Position Paper, Values, ethics and sustainability shape much of the more interesting news events. Scan the press, magazines, on-line news sources, idea handout sheets, and TV, and choose a topic from

Despite the importance of the managerial role in implementing CSR, prior studies generally have treated the role of managers as an exogenous factor. This study builds on recent studies on the managerial characteristics in studies on CSR by examining how managerial eciency inuences the outcomes of CSR. Using a newly developed measure of managerial eciency, we nd that, on average, managerial eciency is positively associated with a subsequent change in corporate social performance (CSP), although the association is weak in the level of total CSP. We nd that ecient managers are more likely to engage in the product-related CSR that directly connects to corporate nancial performance (CFP) but are less likely to engage in environment-related CSR. We also nd that CSP is positively associated with CFP with ecient managers. Our ndings contribute to management and other stakeholders’ understanding of the association of CSR to its outcomes, CSP and/or CFP, which is hinged by the indispensable moderating role of managerial eciency.

Keywords Managerial eciency · Corporate social responsibility · Corporate social performance · Corporate nancial performance · Ethical corporate social responsibility JEL Classi cation G30 · G34 · L21 · M14 * Cheol Lee [email protected] Seong Y. Cho [email protected] 1 Department of Accounting and Finance, School of Business Administration, Oakland University, Rochester, MI, USA 2 Department of Accounting, Mike Ilitch School of Business, Wayne State University, 207 Rands House, 5229 Cass Avenue, Detroit, MI 48202, USA Abbreviations Corporate social responsibility (CSR) CSR refers to a rm’s engage - ment in areas that benet society in general, such as environ - mental protection, community contribution, customer relation - ship, labor issues, and diversity of employment Corporate social performance (CSP) CSP refers to the outcome (or overall quality) of a rm’s pro - grams or investment in CSR and is judged by a third-party rating agency or organization in lieu of stakeholders’ assessments Corporate nancial performance (CFP) CFP represents rm value in terms of the joint e ect of nan - cial (tangible) and nonnancial (intangible) value drivers Introduction Over the past few decades, corporate social responsibility (CSR) activities have become increasingly essential to a rm’s operation. While there is no single agreed-upon de - nition, CSR has traditionally referred to actions to enhance social goods at least partially beyond the rm’s immediate Vol.:(0123456789) 1 3 468 S. Y. Cho, C. Lee economic interests. It can be narrowly limited to purely vol- untary philanthropic or ethical engagements to society but CSR may extend to actions triggered by economic orienta- tion and even legal requirements (e.g., Carroll 1991, 2016).

In this study, consistent with Carroll’s view ( 1991, 2016 ), we take a comprehensive denition of CSR including all exter - nal and internal actions involving environmental protection, community contribution, customer relationship, labor issues, product quality, and diversity of employment. Also, along with this comprehensive denition of CSR, we view that several di erent incentives of CSR can coexist in a rm’s CSR activity. Indeed, the importance of CSR goes beyond the philan- thropic endeavors, which was treated as an integral part of CSR, and expands to a rm’s strategic and business case for CSR (e.g., economic-oriented CSR). For example, as demonstrated by Servaes and Tamayo (2013) and Lev et al.

( 2010), customer loyalty tends to be signicantly stronger toward corporations that incorporate CSR. Moreover, assets of an international network of CSR-focused investors, Prin- ciples for Responsible Investment, increased to a total of $59 trillion in fiscal year 2015. To match this growing demand for CSR in recent years, more managers of many U.S. rms have begun allocating resources to CSR more than ever before. 1 However, even for top managers such as the chief executive ocer (CEO), who has the most control over resource allocation, it is dicult to pursue CSR because most CSR investment is frequently denigrated or criticized due to limited evidence that such spending has a positive inuence on the rm’s long-term protability or survival.

Moreover, CEOs are confronted by the dilemma in CSR investment when they deal with acute conicts and compe- titions among di erent stakeholders. For example, corporate pressure on suppliers can violate human rights by pushing down workers’ compensation of suppliers, mishandled labor relations can cause discrimination lawsuits, or a large-scale industrial dump can cause irreversible environmental dam- age. In such cases, business ethics can give some guidance in resolving the outright conicts. Therefore, as powerful and nal decision makers, most CEOs constantly calibrate their decisions to align with socially acceptable ethical standards.

Obviously, investment in CSR is one area that involves busi- ness ethics. Considering the importance of business ethics in CSR investment, a long-standing concern in academia is whether CSR activities provide nancial gain to the rm. Many researchers have investigated the consequences of CSR by focusing on corporate social performance (CSP), 2 which is rated by independent third-party agencies, through the lens of corporate nancial performance (CFP) such as account- ing earnings. The ndings from prior studies on the rela- tionship between CSP and CFP, however, are mixed. For example, Aupperle et al. (1985) nd no relation between the two, whereas Barnett and Salomon (2006) document a positive relation between CSP and CFP. On the contrary, Lopez et al. (2007) show a negative relation between CSP and CFP. Such a lack of conclusive evidence may be attrib- utable to the seemingly contradictory incentives that drive CSR. That is, the managerial incentives that precipitate CSR spending are encapsulated in a wide spectrum, dominated on one extreme by purely prot-oriented motivations and dominated on the other by altruistic motivations. Another reason for the mixed nding of the relation between CSP and CFP is that CFP does not work as a proxy to capture the e ect of CSP beyond direct monetary benets. Accord- ingly, understanding the underlying managerial motivations of various CSR and their consequences is important to shed light on the relation between CSP and CFP. Stakeholder theory (Jensen 2002) suggests that a rm engages in CSR with prot-oriented strategic objectives in response to the needs of customers, investors, employees, and other stakeholders, which may incur short-term costs but will result in social net benets as well as the rm’s long- term prot (Benabou and Tirole 2010). Prior studies sug- gest that managers’ strategic use of CSR in business opera- tions increases rm value (e.g., Baron 2001; McWilliams and Siegel 2001; Bagnoli and Watts 2003). Furthermore, nonmonetary dimensions are to be considered as benets of CSR. For example, rms with a high level of CSP may face fewer cases of labor-related litigation, gain recognition in the community via donations, retain the loyalty of customers with higher quality products, and promote employee morale through better labor relationships. On the contrary, managers’ altruistic CSR stems from their eagerness to be sincere and socially responsible citi- zens. As Baron (2001) suggests, the altruistic manager cares so much about pollution that she may spend extra money to reduce pollutants far below a compliance level that is already good enough to comply with regulation and meet stakehold- ers’ expectations. Thus, in contrast with the prot-oriented 1 The Principles for Responsible Investment is the international network of CSR investment signatories cofounded by the European Commission and supported by the United Nations. Currently, more than 1300 signatories are involved in CSR investment. See http:// www.unpri.org for information on CSR investment. 2 Unlike CSR, which refers to a rm’s programs or investment regarding social issues, CSP represents the overall quality of such CSR-related investments as judged by a third-party rating agency or organization in lieu of stakeholders’ assessments, because each stake- holder has a limit to accessing and evaluating the full range of CSR activities. In addition, in contrast to CSR’s nature as a one-time com- mitment, CSP is cumulative and composed of continual CSR invest- ment (Barnett 2007).

