_it,it−1 = CR_it − CR_it−1 = 0). Table 1 provides an overview of the composition of these four distinct groups.

We employ the first-difference estimation approach to control the impacts of unobservable firm characteristics. This approach helps address the influence of unobservable factors. We assume a linear relationship between the dependent variable and the independent variables, as follows:

$$y_{it}=x_{it}\times \beta +{\delta }_i+{\epsilon }_{it}$$

The lagged value of y_it is $y_{it-1}=x_{it-1}\times \beta +{\delta }_i+{\epsilon }_{it-1}$. We take the difference of the dependent variable by subtracting y_it−1 from y_it:

$$y_{it}-y_{it-1}={(x}_{it}-x_{it-1})\ast \beta +{(\delta }_i-{\delta }_i)+({\epsilon }_{it}-{\epsilon }_{it-1})=∆y_{it}=∆x_{it}\ast \beta +∆{\epsilon }_{it}$$

(1)

The application of first-differencing in our model enables the production of consistent estimates for β by eliminating the influences of unobservable characteristics (δi) [104]. Following this methodology, we measure both the dependent and independent variables employed in our empirical analyses to assess the proposed hypotheses.

4.6. Empirical Model

To investigate our research question about how firms’ strategic positioning regarding CSR/CSIR affects their financial performance, we specify the following empirical model:

$$\begin{gathered} {∆Tobi{n}^{\prime }s Q}_{it}=\alpha +{\beta }_1\ast {Treatment1}_{it}+{\beta }_2\ast {Treatment2}_{it}+{\beta }_3\ast {Treatment3}_{it}+{\gamma }_1\ast {\mathit{log}\left(AT\right)}_{it}+{\gamma }_2 \\ \ast {Market-to-book}_{it}+{\gamma }_3\ast {Cash-holding}_{it}+{\gamma }_4\ast {Leverage}_{it}+{\gamma }_5\ast {ROA}_{it}+{\gamma }_6 \\ \ast {Lagged strength rating}_{it}+{\gamma }_7\ast {Lagged concern rating}_{it}+\sum {\delta }_j\ast {Iudustry Dummy}_j \\ +\sum {\theta }_t\ast {Year Dummy}_t+{\epsilon }_{it} , \end{gathered}$$

(2)

where Treatment1 is the firm group with a proactive strategy, Treatment2 is the firm group with a rectification strategy, and Treatment3 is the firm group with an aggressive strategy. To control for factors that could affect firm financial performance, we include several control variables based on prior literature [43,105]. These variables comprise the logarithm of total assets, market-to-book ratio, cash-holding, leverage, and return on assets (ROA).

The logarithms of total assets and ROA serve as measures of firm size and profitability, respectively, while the market-to-book ratio reflects growth opportunities. We incorporate these controls to mitigate the influence of differences in firm size, profitability, and growth opportunities on Tobin’s Q, as observed in previous studies [102,106,107,108].

Additionally, we consider cash holdings and leverage to address selection bias in our research, controlling for any potential effects of variations in these variables across firms on firm performance [109,110].

Furthermore, we include lagged strength and concern ratings (i.e., ratings from the previous year, t − 1) to capture the long-term impact of firms’ CSR investments on financial performance [11,12,111,112].

To account for specific temporal and industry effects, our empirical model incorporates year-specific and industry-specific dummy variables. In this context, ‘i’ represents firms, ‘j’ represents industries based on SIC-2 digits, and ‘t’ represents years. Further details regarding variable definitions can be found in

Table 2. Definition of main variables.

Data Sources	Variable	Definition
KLD STATS	Strength	Sum of all the strengths (strength ratings: SR) over the seven dimensions: environment, community, human rights, employee relations, diversity, products, and governance
	Weakness	Sum of all the concerns (concern ratings: CR) over the seven dimensions: environment, community, human rights, employee relations, diversity, products, and governance (The details of the dimensions are provided in the Appendix A (Table A1))
Compustat database	Tobin’s Q	$(\mathrm{A}\mathrm{T}-\mathrm{M}\mathrm{V}\mathrm{E}-\mathrm{C}\mathrm{E}\mathrm{Q}-\mathrm{T}\mathrm{X}\mathrm{D}\mathrm{B})$/AT
	AT	Total Assets
	MVE	Market value measured as common shares outstanding (CSHO) divided by closing price (PRCC_F)
	CEQ	Book value of equity
	TXDB	Deferred taxes
	Market-to-Book ratio	Market value (MVE) divided by book value (CEQ)
	Cash-holding	Cash and marketable securities (CHE) divided by total assets (AT)
	Leverage	Long-term debt (DLTT) divided by total assets (AT)
	ROA	Net income (NI) divided by total assets (AT)

.

Table 3. Descriptive statistics.

Variable	N	Mean	STD	p25	Median	p75
∆Tobin’s Q	12,567	−0.06	0.86	−0.21	−0.00	0.17
log(AT)	12,567	7.37	1.63	6.16	7.29	8.44
Market-to-book	12,567	3.30	9.60	1.39	2.13	3.42
Cash-holding	12,567	0.17	0.21	0.02	0.09	0.24
Leverage	12,567	0.18	0.19	0.01	0.14	0.29
ROA	12,567	0.02	0.15	0.01	0.04	0.08
Lagged strength rating	12,567	1.29	2.17	0.00	1.00	2.00
Lagged concern rating	12,567	2.06	1.92	1.00	2.00	3.00

provides descriptive statistics for the primary variables utilized in this study. The mean (median) change in Tobin’s Q is −0.06 (0.00), while the mean (median) ROA is 0.02 (0.04).

Additionally, it is worth noting that the mean values of the logarithm of total assets (log(AT)), market-to-book ratio, and cash-holding suggest that the sample firms are predominantly large companies. This observation aligns with the fact that KLD STATS encompasses the largest 3000 publicly traded companies based on market capitalization.

Table 4. Correlation matrix.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1) ∆Tobin’s Q	1.00
(2) Log(AT)	0.04	1.00
(3) Market-to-book	0.07	−0.04	1.00
(4) Cash-holding	−0.03	−0.35	0.07	1.00
(5) Leverage	0.02	0.31	0.04	−0.17	1.00
(6) ROA	−0.06	0.19	−0.03	−0.16	−0.04	1.00
(7) Lagged strength rating	0.00	0.44	0.02	0.01	0.09	0.08	1.00
(8) Lagged concern rating	0.03	0.42	0.00	−0.04	0.16	0.04	0.35	1.00

Notes: Pearson correlation coefficients are reported. Correlation coefficients in bold are statistically significant at a significance level of p < 0.01.

presents the correlation matrix among the primary variables. Notably, Tobin’s Q exhibits a positive correlation with the logarithm of total assets, market-to-book value, and leverage, indicating that larger companies with a higher book-to-market ratio and leverage ratio tend to have higher Tobin’s Q.

Furthermore, the correlation coefficients between Tobin’s Q and the strength rating and concern rating in t − 1 are positive. However, it is worth noting that the relationship between Tobin’s Q and strength ratings is statistically insignificant. This suggests a potentially long-term association between Tobin’s Q and the concern rating.

In the subsequent section, we conduct a multivariate analysis, incorporating various control variables to further explore these relationships.

5. Results

Table 5. Empirical results of the impacts of different strategies.

Variable	DV: ∆Tobin’s Q
	Model (1)	Model (2)
Treatment1 (Proactive Strategy)	0.06 *** (0.02)	0.05 ** (0.02)
Treatment2 (Rectification Strategy)	0.13 *** (0.02)	0.13 *** (0.02)
Treatment3 (Aggressive strategy)	0.16 *** (0.02)	0.15 *** (0.02)
log(AT)	0.03 *** (0.01)	0.03 *** (0.01)
Market-to-book	0.01 * (0.00)	0.01 * (0.01)
Cash-holding	−0.12 * (0.06)	−0.08 (0.07)
Leverage	−0.02 (0.04)	−0.07 (0.05)
ROA	−0.45 *** (0.16)	−0.49 *** (0.16)
Lagged strength rating	−0.01 (0.01)	−0.01 (0.01)
Lagged concern rating	−0.01 (0.01)	−0.01 (0.01)
Industry-fixed Dummy	-	Included
Year-fixed Dummy	-	Included
Constant	Included	Included
R2	1.96%	3.84%
Number of Observations	12,567	12,567