1 3 469 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance strategic CSR aimed at enhancing monetary or nonmonetary benets, the underlying motivation of altruistic CSR may be disconnected from a rm’s economic gains. Prior literature (e.g., Baron 2001; Baron and Diermeier 2007; Hillman and Keim 2001) demonstrates that CSR implemented by altruis- tic managers may achieve high CSP but does not necessarily increase rm value. Baron and Diermeier (2007) view this kind of CSR cost as “an imperfect substitute for individual social giving” (p. 685). As a middle ground of two contrasting motivations, Car - roll (1991) proposes another form of social responsibility incentive that is rooted in the fulllment of legal require- ments or ethical responsibilities. He denes this type of CSR as performing actions in a manner consistent with various expectations of regulators or standards which are not codi- ed (Carroll 2016). Therefore, this kind of ethical or socially responsible CSR can partly provide nonmonetary benets by reducing the risk of losing reputation. It can also reduce unnecessary costs by avoiding penalties. In sum, the litera- ture characterizes objectives of CSR in several dimensions, and the outcome or benets of CSR can be di erent. The lit- erature also suggests that, coupled with an ethical dilemma, these diverse managerial incentives highlight the important role of managerial discretion in CSR. We propose that managers’ particular traits, which have been omitted from consideration in prior studies, are an important factor to determine CSP and a ect the relation between CSP and CFP. Among the many traits of manag- ers (e.g., reputation, leadership, demographic character - istics, style), we focus on a feasible measure of manage- rial eciency that refers to the manager-specic ability to eciently use a rm’s tangible and intangible resources to generate nancial outcomes (Demerjian et al. 2012). As prior studies suggest (Slater and Dixon-Fowler 2009; Attig and Cleary 2015; Osagie et al. 2016), we view managerial eciency as a key characteristic in the process of adopting CSR to organize, implement, execute, and interact and com- municate with the stakeholders of the community outside the rm. Thus, we predict that managerial eciency a ects the choice of CSR and inuences the relation between CSP and CFP via managerial competence in the selected CSR activities. Specically, we investigate the following two research questions: (1) whether managerial eciency plays a role in the execution of CSR and the achievement of high CSP and (2) how managerial eciency inuences the relation- ship between CSP and CFP. To indirectly capture underly - ing managerial incentives, we consider two distinct CSR activities: (1) product-related CSR that is associated with product quality, customer satisfaction, and safety, which will eventually govern the generation of revenue in the future, and (2) environment-related CSR such as installation of antipollution devices to reduce toxic wastes, which is less likely to generate prots but possibly avoids nes or pen- alties. The product-related CSR activities (e.g., long-term quality programs or notable innovation) are an example of relatively more strategic or economic-oriented activities that are expected to provide signal product quality to consum- ers, contrasting with purely prot-oriented business behav - ior producing products of lower quality with less costs. We contend that the product-related CSR may lead more direct nancial benets than environment-related CSR activities that are more likely driven by altruistic and (or) passive ethi- cal or legal motivation. Hence, we predict that managerial eciency di erently inuences these two CSR activities. To test our research questions, we consider CSR activi- ties when they are rated by Kinder, Lydenberg, and Domini (KLD) Research and Analytics, Inc. We use the CSR rat- ing score published by KLD to measure CSP from 2003 to 2011. The KLD CSP score is composed of seven areas (environment, product, labor relations, diversity, commu - nity, corporate governance, and humanities). 3 For manage- rial eciency, we use the measure quantied by Demerjian et al. (2012), who introduced a new measure of managerial ability by quantifying the degree of a rm’s eciency in terms of the relative yield rate of transforming the rm’s resources into output measure (revenue, in this case, control- ling for size and industry). For the empirical proxy of CFP, we use Total Q (Peters and Taylor, 2017) to better incorpo- rate the e ect of nancial and nonnancial aspects of CSR on rm value. 4 Methodologically, we handle the causality issue between CSP and managerial eciency by employing a lead-lag approach (i.e., current CSP explained by prior- year managerial eciency). We also check the robustness with both level and change regression models of CSP. We 3 The score is published by the end of each calendar year by the third party based on the observations and collections of performance data independent of individual rms. Each area of the KLD score is com- posed of strength elements and concerns scores, and the net score of these positive and negative scores is used in similar studies. For example, in the environment area, the aspects measured for the pos- itive aspects are the use of a pollution prevention program, a recy - cling procedure, and deployment of clean energy. If a company uses equipment or has instituted a program to prevent pollution but the company does not have a recycling procedure or use recycled mate- rial or a clean energy source, the score will be 1 for the environment.

For the negative aspects, such as generating hazardous waste; having problems in regulatory compliance; and producing ozone-depleting chemicals, substantial emissions, and agriculturally toxic chemicals, the same company can score 2 by paying nes for the noncomplying emission control problem. Then, an overall environment CSP score for this company will be −1 (1–2). The KLD score stacks all scores from the seven areas. The full description of the seven areas and area- specic measures are provided in “ Appendix”.

4 Peters and Taylor (2017) demonstrate that this measure is better to incorporate intangible investments in rm value than the traditional Tobin’s Q.

1 3 470 S. Y. Cho, C. Lee perform post hoc probing analysis for the moderating e ects of managerial eciency between CSP and CFP.

On average, we nd that managerial eciency does not signicantly a ect the level of aggregated CSP, which is measured by the net total KLD CSR rating score. We inter - pret that this insignicance does not imply the irrelevance of managerial eciency to CSP. Rather, it is plausible that the results reect the outcome of the blending nature of di erent types of CSR. Further analyses document that the moderat- ing role of managerial eciency appears for product-related CSR. On the other hand, we nd that there is no moderating role of managerial eciency on environment-related CSR.

These results suggest that managerial eciency is working in places related to product CSR issues to enhance CSP, not in environmental issues. To control for potential noise in the level of CSR measures, we revisit this issue by using the change of CSP measure. In contrast with the results of level analysis, we nd the positive relation between the change in CSP and the degree of managerial eciency, interpreting as evidence that more-ecient managers make more of an e ort to improve CSP than less ecient managers. With respect to the role of managerial eciency in the link between CSP and CFP, we nd that the link is stronger for rms with high managerial eciency. When we consider the di erential e ect of managerial eciency on two di er - ent types of CSP, we observe a similar moderating role of managerial eciency on product-related CSP to generate high long-term benets. These di erential e ects of mana- gerial eciency to capitalize CSP into CFP suggest that the degree of managerial eciency is one of the important man - agerial characteristics to drive, choose, and execute CSR to generate future benets. In addition, the results of post hoc probing of moderating e ects are generally consistent with those of our main regression analyses. Overall, our results imply that degree of managerial eciency can be an ex ante indicator of whether engaging in CSR is value creating or not. This paper contributes to the ongoing debate regarding the relationship between CSP and CFP in several ways. First, our study introduces and conrms the existence of mana- gerial eciency among several characteristics that govern outcomes of CSR. Our ndings show that managerial e- ciency is an important factor to choose and execute a rm’s CSR. Then, we further trace its inuence on the level of long-term benets. We examine the CSR decision and its consequences at the individual level, as Waddock and Graves ( 1997b) suggest that management quality is a fundamental but unexplored factor to hinge the links between the input and its outcome. They also call for future research with bet- ter measurement of management quality than survey-based data (e.g., a Fortune magazine rating). Our study uncov - ers this unexplored area with managerial eciency that is more comprehensive, estimated from a larger sample, and comparable across industries, which is easily applicable in experimental settings. Our ndings suggest that CSR activi- ties generate social benets when implemented by ecient managers. Second, our analyses provide insights on how to interpret empirical ndings in line with competing CSR theories. Prior mixed empirical implications result from the complicated nature of CSR practices, which are simultane- ously motivated by di erent theories. We identify mana - gerial eciency that moderates this unclear relationship between CSP and CFP. Prior Literature and Hypotheses Development In this section, we rst dene CSR and CSP before discuss- ing the relationship between CSP and CFP as addressed in prior literature. Then, we hypothesize that managerial e- ciency a ects the decision to spend on CSR and plays a moderating role in executing CSR to create rm value.

CSR and CSP Prior literature has dened CSR as a set of voluntary corpo- rate actions beyond legal requirements designed to improve social conditions and social goods (e.g., McWilliams and Siegel 2001; Mackey et al. 2007). This broadly dened CSR can be further outlined as several di erent underlying mana- gerial objectives. Carroll (1991) identies CSR as a spec- trum of various types between two distinct and contrasting styles: economic CSR and altruistic (or philanthropic) CSR.

For example, the objective of economic CSR is in line with stakeholder theory that views CSR as managers’ response to the demand of stakeholders such as employees, suppliers, community, directors, creditors, and investors, who are ulti- mately interested in the use of a rm’s resource to enhance rm value. Hence, the underlying objective of a so-called economic CSR is to create value for the rm, particularly long-term monetary benets. In contrast to more business-oriented economic CSR, altruistic CSR has little connection to such prot-oriented motivation. Instead, altruistic CSR refers to voluntary behavior (e.g., charitable contribution or donation) beyond what business is required to accomplish business goals or to prevent possible harm to public. Accordingly, prior stud- ies characterize altruistic CSR as managerial investment with value-irrelevant or value-decreasing activities. Carroll ( 1991) names another dimension between economic and altruistic CSR: ethical or legal CSR (e.g., community dona - tion and extra spending on environmental protection), which typically responds to legal or ethical fulllment. This passive CSR spending may result in a high level of sustainability, but the e ect of such CSR spending on rm value is not clear.