Notes: Standard errors are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

presents the empirical results based on Equation (2). In the first model (Model 1), we do not consider industry and year-specific dummy variables. In the second model (Model 2), we include those dummy variables in the equation.

In Model 1, we observe that the estimated coefficients for all three strategies firms may adopt are significantly positive (β_{Treatment1(Proactive Strategy)} = 0.06, p-value < 0.01; β_{Treatment2(Rectification Strategy)} = 0.13, p-value < 0.01; β_{Treatment3(Aggressive Strategy)} = 0.16, p-value < 0.01). These results indicate that all three strategies effectively improve Tobin’s Q when compared to the control group (firms with no changes in CSR/CSIR).

In Model 2, where we include industry and year-specific dummy variables, the three strategies remain effective in improving Tobin’s Q (β_{Treatment1(Proactive Strategy)} = 0.05, p-value < 0.01; β_{Treatment2(Rectification Strategy)} = 0.13, p-value < 0.01; β_{Treatment3(Aggressive Strategy)} = 0.15, p-value < 0.01).

These findings suggest that firms’ proactive and rectification strategies are positively associated with Corporate Financial Performance (CFP), indicating that efforts to engage in socially responsible practices can lead to financial benefits. Therefore, our hypotheses, H1a and H1b, are supported.

The estimated coefficient of the aggressive strategy is higher than those of the other two strategies, indicating that firms can achieve greater financial benefits when they simultaneously engage in both positive and negative CSR/CSIR actions. These results provide initial support for hypotheses H2a and H2b.

In the following sections, we will conduct pairwise comparison analyses to further examine whether hypotheses H2a and H2b are supported. Pairwise comparison analysis is a valuable method for comparing and ranking a set of options or alternatives. This approach helps mitigate bias and provides a structured framework for decision-making. It has been widely utilized in academic research [113].

To test H3, we examine whether the two estimated coefficients in Model (2) of Table 5 (a more comprehensive model including fixed effects) are equal to each other (β_{Treatment1(Proactive Strategy)} = β_{Treatment2(Rectification Strategy)}). To do this, we conduct a Wald test with a null hypothesis that βTreatment1 is equal to βTreatment2 following the regression estimation of Model (2). The Wald test result reveals an F-value of 8.64, with a corresponding p-value of 0.00, indicating the rejection of the null hypothesis. Consequently, we conclude that the estimated coefficient of Treatment 2 (Rectification Strategy) is statistically and significantly greater than that of Treatment 1 (Proactive Strategy). This result provides support for H3, indicating that the rectification strategy is more effective in enhancing firm performance than the proactive strategy.

When interpreting the estimated coefficients of the control variables in Equation (2), several findings emerge. First, we observe that the logarithm of total assets (log(AT)) has a positive influence on the change in Tobin’s Q (β_log(AT) = 0.03, p-value < 0.01). This suggests that larger firms are more likely to exhibit higher profitability and future growth potential.

Second, the market-to-book ratio shows a positive relationship with the change in Tobin’s Q (β_{Market-to-book} = 0.01, p-value < 0.10 in Model (2)). On the other hand, return on assets (ROA) demonstrates a negative impact on the increase in Tobin’s Q (β_ROA = −0.49, p-value < 0.01 in Model (2)). These results imply that, generally, firms with higher market-to-book ratios tend to be more profitable, while firms with higher ROA values are efficient in asset utilization, which leads to a diminishing effect on the increase in performance (change in Tobin’s Q).