1 3 471 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance In this study, we view that potential objectives related to CSR vary between the two extreme types. That is, managers decide on CSR investment on the basis of di erent motiva- tions or by considering multiple objectives simultaneously.

While there are various underlying motivations and types of CSR, the outcomes of CSR are usually evaluated by inde- pendent rating agencies or organizations. In the literature, CSR rating scores by such independent entities refer to CSP.

Even though these scores cannot fully capture the actual consequences, researchers and practitioners generally use these scores as proxies for CSP. The reason is that the rating agencies, including KLD, use numerous sources within the full range of a rm’s CSR activities, to which other stake- holders have very limited access. The sources that rating agencies utilize include not only a rm’s nancial reports but also the agencies’ own analyses of media, surveys, and government reports as well. As in the prior studies, we use the KLD rating score as a proxy for a rm’s CSP.

Managerial E ciency and CSP (H1) We consider managerial discretion as a key moderator in CSR investment based on the upper echelons theory (Ham- brick and Mason 1984; Hambrick 2007 ) depending on the context of CSR. The theory suggests that qualities and characteristics of top managers (i.e., the upper echelon of an organization) determine a rm’s strategic decisions. 5 According to Hambrick and Mason (1984), the character - istics of the CEO or senior management teams are associ- ated with past individual experience, value, and educational background. This tangible and intangible expert knowledge and these characteristics enable top-level managers to make ecient and valuable decisions. For CSR investment, the top manager formulates a rm’s strategic choice and her beliefs of social responsi- bility. McGuire et al. (2003) conclude that “strong social performance may be primarily driven by managerial belief.” Moreover, the unique aspect of CSR investment can high- light the manager’s role because the outcome of CSR is not always immediately measurable and has an intangible nature. Instead, potential benets of CSR are not instantly visible or realized, but are often accompanied by a costly nancial burden. Thus, CSR spending calls for managerial discretion to make decisions on whether to engage in CSR and which CSR activities to choose.

In this study, we use managerial eciency dened as the degree of management’s specic ability to generate revenue using a rm’s tangible and intangible resources consistent with Demerjian et al. ( 2012). Managerial eciency can inu - ence CSR in two ways. First, given di erent objectives of CSR, managerial eciency becomes a factor in determining CSR investment and selecting certain areas of CSR. For exam- ple, because a rm usually has fewer resources than needed to satisfy needs of all stakeholders, an ecient manager may seek CSR as a corporate strategy (i.e., economic CSR) or may forgo CSR investment. 6 Next, if managers determine CSR investment, then they decide on which area to focus, such as economic CSR, altruistic CSR, or a combination of these two with ethical consideration. Second, managerial e- ciency a ects the quality of the execution of CSR. Prior stud- ies (e.g., Chakravarthy 1986; Koch and Cebula 1994) suggest that e ective managers should be able to meet or adapt to the expectations of a new environment, such as the stakehold- ers or the society in which the corporation operates. Thus, the degree of managerial eciency can ultimately a ect rm value via the manager’s competence with the underlying moti - vation of CSP. Figure 1 summarizes the conceptual frame - work of our study regarding the role of managerial eciency. One potential inuence of managerial eciency is a posi- tive e ect on achieving high CSP. Several studies (Slater and Dixon-Fowler 2009; Attig and Cleary 2015; Osagie et al.

2016) suggest that characteristics of management determine the outcome of CSR. Slater and Dixon-Fowler (2009) show that the past experience of the CEO increases CSP with the allocation of limited resources. They demonstrate the need of human capital to command the responsible, innovative execution of complex tasks. Attig and Cleary (2015) also show that high-quality management practice leads to high CSP. Their study views management quality as organiza- tional capital that can eciently derive outcomes. Osagie et al. (2016) summarized managerial competence as a key moderator for CSR performance in their meta-analysis. From these ndings, we conrm that managerial eciency can be a moderator to determine CSR projects to achieve high CSP. 5 Corporate governance is another important intermediary that a ects the relationship between CSP and corporate nancial per - formance. As stated in Hong et al. (2016), if managers use CSR to maximize all stakeholders’ value, corporate governance is likely to promote CSR and thus improve corporate nancial performance. On the other hand, if managers’ engagement in CSR is an agency cost, corporate governance is likely to reduce CSR activities. In this study, we do not directly test the role of corporate governance because of our interest in managerial eciency. However, we conduct a sensitiv - ity test by using a separate independent variable of the governance KLD score from our main variable of CSP. Untabulated results indi- cate that results based on the new measure of CSP (excluding govern- ance score) are consistent with our main ndings. 6 It is worth noting that we use the term “limited resource” to refer the rm’s general and fundamental economic situation in which a rm has fewer resources to ll all (unlimited) needs from the rm’s stakeholders (i.e., economic scarcity). Therefore, the term limited resource does not necessarily mean a rm’s nancial constraint, which indicates a rm’s situation facing high costs of external nanc- ing (or a shortage of internal funding), or a rm’s nancial distress indicating a diculty to pay o  a rm’s nancial obligation.

1 3 472 S. Y. Cho, C. Lee On the other hand, some CSR activities are ruled out by the strategic choices that pursue economic eciency. As noted in the previous study by Carroll (1991), CSR choices have a wide spectrum, and at one extreme end is the altru- istic choice of CSR. The altruistic CSR includes a donation to education or social programs or participation in envi- ronmental protection programs that are rarely picked up by strategy-oriented managers who act based on the principle of eciency. As a result, an ecient manager is less likely to achieve the goals of certain CSR such as the community contribution or environmental protection aspect of CSP. In this case, the managerial eciency will negatively inu- ence CSP. Carroll (1991) suggests a mixed incentive that comes from the middle ground of strategic and altruistic CSR choices that does not necessarily reect managerial eciency. Using scarce resources to just comply with gov - ernment regulations would not be connected with eciency.

So, either from altruistic purposes or to meet the ethical requirements, non-economic CSP will not generate any ben- ets, and it will be considered inecient. In sum, managerial eciency is viewed as one of the main characteristics that moderates the process of adopting CSR and achieving high CSP, because underlying manage- rial incentives (strategic, altruistic, or ethical) govern the choices of CSR that result in a higher outcome of CSP. So, assuming that a rm usually deals with limited resources, the strategic choice of CSR predicts a positive inuence of managerial eciency on the CSR-CSP relation, while the altruistic or ethical choice of CSR predicts a negative inu- ence of managerial eciency on CSR-CSP. Therefore, we formulate our rst hypothesis on the relationship between managerial eciency and CSP as follows with no prediction on the direction:

H1a The degree of managerial eciency a ects the level of CSP. To provide additional insight into the rst hypothesis (H1a), we examine two specic dimensions of CSP that reect the distinct underlying incentives for CSR: product- related CSP and environment-related CSP. We view the former as the choice of strategic or economic CSR and the latter as the altruistic or ethical CSR. Product-related CSP captures CSR activities related to new product (service) development, logistical supply chain enhancement, and risk- reducing product line initiatives that directly increase e- ciency and save cost. This type of CSR investment is distinct from other product-related activities lacking CSR, such as production of low-quality products with minimal costs and without innovative R&D. In the short run, the net nancial benets from product-related CSP may be lower than those from product-related activities without CSR engagement, but product-related CSP may lead to an increase in product loyalty and build up positive image of the product, resulting in valuable signals to customers. So, product-related CSP is likely to directly connect to a rm’s value through poten- tial nancial benets that require ecient use of resources.

On the other hand, we view environment-related CSP as relatively ethical or altruistic CSR (non-economic CSR) that requires a relatively passive commitment but hurts the overall eciency of resource allocation. Many rms engage in environmental CSR activities to comply with governmen - tal regulations or social beliefs, but they do not require any sophisticated operation to achieve maximum benet. 7 In that Corporate Social Responsibility (CSR ) Strategic CSR Altruistic CS R Corporate Social Performance (CSP ) (Product-related CSP & Environment-related CSP) KLD Scor e Corporate Financial Performance (CFP) Total Q Managerial Efficiency Fig. 1 Conceptual Model 7 The response from managers to the regulation can vary according to how managerial discretion is exercised. If a rm chooses a mini- mum level of environmental CSR investments up to the expected penalties or nes on the non-compliance, the benets (i.e., saving penalties or nes) from this CSR can be o set with the cost of CSR (i.e., investment on CSR), resulting in zero prot. Likewise, if a man- ager voluntarily chooses higher environmental standards than the regulated ones, the costs of the environment-related CSR exceed the potential direct benets.

1 3 473 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance case, the ultimate benet of environmental protection will be shared by the community, but it is not a direct nancial benet to the rm.

We view CSR activities as a combination of relatively economic CSR and non-economic CSR and then predict that more-ecient managers are likely to use their resources for relatively economic-oriented CSR rather than relatively non- economic CSR. We argue that CSR spending, for managers pursuing high eciency, is less likely to be a top priority of a rm’s investment opportunity set because a rm’s resources are limited and the expected benets from CSR investment are uncertain. Indeed, ceteris paribus, ecient managers will use limited resources to invest xed assets or research and development prior to spending in CSR. Moreover, if managers have a strong ability to eciently use their assets for busi- ness operations, they will choose more strategic CSR such as product-related CSR rather than ethical or altruistic environ- ment-related CSR. A line of studies (e.g., Hillman and Keim 2001; Jayachandran et al. 2013) support our prediction by suggesting that the associations between managerial eciency and CSP depend on the specic facets of CSR activities.

However, our argument does not mean that the main objec- tive of CSR is only monetary-driven action and that ecient managers ignore non-economic CSR. Instead, we conjecture that managerial eciency may drive managers to subordinate non-economic CSR in a rm’s ecient frontier. Based on the above argument, we propose the following hypothesis:

H1b Managerial efficiency differently affects product- related CSP and environment-related CSP.

Managerial E ciency and the CSP–CFP Relation (H2) Margolis et al. (2009) indicate that the common average positive relationship between CSP and CFP is surprisingly small (r = 0.13), or only 28% of the studies show a positive relation, whereas 59% of studies show insignicant results.

Even with various moderators and conditions to study the CSP–CFP relation, the prior studies overlooked the role of managerial eciency as an important moderator in the CSP–CFP relation. As discussed in the previous section, we argue that managerial eciency can moderate this rela- tion as two channels of a determinant of the heterogeneous aspects of CSR activities rooted in various nancial and non- nancial benets and the high degree of competence in the selected CSR aligning with underlying objectives. First, in connecting to H1, the degree of managerial e- ciency a ects CFP via a particular type of CSP motivated by managerial incentives. Carroll and Shabana (2010) indicate that certain CSR activities reduce business risks related to a rm’s technical, market, and political aspects in running a business. This immediate monetary benet from CSR gives managers incentives to engage in CSR for their compensa- tion tied to the rm’s nancial performance. In this case, more-ecient managers can select strategic CSR as a sus- tainable business strategy not only because of the demand of stakeholders but also for the managers’ own interests, suggesting the potential positive inuence of managerial eciency on strategic CSR–CFP relation.

Managerial eciency also can garner nonnancial ben- ets from strategic CSR. It is well known that another con- sequence of CSR is future long-term economic benets that are not captured as immediate monetary rewards (e.g., Lubin and Esty 2010; Kiron et al. 2012). These nonmonetary ben- ets of CSR incur temporary costs to rms, but they are expected to bring future benets that exceed current costs in the long run. Intangible and nonmonetary benets include an inuence on attracting customers that results in cus- tomer loyalty (Bhattacharya and Sen 2003, Barnett 2007 ), employee morale and retention that will eventually increase productivity, a rm reputation among the public that will decrease the risk of litigation (Godfrey 2005, Williams and Barrett 2000), or product di erentiation and the intangible benets of improving customer loyalty (McWilliams and Siegel 2001). Because ecient managers will care for the use of resources and its benets, these potential nonmon- etary benets of CSR will motivate such managers to allo- cate resources in CSR. Thus, we predict that more-ecient managers pursue CSR as long as the benets of CSR can improve CFP. This leads us to predict the positive inuence of managerial eciency on the CSP–CFP relation. On the contrary, the role of managerial eciency can di er when managers have altruistic motivations. Especially when a rm has abundant resources, the altruistic manager of the rm is more likely to use the extra cash to create CSR programs not to generate any benet but to act as a good citizen (Kang et al. 2016). Thus, the CSR has priority in the altruistic decision ahead of benet-generating alterna- tives, which may result in the inecient use of resources. In contrast with ecient strategic CSR activities, philanthropic incentives of CSR can result in bad nancial performance without any other benets of securing sustainable busi- ness. Indeed, pure altruism leads managers to use too many resources on CSR activities beyond the optimal level; in other words, the marginal costs of altruistic CSR are greater than marginal benets of such CSR. 8 Thus, this altruistic CSR spending can have a negative impact on the wealth of stakeholders (Baron 2001). So, in this view, CSR activities 8 We refer to the optimal level in the context of economic optimality that marginal benets are equal to marginal costs. Since the economic marginal benet from pure altruism is likely to lower than economic marginal costs of such CSR, the CSR investment rooted in pure altru- ism leads to sub-optimal investment.

1 3 474 S. Y. Cho, C. Lee result in nonproductive investments and increase proprietary costs, leading to the loss of a competitive edge in the market, ultimately decreasing the rm’s value and imposing addi- tional stakeholder costs. Under the realistic business envi- ronment with limited resources, this altruistic management will not be generally viewed as an ecient manager in terms of resource allocation. 9 Importantly, another channel of the moderating role of manager is managerial competency in the selected CSR activities, because more-ecient managers are likely to be competent, and they are therefore expected to implement CSR investment more e ectively than inecient managers.

Osagie et al. (2016) suggest a manager’s individual compe- tency can inuence the e ectiveness of CSP. In line with Osagie et al. (2016), we view managerial eciency as the key attribute of management competency. Our measure of managerial eciency is supposed to capture how well a manager executes business operations with all of the rm’s available resources to generate ultimate financial gain.

Accordingly, if a manager chooses economically oriented CSRs, we predict that the CSP–CFP relation would be much stronger when ecient managers implement CSR. This pre- diction can also work for altruistic CSP. For example, when managers can primarily pursue altruistic CSR with resource slack, the contribution to CFP is not clear in conjunction with the degree of managerial eciency. However, even if managers spend on altruistic CSR, more-ecient managers may implement altruistic CSR for “insurance-like” e ects, 10 which makes managers immune to future loss because of the risk to their reputation (e.g., Godfrey 2005; Shiu and Yang 2017), rather than for managerial self-interest with value- irrelevant or value-decreasing spending. We predict that the relation between CSP and CFP can be stronger when more- ecient managers invest in CSR because of more strategic choices and high managerial competency While our main interest is the role of managerial e- ciency in the relation between CSP and CFP, we rst state a general hypothesis regarding the relation between CSP and CFP (H2). As discussed earlier, because the relation between CSP and CFP is inconclusive, our prediction on this relation is also non-directional. We then extend our general hypoth- esis, specically aiming at the role of managerial eciency as H2a and H2b. These discussions lead to the following hypotheses:

H2 A rm’s CSP is associated with its CFP.

H2a The association between CSP and CFP is stronger for rms with ecient managers than for rms with inecient managers.