5.1. Pairwise Comparison—Proactive versus Aggressive Strategies

To empirically test H2a, we conducted pairwise comparison analyses. In this section, we compared the effectiveness of two strategy groups, namely the proactive and aggressive strategies. To do this, we excluded rectification strategy firms from the sample, resulting in a dataset that includes only proactive and aggressive strategy firms. We constructed the variable ‘Aggressive Strategy’, assigning a value of 1 for firms adopting an aggressive strategy and 0 for those employing a proactive strategy. Consequently, proactive strategy firms served as the baseline group. Each column incorporated the same control variables as in the previous analyses presented in Table 5.

The results of the pairwise comparison between proactive and aggressive strategies are displayed in

Table 6. Empirical results of the impacts of proactive versus aggressive strategies.

Variable	DV: ∆Tobin’s Q
	Model (1)	Model (2)
Aggressive Strategy	0.10 *** (0.02)	0.06 *** (0.03)
log(AT)	0.01 (0.01)	0.01 (0.01)
market-to-book	0.01 (0.01)	0.01 (0.01)
Cash-holding	0.04 (0.14)	−0.09 (0.16)
Leverage	−0.01 (0.07)	−0.01 (0.10)
ROA	−1.27 *** (0.54)	−1.42 *** (0.55)
lagged strength rating	0.01 (0.01)	0.01 (0.01)
lagged concern rating	−0.01 (0.01)	0.01 * (0.00)
Industry-fixed Dummy	-	Included
Year-fixed Dummy	-	Included
Constant	Included	Included
R2	3.59%	10.36%
Number of Observations	2931	2931

Notes: Standard errors are in parentheses. ***, and * indicate significance at the 1% and 10% level, respectively.

. In the first column of Model (1), the estimated coefficient of the aggressive strategy is significantly positive (β_{Aggressive Strategy} = 0.10, p-value < 0.01). This finding remains consistent in the second column (β_{Aggressive Strategy} = 0.06, p-value < 0.01) for Model (2), which controls for industry-fixed and year-fixed effects. Based on these results, we can conclude that an aggressive strategy is more effective than a proactive strategy in enhancing Tobin’s Q, indicating that simultaneously increasing CSR and reducing CSIR can benefit the firm more than increasing CSR alone. Therefore, H2a is supported.

5.2. Pairwise Comparison—Rectification versus Aggressive Strategy

To test hypothesis H2b, we compared the effects of two strategies, namely the rectification and aggressive strategies. The analysis sample consisted of firms employing either rectification or aggressive strategies. We constructed the variable ‘Aggressive Strategy’, assigning a value of 1 for firms adopting an aggressive strategy and 0 for those employing a rectification strategy. Firms implementing a rectification strategy were used as the baseline group in the analysis. Both models in

Table 7. Empirical results of the impacts of rectification versus aggressive strategies.

Variable	DV: ∆Tobin’s Q
	Model (1)	Model (2)
Aggressive Strategy	0.07 *** (0.02)	0.06 ** (0.03)
log(AT)	−0.01 (0.01)	−0.01 (0.01)
Market-to-book	0.01 (0.01)	0.01 (0.01)
Cash-holding	0.08 (0.10)	0.05 (0.10)
Leverage	−0.14 *** (0.05)	−0.24 *** (0.07)
ROA	−1.03 *** (0.34)	−1.07 *** (0.35)
Lagged strength rating	0.01 (0.01)	0.01 (0.01)
Lagged concern rating	0.01 (0.01)	0.01 ** (0.01)
Year-fixed Dummy	-	Included
Industry-fixed Dummy	-	Included
Constant	Included	Included
R2	5.01%	8.90%
Number of Observations	3513	3513

Notes: Standard errors are in parentheses. ***, and ** indicate significance at the 1% and 5%level, respectively.

included the same control variables as in the previous analyses.

In the first model (1) of Table 7, which did not consider industry and year dummies, the estimated coefficient for the aggressive strategy was significantly positive (β_{Aggressive Strategy} = 0.07, p-value < 0.01). Similarly, in the second column (2), the estimated coefficient (β_{Aggressive Strategy} = 0.06, p-value < 0.05) was also significant and positive. These results indicate that the aggressive strategy outperforms the rectification strategy, suggesting that when companies aim to enhance Tobin’s Q, it is more advantageous to increase CSR and decrease CSIR simultaneously rather than solely reducing CSIR. Therefore, hypothesis H2b is supported.

In summary, based on the analyses presented in Table 6 and Table 7, the aggressive strategy is shown to be more effective than both the proactive and rectification strategies in improving Tobin’s Q. In the following section, we directly compare the two groups of firms employing either a proactive or rectification strategy to gain insights into which strategy is more effective when firms face constraints and can choose between doing more good or doing less bad.

5.3. Pairwise Comparison—Proactive versus Rectification Strategies

In this section, the sample comprises firms employing either a proactive or rectification strategy exclusively. We created the variable ‘Rectification Strategy’, assigning a value of 1 for firms adopting a rectification strategy and 0 otherwise, with firms implementing a proactive strategy serving as the baseline group. The results are presented in

Table 8. Empirical results of the impacts of proactive versus rectification strategies.

Variable	DV: ∆Tobin’s Q
	Model (1)	Model (2)
Rectification Strategy	0.03 (0.02)	0.04 * (0.02)
log(AT)	0.01 (0.01)	0.01 (0.01)
Market-to-book	0.01 ** (0.00)	0.01 ** (0.00)
Cash-holding	−0.04 (0.10)	−0.08 (0.10)
Leverage	−0.09 (0.05)	−0.15 ** (0.07)
ROA	−1.07 *** (0.27)	−1.13 *** (0.28)
Lagged strength rating	−0.01 (0.01)	0.01 (0.01)
Lagged concern rating	0.01 (0.01)	0.02 *** (0.01)
Industry-fixed Dummy	-	Included
Year-fixed Dummy	-	Included
Constant	Included	Included
R2	5.05%	7.01%
Number of Observations	3840	3840

Notes: Standard errors are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

. The first column of Model (1) displays the outcomes without considering industry and year-specific dummy variables. In Model (1), the estimated coefficient is positive but not statistically significant (β_{Rectification Strategy} = 0.03, p-value > 0.10). Model (2) incorporates industry and year-specific dummy variables, leading to a significantly positive estimated coefficient (β_{Rectification Strategy} = 0.04, p-value < 0.10). While the significance level may not be high, this result from the pairwise comparison suggests that the rectification strategy is more effective in enhancing Tobin’s Q. Therefore, hypothesis H3 receives support.

5.4. Robustness Check

We conducted three additional tests based on Equation (2) to verify the robustness of our empirical findings. First, we performed a regression analysis using Newey–West estimation to control for potential autocorrelation and heteroskedasticity effects. The results are presented in

Table 9. Empirical results of the impacts of different strategies with Newey–West, bootstrap, and median based estimations.

Variable	DV: ∆Tobin’s Q
	Newey-West		Bootstrap		Quantile
	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)	Model (6)
Treatment1 (Proactive)	0.06 *** (0.02)	0.05 ** (0.02)	0.06 *** (0.02)	0.05 ** (0.02)	0.02 *** (0.00)	0.02 *** (0.00)
Treatment2 (Rectification)	0.13 *** (0.02)	0.13 ** (0.02)	0.13 *** (0.02)	0.13 *** (0.01)	0.03 *** (0.00)	0.03 *** (0.00)
Treatment3 (Aggressive)	0.16 *** (0.02)	0.15 *** (0.02)	0.16 *** (0.02)	0.15 *** (0.02)	0.04 *** (0.00)	0.05 *** (0.00)
log(AT)	0.03 *** (0.00)	0.03 *** (0.01)	0.03 *** (0.00)	0.03 *** (0.01)	0.01 *** (0.00)	0.01 *** (0.00)
Market-to-book	0.00 * (0.00)	0.00 * (0.00)	0.00 (0.00)	0.00 (0.00)	0.01 ** (0.00)	0.01 ** (0.00)
Cash-holding	−0.12 ** (0.06)	−0.07 (0.06)	−0.12 * (0.06)	−0.07 (0.07)	−0.00 (0.03)	0.01 (0.02)
Leverage	−0.01 (0.03)	−0.07 (0.05)	−0.01 (0.04)	−0.07 * (0.04)	0.01 (0.01)	−0.02 (0.01)
ROA	−0.45 *** (0.16)	−0.49 *** (0.16)	−0.45 *** (0.17)	−0.49 *** (0.16)	−0.15 *** (0.04)	−0.18 *** (0.05)
Lagged strength rating	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 *** (0.00)	−0.00 *** (0.00)
Lagged concern rating	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 ** (0.00)
Industry-fixed Dummy	-	Included	-	Included	-	Included
Year-fixed Dummy	-	Included	-	Included	-	Included
Constant	Included	Included	Included	Included	Included	Included
F-value	10.90	4.80	-	-	-	-
p-value	0.00	0.00	-	-	-	-
R2	-	-	1.96%	3.84%	-	-
Pseudo—R2	-	-	-	-	0.83%	1.96%
Number of Observations	12,567	12,567	12,567	12,567	12,567	12,567