For the two subdimensions of CSP, product-related CSP and environment-related CSP, we can use a similar logic employed to derive the prior hypotheses. Given di erent objectives of the two CSPs, we expect that the relation of product-related CSP and CFP is stronger than that of envi- ronment-related CSP and CFP. Moreover, when we consider the moderating role of managerial eciency, we predict that the e ects of product-related CSP with ecient managers on CFP may di er from those with inecient managers on CFP because of the managerial e ectiveness of implementing CSR. However, we do not predict a di erence in eciency e ects on the relation of environment-related CSP and CFP because the relation is inuenced by other factors, such as altruism, rather than eciency. Accordingly, we formulate another hypothesis as follows:

H2b The positive association between product-related CSP and CFP is stronger for rms with ecient mangers than for rms with inecient managers. Sample and Research Design Sample Our initial sample includes rms common to the Compustat annual le, Center for Research in Securities Prices (CRSP) monthly returns le, and KLD database for the years 2003 through 2011. We use Compustat data to estimate the vari- able of managerial eciency and necessary control vari- ables. To measure CSP, we use the KLD CSR performance scores (strengths and concerns) from seven categories. We also use the CRSP return le to calculate standard deviation and mean value of stock returns for a proxy of the rm’s information environment. From this initial sample, we delete rms in nancial institutions (SIC codes between 6000 and 6999) because of di erent operations of rms in the nan- cial sector. We also exclude rms with negative book value of equity. Finally, we require the lagged variable of KLD 10 The insurance e ects of CSR refer to a case in which CSR activi- ties can mitigate or reduce the impact of the occurrence of a nega- tive event. For example, building a good reputation via long-term CSR engagement (i.e., insurance premium) can mitigate the decrease in stock price (i.e., insurance coverage) when the rm announces bad news (e.g., the SEC’s investigation of securities fraud).

9 Outcomes that are similar to those from altruistic incentives can be caused by management entrenchment, too. For example, Johnston (2005) argues that managers use CSR to greenwash unethical choices that rms make or to hide poor nancial performance. Furthermore, in line with agency theory, top managers’ inated self-condence due to hubris may drive them to underestimate strategic need but to overestimate the resources in their hands to promote their own empire building (Malmendier and Tate 2005). Similarly, CSR motivated by management entrenchment will result in an inecient management of resources such that even with high CSP, there will be a negative inu- ence on the CSP–CFP relation.

1 3 475 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance scores to assess the role of managerial eciency in change of CSP. This sample selection procedure yields 11,037 rm- year observations.

Corporate Social Performance (CSP) To test our research questions, we use the KLD STATS score to measure CSP from 2003 to 2011 as commonly used in prior studies (e.g., Waddock and Graves 1997a; Hillman and Keim 2001; Deng et al. 2013). KLD is one of the oldest and most inuential social responsibility rating agencies. While stakeholders cannot catch the full range of numerous CSR activities and they have limited access to the actual outcome of such CSR investments, the rating agency intermediates transparent and comprehensive information about CSP measure. Using various nancial and nonnancial report- ing sources such as surveys, media, articles in academic test for the di erential e ects in specic areas of interest among seven KLD summary categories (H1b and H2b), we use a single category score of product-related and environ- mental issues, which ranges from − 1 to + 1. Managerial E ciency We use the managerial eciency (ME) measure developed by Demerjian et al. (2012) to assess the important trait of manager. Demerjian et al. modeled this measure by employ - ing data envelopment analysis (DEA), which is used for assessing the relative eciency of business entities, to esti- mate rm eciency in each industry. They dene the DEA objective function with input factors of rm eciency (i.e., cost of goods sold; selling and administrative expenses; net property, plant, and equipment [PP&E]; net operating leases; net R&D; purchased goodwill; and other intangible assets) to generate the output factor of sales:

This optimization process generates the total relative e- ciency measure at rm level in its estimation group using the optimal combination of input factors. Demerjian et al.

( 2012) assign one for the most ecient rm and zero for the least ecient one. Next, because the total rm eciency is driven by rm- specic factors (e.g., rm size, market share, positive free cash ows, life cycle of rm, and diversication or com - plexity of rm operations) and management-specic factors, Demerjian et al. (2012) attempt to separate management- specic factors from total rm eciency. They capture man- agement-specic rm eciency from the following Tobit regression as regression residual value:

Consistent with Demerjian et al. (2012), we use the raw value of the residual and decile ranked regression residual value across the rm-years and industry in our main analy - ses. Recent studies conrm the validity of this measure in terms of credible management disclosure (Baik et al. 2011) and high earnings quality (Demerjian et al. 2013).

Corporate Financial Performance (CFP) When we operationalize the variable of CFP, the main con- cern is how the empirical proxy of CFP can better capture (1) max . = Sales 1COGS + 2SG&A + 3PPE + 4OP\fEASE + 5R&D + \bGoodwill + 7In tan. (2) Firm e ciency it= 1+ 2Size it+ 3Market s\fare it + 4Positi\be Free cas\f ow it + 5Ln(Age ) it+ 6Business Concentration + 7Foreign Currency indicator it +Year indicator + it. journals, and government reports, KLD evaluates CSP in seven categories of CSR: corporate governance, commu- nity, diversity, employee relations, environment, products, and human rights. For each category, KLD collects scores for positive (strengths) and negative (concerns) aspects of CSR (high strengths are high CSP and high concerns are low CSP). The scores are the sum of zero or one, with numbers given for several elements in each category (see "Appen- dix" for details). For example, the total number of positive elements in the product category is four for both strength and concerns elements. Thus, the range of category score for strength or concerns score is from zero to four and the category score can range from − 4 to + 4 when we calculate the net score (strength score minus concerns score) for the product category as other studies use. Likewise, the KLD score provides both categorical aggregation of elements binary score (zero or one for each element) in strengths and concerns for seven categories.

One diculty of using these raw scores of KLD per - formance indicators is its comove in each year (Manescu 2009). Thus, it is dicult to compare CSP across years and dimensions. To address this concern, we calculate the net weighted adjusted CSP score, converting each categorical strength and concern as the proportion to the maximum score for the category. Then, we calculate the net weight as (weighted adjusted CSR strengths score—weighted adjusted CSR concerns score), following Deng et al. (2013). So, this net weight treats each category equally and each category score will be standardized to range from − 1 to 1. When aggregated as total net CSP, it ranges from − 7 to + 7. To 1 3 476 S. Y. Cho, C. Lee the e ect of dynamic CSR with nancial and nonnancial outcomes on rm values. In this study, we use Total Q (Peters and Taylor 2017) as a proxy of CFP associated with CSP. Peters and Taylor (2017) introduce this new metric as rm value deated by the sum of physical and intangible capital as follows:

where Kphy it is the book value of property, plant, and equip- ment (Compustat item PPEGT) and Kint it is the book value of intangible assets (Compustat item ITAN), and Vit is the mar - ket value of equity (Compustat items PRCC_F times CSHO) plus book value of debt (Compustat Items DLTT+DLC) minus current assets (Compustat item ACT). Total Q is intended to overcome the limitation of Tobin’s Q , a popular measure of capturing rm value assessing mainly by the physical investment, because Total Q incorporates equally the investment of physical assets and intangible assets. We believe that Total Q is a better measure than Tobin’s Q in our test setting because: (1) one of the main elements of Total Q is market value of equity in the stock market, and (2) Total Q considers intangible assets as a key driver of rm value. It is well known that the investor reacts to not only current nancial indicators but also nonnancial and long-term, forward-looking qualitative value indicators. For example, CSR activity aiming at product innovation in a socially responsible manner or altruistic funding for the local community is less likely to contribute to the current nan- cial outcome (e.g., earnings), but is likely to improve future economic performance because successful CSR activity can increase customer satisfaction or enhance brand image.

Thus, Total Q is expected to capture this kind of future value through stock price. In addition, like most intangible assets, which are signicant and growing value drivers but rarely quantiable resources, the nature and potential benets of CSR are similar to intangibles. Taken together, we expect that Total Q can be a better measure to reect the intangible nature of CSR. 11 Total Q = Vit Kphyit +K intit , Regression Models To test the rst hypothesis, we estimate the following regres- sion model:

where CSPmeasure is the total aggregated net score of weighted KLD performance scores and ME is the measure of managerial eciency as dened before. As we elaborated in our hypothesis development, we use the managerial e- ciency of the prior year. 12 Next, we control for rm perfor - mance in accounting prot (ROA) and mean value of annual buy-and-hold stock returns (M_ RET) for the scal period.

Also, as suggested in the information asymmetry and CSP literature (Cho et al. 2013), we control for the percentage of institutional ownership (INST), risk measures of leverage (LEV), and annual standard deviation of daily stock returns (STD_ RET). Finally, we include the size variable measured by the log of lagged total assets (SIZE) to control for poten- tial endogeneity of CSPmeasure determined by a rm’s size.