Notes: Standard errors are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

, and they align with the outcomes in Table 5. Newey–West standard errors were used for coefficients in the first two models (1) and (2). The estimated coefficient for Treatment 1 (Proactive strategy) is significantly positive (β_Treatment1 = 0.06, p-value < 0.01 in Model (1), and β_Treatment1 = 0.05, p-value < 0.05 in Model (2)). Similarly, the estimated coefficients for Treatment 2 (rectification strategy) are positive (β_Treatment2 = 0.13, p-value < 0.01 in Model (1), and β_Treatment2 = 0.13, p-value < 0.05 in Model (2)). Notably, the estimated coefficients for Treatment 3 (aggressive strategy) remain significantly positive and greater than those of Treatment 1 (Proactive) and Treatment 2 (rectification) (β_Treatment3 = 0.16, p-value < 0.01 in Model (1), and β_Treatment3 = 0.15, p-value < 0.01 in Model (2)).

Secondly, to further substantiate our consistent empirical findings, we employed nonparametric bootstrap estimation, providing a robust alternative to classic parametric estimation, particularly when considering potential deviations from the normality of the error term. We conducted 400 bootstrap replications in line with prior literature recommendations [114]. The results from our bootstrap estimation reaffirmed the robustness of our conclusions. The outcomes are presented in the subsequent two models, (3) and (4). In Model (3), the estimated coefficients for Treatment 1 (Proactive strategy), Treatment 2 (Rectification strategy), and Treatment 3 (Aggressive strategy) all remain significantly positive (β_Treatment1 = 0.06, p-value < 0.01, β_Treatment2 = 0.13, p-value < 0.01, and β_Treatment3 = 0.16, p-value < 0.01, respectively in Model (3)). When industry- and year-fixed effects are included in the model specification, the estimated coefficients continue to be statistically significant and positive (β_Treatment1 = 0.05, p-value < 0.05, β_Treatment2 = 0.13, p-value < 0.01, and β_Treatment3 = 0.15, p-value < 0.01, respectively in Model (4)).

Thirdly, we assessed the sensitivity of our results through median-based estimation using Quantile Regression. In Quantile Regression, there is no strict requirement for the assumption of the distribution of residuals, making it a valuable approach for robustness checks. The estimated coefficients for Treatment 1 (Proactive strategy), Treatment 2 (Rectification strategy), and Treatment 3 (Aggressive strategy) all remained significantly positive (β_Treatment1 = 0.02, p-value < 0.01, β_Treatment2 = 0.03, p-value < 0.01, β_Treatment3 = 0.04, p-value < 0.01, respectively in Model (5)). The estimated coefficients for Treatment 1 (Proactive strategy), Treatment 2 (Rectification strategy), and Treatment 3 (Aggressive strategy) in Model (6) also remained significantly positive when we introduced industry- and year-fixed effects (β_Treatment1 = 0.02, p-value < 0.01, β_Treatment2 = 0.03, p-value < 0.01, β_Treatment3 = 0.05, p-value < 0.01, respectively in Model (5)). The results from Models (5) and (6) in Table 9 provide strong evidence that our empirical findings are robust, with no significant deviations attributable to the median-based estimation.