In Eq. (3 ), if rms with more-ecient managers are more (or less) likely to engage in CSR, we predict that the coef- cient of our measure of managerial eciency, (ME), 2 , will be positive (negative). We run the same regression for the single-category-based CSP measures: product-related CSP - measure and environment-related CSPmeasure. With these variables, we can test whether there is any dierence in the inuence of managerial eciency on CSP. To test our second research question about the moderating role of managerial eciency on the relationship between CSP and CFP, we run a regression for two subgroups of ecient and inecient managers partitioned based on the median value of managerial eciency (ME). Then, we can compare the CSP–CFP relations from each regression to test for the dierential eect of managerial eciency on achiev - ing high CFP with high CSP. For CFP, we use the long-term perspective measure of Total Q (TOTALQ) that includes the value of tangible and intangible future growth potential.

Thus, we form the following model using Total Q as our measure for rm performance: (3) CSPmeasure it= 1+ 2ME it 1 + 3ROA it+ \fSIZE it 1 S 5\bEV it+ 6INST it+ 7M_RET it + 8STD_RET it+ it, 11 An alternative proxy of CFP is stock return (or cumulated abnor - mal return) regarding the report of the KLD CSR score. While change in stock price can be a good indicator for both current nan- cial performance and future nonnancial performance, we use Total Q rather than stock returns or any short-term window abnormal returns because stock returns without a clear event window provide a “noise” indicator. Unlike an accounting earnings announcement, KLD does not provide any specic announcement date, so it is dif- cult to match a relevant testing returns window. 12 While we adopt the lag-lead testing approach to clarify the causal relationship between managerial eciency and CSP (and corporate nancial performance), there is still a potential endogeneity issue when able managers engage in CSR activities. We conduct additional tests using a two-stage regression model to control for potential endo- geneity. In the rst-stage regression, we obtain a predictive value of CSP and use it in the second-stage regression on corporate nancial performance. The untabulated results are qualitatively consistent with our main ndings.

1 3 477 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance where GROWTH is sales growth.

From each subgroup, we obtain coecients on the CSP term, 2 , which captures the relation of CSP and CFP. Then, we test for the moderating role of managerial eciency on CSR by comparing those coecients across two groups (ecient and inecient managers). The prior literature sug- gests that CSP is a forward-looking measure that indicates future value creation. Thus, we expect a positive association between CSP and TOTALQ. Furthermore, as we argued in the previous section, managerial eciency will play a key role in capitalizing on such CSR to achieve high CFP. As we built our argument in the hypotheses, high managerial eciency can transform the high CSP as value-increasing activities. Thus, we expect 2 to be positive and larger for 2rms with ecient managers than for those with inecient managers. 13 Empirical Results Descriptive Statistics Table 1 shows the summary statistics of the variables. The mean value of the ranked variable of managerial eciency (4) TOTALQ it= 1+ 2CSPmeasure it+ 3\fOA it + 4SIZ\b it 1 + 5L\bV it+ 6INST it + 7G\fOWTH it+ 8STD_\f\bT it+ 2it, (ME) is 5.5. 14 Also, the table shows that our test variable corporate social responsibility performance has a mean value close to zero (− 0.206) and ranges from − 2.238 to 2.202. Similarly, our other single-category-based CSP meas - ures show similar characteristics (mean values − 0.036 for product-related CSP and 0.021 for environment-related CSP) with narrower ranges (from − 1 to 0.4 for product-related CSP and from − 0.454 to 0.714 for environment-related CSP). Table 2 shows the correlation coecients among major variables, and it shows that managerial eciency is posi- tively related to CSP, the overall performance measure, but it is insignicant. Meanwhile, managerial eciency is positively associated with the product category performance measure and is statistically signicant (at the 5% level).

However, environment-related CSP is negatively associated (at the 1% level). From these correlation coecients, we can infer a di erent e ect of managerial eciency by CSP type.

Most notably, the product-related performance measure shows a positive association, while the environment-related measure shows a negative association. Regression Results Table 3 shows the test results of running the ordinary least square (OLS) Eq. (3 ) on three di erent types of dependent variables: CSP for overall score, PCSP for product-related CSP score, and ECSP for environment-related CSP score. 15 The table shows that when the dependent variable is the overall level of performance measure (CSP), the coecient on managerial eciency (ME) is negative and marginally signicant (− 0.049 and signicant at the 10% level). How - ever, for other specic CSP variables (PCSP and ECSP), as suggested in other studies (e.g., Jayachandran et al. 2013), we nd a positive inuence of managerial eciency on the product-related performance measure (PCSP) (0.032 and signicant at the 1% level). On the other hand, for the environment-related performance measure (ECSP) as the dependent variable, the coecient on managerial eciency ( ME) becomes negative (− 0.049) and statistically signi- cant at the 1% level, suggesting that ecient managers are less likely to choose environment-related CSR. Also, overall results suggest that CSP is driven largely by the environmen- tal component. These ndings support the second part of our rst hypothesis (H1b). In particular, these results show that 13 We also add the KZ index (Kaplan and Zingales 1997) as a proxy of nancial constraint to a set of control variables. The KZ index is based on the following ve factor model:

where Cash ow is the sum of operating income (data item IB) and depreciation (data item DP) scaled by beginning balance of property, plant, and equipment (data item PPENT); Q is Tobin’s Q measured as the sum of [assets (data item AT)—book value of common equity (data item CEQ)—deferred tax (data item TXDB) + market value of equity (data item CSHO * data item PRCC_F)] scaled by book value of common equity (data item CEQ); LEV is long-term debt ratio measured as the sum of [short-term debt (data item DLTT) + long- term debt (data item DLC)] scaled by the sum of [short-term debt (data item DLTT) + long-term debt (data item DLC) + stockhold- ers’ equity (data item SEQ)] (if stockholders’ equity is negative, we set LEV equal to 1); DIV is dividend payment for common stock and preferred stock shareholders (data item DVP + data item DVC) scaled by beginning balance of property, plant, and equipment (data item PPENT); Cash holding is measured as cash (data item CH) scaled by beginning balance of property, plant, and equipment (data item PPENT). Larger (smaller) values of KZ Index correspond to higher (lower) levels of nancial constraints. Untabulated results of this robust test show that our main ndings are not sensitive to the additional variable of nancial constraint.

KZ Index =− 1.002 ICash ow n0.28\f IQ\f.1\f9 ILE\b Z\f9.\f68 IDI\b Z1.\f15 ICash holding, 14 The untabulated mean (median) value of a rm’s eciency from the whole sample from 2003 to 2011 is 0.587 (0.608). The manage- rial eciency data are available at Professor Sarah McVay’s website (http://faculty.washington.edu/smcvay/research.html).

15 For the robustness of our tests, we used a ranked score of CSP.

Untabulated results show that the results based on the ranked variable of CSP are consistent with those based on the raw value of CSP.

1 3 478 S. Y. Cho, C. Lee managerial eciency boosts product-related CSP (PCSP), but managerial eciency harms environment-related CSP (ECSP). We interpret that due to these di erential inu- ences of managerial eciency on various aspects of the CSP measure, the combined e ect of managerial eciency on the overall CSP becomes ambiguous. One possible rea- son for these inconclusive results is the regression model specication that uses variables measured at the level CSP model because the variable rarely changes over the years.

We revisit this issue in a later section by adopting a CSP change model instead of level.

Next, to nd the evidence on our hypothesized relation- ship between rm value and inuence of managerial e- ciency on CSR activities to enhance the rm’s value, we use CSP as the proxy for such activities and tests to compare the cross-sectional di erence for di erent managerial eciency levels. Table 4 shows our tests for the three di erent sets of our proxy for CSP (CSP, PCSP, and ECSP). We divide the sample into two subgroups (ecient and inecient manager groups) using the median value of managerial eciency. For each group, we run the regression of rm value (TOTAL_Q) on the CSP measures (CSP, PCSP, and ECSP) for each group by managerial eciency.