Table 10. Summary of hypothesis test.

	Hypothesis	Test Results
H1a	Firms’ proactive strategy of increasing CSR (i.e., doing more good) is positively related to CFP.	Supported
H1b	Firms’ rectification strategy of decreasing CSIR (i.e., doing less bad) is positively related to CFP.	Supported
H2a	Aggressive strategy (i.e., doing more good and doing less bad) is more effective than proactive strategy (i.e., doing more good) merely.	Supported
H2b	Aggressive strategy (i.e., Doing more good and doing less bad) is more effective than rectification strategy (i.e., doing less bad) merely.	Supported
H3	Ceteris paribus, rectification strategy (i.e., reducing CSIR, doing less bad) leads to more financial benefits than proactive strategy (i.e., increasing CSR, doing more good).	Supported

provides a summary of the hypothesis testing outcomes.

6. Discussion

Firms are increasingly engaging in both CSR and CSIR initiatives, often using CSR to mitigate the adverse effects of CSIR [20]. A comprehensive study spanning 15 years and involving 3000 publicly listed US firms conducted by Kotchen and Moon (2012) [22] reveals that firms frequently employ CSR to counterbalance their CSIR activities. Scholars have emphasized that CSIR poses a significant challenge for business practitioners, scholars, and society at large, as it can be detrimental to stakeholders, society, and the environment [115,116]. Consequently, it becomes imperative to provide compelling guidance for managers to develop effective strategies concerning their CSR and CSIR endeavors. Nevertheless, the existing literature has thus far not yet delivered such guidance, as it has yielded inconsistent and, consequently, inconclusive findings regarding the impact of CSR and CSIR on CFP.

In the pursuit of offering valuable and compelling guidance to businesses in shaping their CSR/CSIR strategies, this study has developed a comprehensive theoretical framework. This framework aims to elucidate the pivotal role that firms’ CSR/CSIR strategies play in achieving financial gains by taking into account stakeholder responses and behaviors. The underpinning of this framework draws from EDT and Stakeholder Theory while incorporating a theoretical tradition rooted in the examination of negativity bias. Through this integrative framework, the study discerns three distinct strategic positioning alternatives for firms to adopt: the proactive strategy (entailing an emphasis on doing more good by augmenting CSR), the rectification strategy (focusing on doing less harm by mitigating CSIR), and the aggressive strategy (simultaneously pursuing both enhanced positive impact and reduced negative impact).

Subsequently, we bolster our research with empirical evidence aimed at discerning the most effective strategy for firms to leverage their socially responsible endeavors. The empirical results substantiate that the rectification strategy, centered on reducing negative impacts, outperforms the proactive strategy, emphasizing positive contributions. Additionally, the findings underscore the exceptional value of the aggressive strategy, where firms simultaneously focus on both amplifying their positive impact and diminishing negative effects. Notably, this comprehensive approach surpasses the individual strategies of doing more good or doing less bad, respectively.

7. Conclusions

This study is among the pioneering endeavors that systematically investigate the available strategic options concerning firms’ CSR/CSIR, offering empirical insights that underscore the superior efficacy of minimizing negative impact compared to accentuating positive contributions.

In a theoretical context, we introduce EDT to the domain of the CSP-CFP relationship. This adaptation extends the purview of EDT, shifting from its original application to customer dynamics and broadening its scope to encompass a diverse array of stakeholders. This expansion facilitates the inclusion of an intermediary factor known as stakeholder expectancy, which plays a pivotal role in substantiating our assertions. It enhances our understanding of the CSP-CFP link and advances the EDT framework within CSR literature. Consequently, our endeavor makes a noteworthy contribution to the evolving landscape of EDT and the broader field of CSR literature.

Furthermore, this study responds aptly to the imperative to amplify our focus on CSIR, as advocated by scholars such as Greenwood (2007) [117] and Lin-Hi and Müller (2013) [11]. It contributes significantly to the CSR/CSIR-CFP research domain by scrutinizing and contrasting the financial implications of firms’ three distinct strategic options concerning CSR/CSIR. This research departs from the conventional discourse centered on whether firms should intensify their CSR efforts to encompass the equally crucial inquiry of whether firms should mitigate their CSIR activities. By examining the financial implications of firms’ strategic choices regarding CSR/CSIR, we depart from the conventional discourse and highlight the crucial disparity between CSR and CSIR effects on financial performance. Most notably, this study elucidates three strategic positioning alternatives for firms, particularly those concurrently involved in CSR and CSIR, to inform their CSR/CSIR strategies.

Moreover, the outcomes of this study carry profound implications for business practitioners, which, in turn, hold significant relevance for the environment and society. On one hand, the findings provide managerial insights, highlighting the importance of prioritizing the rectification strategy and focusing on reducing CSIR as an initial step. On the other hand, these implications extend beyond the corporate sphere and are critically significant for the environment and society at large. CSIR incidents, such as BP’s oil spill and Johnson & Johnson’s harmful baby powder, have demonstrated the catastrophic consequences they can inflict [26]. Thus, addressing and mitigating CSIR to avert or minimize these dire repercussions take precedence and hold more substantial meaning than merely intensifying CSR efforts. This underscores the urgency of addressing the ‘doing less bad’ aspect, which ultimately serves the greater good for both businesses and society.

8. Limitation

It is crucial to acknowledge the limitations of this study. A primary constraint stems from the data source employed, namely the KLD data. This data source lacks the granularity to differentiate the magnitude of benefits and harms caused by strength (interpreted as CSR) and concerns (interpreted as CSIR). As noted, “Corporations perform actions that can inflict harm with different levels of intensity, from death to material loss, to both companies’ internal and external stakeholders” [118] (p. 285). To enhance the robustness of our findings, future research endeavors could benefit from utilizing data sources that provide a more nuanced perspective on both CSR and CSIR, enabling a more comprehensive assessment of their impact.

Furthermore, it’s worth noting that the KLD data employed in our study does not allow for a nuanced assessment of whether firms’ efforts in doing more good or doing less bad are substantive or merely symbolic, nor does it distinguish between voluntary and forced actions. As highlighted by Clark et al. (2022) [54], rectification efforts can be either voluntary or compelled. Subsequent research should strive to capture the distinction between these two forms of behavior related to doing more good and doing less bad. Moreover, our study treats all stakeholders as a homogeneous group, assuming uniform responses to firms’ CSR and CSIR activities. In reality, different stakeholders, such as customers and investors, may exhibit widely varying responses. Future investigations should aim to incorporate the diversity of stakeholders’ interpretations and reactions when data that allows for such differentiation becomes available. This approach would provide a more comprehensive understanding of how different stakeholders engage with firms’ CSR and CSIR strategies.

Another notable limitation of our study is its focus on establishing connections between doing more good and/or less bad and CFP while controlling for key firm-level variables. We do not delve into the exploration of potential mediating mechanisms or moderating factors. Subsequent research endeavors could enhance our understanding of firms’ CSR/CSIR strategies and their financial performance by investigating these mediating and moderating effects in greater detail. Furthermore, it’s worth noting that our study primarily offers insights into the financial impacts of doing more good and/or less bad, providing a set of strategic options and a recommended prioritization. Nonetheless, addressing and reducing CSIR, such as the reduction of sewage discharge, can be complex and may necessitate advanced technology and effective management. Our study falls short of providing concrete and practical suggestions for managers on how to implement these strategies effectively. Future studies could provide more specific, actionable recommendations to assist managers in implementing these strategies. In addition, the analyses in this study are grounded in a sample of firms from the United States. It remains uncertain whether the findings can be extrapolated to other contexts, such as those in Asian or European nations. Subsequent research endeavors could employ analogous methodologies to assess the broader applicability of the study’s conclusions.