Column (1) of Table 4 shows the results for the ecient manager group, and the coecients on CSP, PCSP, and ECSP are all positive (1.106, 1.648, and 2.792, respectively) and statistically signicant (at the 1% level). The results show that ecient managers can capitalize the CSP into the rm value captured by Total Q (TOTAL_Q). However, our interest is to examine di erential e ects of the degree of managerial eciency. For the comparison, we run the same regression on the inecient manager group. Column (2) of Table 4 shows that the coecients on CSP and ECSP are positive (0.670, and 2.911, respectively) and statistically signicant (at the 1% level), but the coecient on PCSP is negative (− 0.127) and statistically insignicant. These results suggest that overall, even inecient managers can make CSP contribute to the rm value but the intermediary role in PCSP is not critical for rm value. To examine how the performance of ecient managers is di erent from that of inecient managers, we compare the coecients on the three variables. The statistical comparison shows that the coecient on CSP of the ecient manager group (1.106) is larger than that of the inecient manager (0.670), and the di erence (0.436) is statistically di erent at the 1% level (Z -statistic 2.17), which suggests that ecient managers can capitalize overall CSP into rm value better than inecient managers. 16 Also, we nd that the e ect of managerial eciency is not statistically di erent for ECSP, whereas the di erence in the coecient of PCSP is signi- cant at the 1% level (Z -statistic 3.13). These results indicate that managerial eciency does not have a moderating role in the environment-related ECSP–CFP relation. Overall, the results suggest that CSP can better contribute to CFP for rms with ecient managers but the moderating role of managerial eciency does not equally apply to all types of CSP, but is driven by more economic-oriented CSP. 17 Table 1 Descriptive statistics Variable denitions:

ME The decile rank of managerial eciency from Demerjian et al.

(2012) by year and industry (two-digit SIC code), CSP corporate social responsibility performance measured by adjusted total strength score minus adjusted total concern score from seven categories of KLD; PCSP product-related social responsibility–adjusted net score from KLD, ECSP environment-related social responsibility–adjusted net score from KLD, TOTAL_Q the market value of equity (Com- pustat items PRCC_F times CSHO) plus book value of debt (Com- pustat items DLTT + DLC) minus current assets (Compustat item ACT) divided by the sum of the book value of property, plant, and equipment (Compustat item PPEGT) and the book value of intangi- ble assets (Compustat item ITAN), SIZE natural logarithm of rm’s total assets of the previous year, ROA net income before extraordi- nary items deated by total assets of the beginning of the year, LEV rm’s leverage measured by long-term debt deated by total assets of the beginning of the year, INST percentage of sum of insti- tutional ownership, M_RET annual buy-and-hold return for the scal year, STD_RET standard deviation of daily stock returns for the scal year, GROWTH sales growth measured by change in sales deated by lagged sales Variable # of Obs.MeanSDMin. Max.

ME 11,037 5.5002.872 1.00010.000 CSP 11,037− 0.2060.399− 2.238 2.202 PCSP 11,037− 0.0360.136− 1.000 0.400 ECSP 11,037 0.0210.106 − 0.454 0.714 TOTAL_Q 11,037 1.1010.918 0.00133.720 SIZE 11,037 6.9981.561 2.05213.589 ROA 11,037 0.0290.128 − 0.648 0.271 LEV 11,037 0.4720.207 0.0720.950 M_RET 11,037 0.0010.002 − 0.011 0.020 STD_RET 11,037 0.2580.411 0.0061.000 INST 11,037 0.5780.379 0.0001.000 GROWTH 11,037 0.1210.294 − 0.523 2.179 17 We conduct additional analysis by using modied Total Q, where the value of R&D expense is added to the numerator of formula of Total Q because R&D expenditure is not capitalized in accounting.

While we lose about 43 percent of the total sample because of blank value or missing value of R&D in Compustat, untabulated results using the modied Total Q show much stronger results compared to results in Table 4. Considering an external validity of our ndings, however, our main results are based on the original measure of Total Q.

16 Following Clogg et al. (1995), we use the following Z-statistics to test whether the coecients are the same between the two partitioned groups:

Z= ( ̂HighME − ̂LowME ) √s2( ̂HighME )+s2( ̂LowhE ), where ̂HighME and ̂LowME are coeocient estimates from the two subsamples and s2() are the squared standard errors of the coe)- cients.

1 3 479 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance Alternative Model Testing: Change in CSP Because rms’ policies on CSR may be sticky over the years, our main analyses based on level of CSP specication may be inuenced by rm-specic unobservable heterogeneity.

As a result, it can be dicult to interpret the presumable cau- sality of managerial eciency on the relationship between CSP and corporate nancial performance. To assess this possibility, we adopt the ordered logistic regression model reecting change in CSP and test the relationship between change in CSP and managerial eciency. This alternative model enables us to better test how the degree of manage- rial eciency at the beginning of the year a ects CSP and whether managerial eciency di erentially inuences cor - porate nancial performance through CSP. Therefore, we modify Eqs. (3 ) and (4) to estimate the changes in the CSP model as follows:

where ΔCSPmeasure in the ordered logistic regression model (3-1) is defined in three categories: one, if CSP decreases; two, if CSP does not change; and three, if CSP increases. In Eq.m(3-1), 1 captures the multiple intercept terms. Table 5 presents the results of regressions of Eqs.(3-1) and (4-1). We do not report the intercepts for dierent out- comes in the ordered logistic models and OLS regression (3-1) ΔCSPmeasure it= 1+ 2ME it−1 + 3ROA it+ \fSIZE it−1 P 5\bEV it+ 6INST it+ 7M_RET it + 8STD_RET it+ it, (4-1) TOTALQ it= + ΔCSP it+ + −1 O 5LE\f it+ 6INST it+ 7\bROWTH it + 8STD_RET it+ 2it, Table 2 Correlation matrix This table shows the Pearson correlation of key variables used in the analyses. Values in bold indicate correlations that are signicant at the 10% level or better. All variables are dened in Table 1 [1] [2][3][4][5][6][7][8][9][10][11][12] [1] ME [2] CSP 0.012 [3] PCSP 0.0180.308 [4] ECSP − 0.0450.542− 0.031 [5] TOTAL_Q 0.0840.0100.0100.003 [6] SIZE 0.0270.119− 0.334 0.252− 0.054 [7] ROA 0.2350.063− 0.052 0.0640.0950.187 [8] LEV − 0.040− 0.017− 0.143 0.053− 0.125 0.431− 0.106 [9] M_RET − 0.045− 0.008 0.017− 0.017 0.049− 0.063 0.048− 0.046 [10] STD_RET − 0.034− 0.086 0.0420.137− 0.035 0.057− 0.028 0.086− 0.119 [11] INST 0.0610.074− 0.034 − 0.107 0.0250.0350.073− 0.034 0.039− 0.861 [12] GROWTH − 0.071− 0.012 0.053− 0.017 0.106− 0.126 0.053− 0.062 − 0.035 0.043− 0.070 Table 3 Managerial eciency and CSP (test: H1a and H1b) The dependent variable is the CSP variable that indicates the total KLD-adjusted net performance score (CSP), product-related KLD- adjusted net performance score (PCSP), and environment-related KLD-adjusted net performance score (ECSP). A two-digit SIC dummy is used to control for industry xed e ects. All variables are dened in Table 1 ***, **, and * denote two-tailed signicance levels of 0.01, 0.05, and 0.10, respectively. The t-statistics reported in brackets are based on heteroskedasticity-robust standard errors. All analyses are controlled for industry and year xed e ects Variable DEP = CSPDEP = PCSP DEP = ECSP Intercept 20.851***− 1.339 − 21.072*** [4.56] [− 0.87] [− 17.97] ME − 0.049*0.032*** − 0.049*** [− 1.85] [3.64] [− 7.41] ROA 0.0490.023*** 0.017*** [1.54] [2.83] [2.87] SIZE 0.041***− 0.031*** 0.019*** [9.87] [− 21.94] [18.16] LEV − 0.145***0.001 − 0.038*** [− 6.98] [0.09] [− 7.48] INST − 0.044**0.043*** − 0.014*** [− 2.22] [6.55] [− 3.49] M_RET − 2.2710.971 − 1.784*** [− 1.01] [1.53] [− 3.83] STD_RET − 0.093***0.053*** − 0.008* [− 4.60] [7.96] [− 1.78] Industry e ects Ye sYe s Ye s Year e ects Ye sYe s Ye s Observations 11,03711,037 11,037 Adj. R 2 (%) 3.20 11.88 11.22 1 3 480 S. Y. Cho, C. Lee for the sake of brevity. Panel A of Table 5 shows the results of an ordered logistic regression. As we anticipated, about 42% of the total sample (4646 observations) has no change in CSP and 2445 (3946) observations reect an increase (a decrease) in CSP. Panel A shows that the change in CSP is positively associated with managerial eciency (coecient 0.029 and signicant at the 1% level). This result suggests that ecient managers increase CSP. In Panel B of Table 5, we report the results of the e ect of change in CSP for two subgroups (ecient and inecient manager groups). Two columns in Panel B show the similar, albeit weak, results as in the level model tests. For both the ecient and inecient groups, the coecients on ΔCSP, ΔPCSP, and ΔECSP are all positive (1.124, 1.865, and 2.058, for the ecient group, and 0.521, 1.611, and 2.191 for the inecient group) and statistically signicant (at the 1 ~ 10% level). However, as we showed in Table 4, there are di erences across ecient and inecient managers. For ΔCSP and ΔPCSP, the coef- cients of the ecient manager group are larger, suggest- ing that ecient managers can do a better job of capital - izing the changes in overall CSP (ΔCSP) and the change in product-related CSP (ΔPCSP) into the rm value. Similar to results in Table 4, while inecient mangers do better than ecient managers in the change in the environmental CSP (ΔECSP), the di erence in coecients is insignicant ( Z -statistic − 0.12). Post hoc Probing: Moderating Role of Managerial E ciency To better understand the moderating role of managerial e- ciency, we perform post hoc probing of the interaction e ect Table 4 Managerial eciency and CSP–CFP relation (test: H2a and H2b) ***, **, and * denote two-tailed signicance levels of 0.01, 0.05, and 0.10, respectively. The t-statistics reported in brackets are based on heter - oskedasticity-robust standard errors. All analyses control for industry and year xed e ects. Z-statistics are computed as follows:

Z= ( ̂HighME − ̂LowME ) √s2( ̂HighME )+s2( ̂LowhE ), where ̂HighME and ̂LowME are coeocient estimates from the two subsamples and s2() are the squared standard errors of the coe)cients The dependent variable is Total Q (TOTAL_Q). The regression analyses are based on weighted adjusted CSR performance metrics and the binary variable of managerial eciency. Columns (1) and (2) present the results for two subsamples partitioned on the median value of manage- rial eciency (HIGH ME vs. LOW ME) by using variant measures of corporate social responsibility performance. A two-digit SIC dummy is used to control industry xed e ects. All variables are dened in Table 1 Variable HIGH ME(1) LOW ME (2) Intercept 250.152 *** [3.04] 267.283 *** [3.22]318.470 *** [3.80] 262.019 *** [3.71]284.109 *** [3.98] 339.253 *** [4.68] CSP 1.106 *** [7.98] 0.670 *** [4.63] PCSP 1.648 *** [3.66]− 0.127 *** [− 0.37] ECSP 2.792 *** [8.32]2.911 *** [6.68] ROA 5.844 *** [6.27] 6.050 *** [6.46]5.918 *** [6.33] − 4.439 *** [− 5.30]− 4.528 *** [− 5.38] − 4.534 *** [− 5.43] SIZE − 0.619 *** [− 10.37] − 0.548 *** [− 9.45]− 0.581 *** [− 9.87] − 0.457 *** [− 8.01]− 0.399 *** [− 7.70] − 0.476 *** [− 8.50] LEV − 6.028 *** [− 17.07] − 5.940 *** [− 16.78]− 5.994 *** [− 16.97] − 3.792 *** [− 12.86]− 3.753 *** [− 12.68] − 3.807 *** [− 12.88] INST 0.061 [0.17] − 0.057 [− 0.16]− 0.039 [− 0.11] − 0.752 ** [− 2.12]− 0.792 ** [− 2.22] − 0.740 ** [− 2.08] GROWTH 3.758 *** [8.84] 3.664 *** [8.61]3.691 *** [9.71] 2.246 *** [7.32]2.249 *** [7.32] 2.261 *** [7.37] STD_RET 0.239 [0.70] 0.054 [0.16]0.036 [0.11] − 0.767 ** [− 2.37]− 0.840 [− 2.57] *** − 0.783 ** [− 2.41] Industry e ect Ye s Ye sYe s Ye sYe s Ye s Year e ect Ye s Ye sYe s Ye sYe s Ye s Observations 5,518 5,5195,519 5,5195,519 5,519 Adj. R 2 (%) 14.75 14.1714.33 11.4211.09 11.51 Di erence in coecient tests (Z-statistics): Di erence in CSP (2.17) Di erence in PCSP (3.13) Di erence in ECSP (− 0.22) 1 3 481 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance of ME and CSP. As reported in the previous section, we use subgroup analysis as a main approach to evaluate the inter - action e ects of managerial eciency and CSP. However, the subgroup analysis may not properly show how the e ect of CSP on CFP is moderated by managerial eciency (ME, e cient manager vs. ine cient manager). As in Holmbeck (2002), we conduct post hoc probing analysis in the following manner. First, we construct two groups (i.e., Group_Low and Group_High) by using the cat- egorized variable of managerial eciency. For Group_Low, the values are equivalent for the original ME variable (0 for inecient group, 1 for ecient group). For Group_High, the values are − 1 for the inecient group and 0 for the ecient group. Then, we construct an interaction term to incorporate these conditional moderator variables: Group_Low*CSP, Group_High*CSP. When we apply our basic CSP and CFP relation regression for each group variable and interaction term, we get the following pairs of estimated results for each CSP measure: All CSP main e ect (CSP) subgroups:

Product-related CSP main e ect (PCSP) subgroups:

Environment-related CSP main e ect (ECSP) subgroups:

As noted, the subgroup estimators are manipulated to generate group-specic equations. The rst regression in each pair estimates the Group_Low (Inecient, ME = 0) specic main equation. After replacing Group_Low and Group_High value as zeros, the main e ect equations will be summarized as follows: All CSP main e ect (CSP) subgroups Total Q =− 0.096 t0.024 CSP t0.023 Group\fLow o0.012 CSP aGroup\fLow TotalQ =0.107 o0.011 CSP o0.023 Group_\figh t0.012 CSP\bGroup_\figh Total Q =− 0.096 t0.008 PCSP t0.203 Grou\f_Low t0.03\b PCSP aGrou\f_Low TotalQ =0.107 o0.045 PCSP o0.203 Grou\f_Hi\bh o0.037 PCSP tGrou\f_Hi\bh TotalQ =− 0.096 t0.027 ECSP t0.204 Grou\f_Low o0.0\b7 ECSP aGrou\f_Low TotalQ =0.108 o0.029 ECSP t0.204 Grou\f_Hi\bh o0.057 ECSP aGrou\f_Hi\bh. Ine cient Group ∶Total Q=− 0.096 +0.0\f4 CS\b (t value =1.74 ) E cient Group∶ TotalQ=0.107 +0.011 \fSP (t value =0.\b8 ) Product-related CSP main e ect (PCSP) subgroups Environment-related CSP main e ect (ECSP) Subgroups Consistent with the results of our main analyses, the results show that the moderating e ect of ME for overall CSP is indistinguishable (0.024 and 0.012, for the ine- cient and ecient groups, respectively), while the moder - ating e ect is positive for PCSP (0.009 and 0.046, for the inecient and ecient groups, respectively) and statistically signicant for the ecient group at the 5% level. Finally, the moderating e ect on ECSP is negative (0.027 and − 0.030, for the inecient and ecient groups, respectively) and sta- tistically signicant at the 10% level. Overall, the results of the post hoc probing analysis are generally consistent with our main OLS regression results but demonstrate the moder - ating e ect of ME on the relation between CSP and CFP. In Fig. 2, we plot the moderating e ects of ME for three CSP measures; CSP, PCSP, and ECSP. Discussion and Conclusion The growing importance of CSR and rapidly increasing funds for socially responsible investment have placed con- siderable attention on the ecacy of CSR activities in terms of contributing to rm value. However, numerous e orts to identify the core of the murky relationship between rms’ CSP and CFP have not provided a clear answer. In this paper, we revisit the relationship between CSP and CFP by incor - porating the role of managerial eciency. We propose that the degree of managerial eciency inuences the choices of CSR activities and then, given a rm’s limited resources, ecient managers choose relatively economic or strategic CSR, which is expected to create rm value, rather than non-economic CSR. Also, we contend that more-ecient managers have high competence in executing CSR aligning with the intended objectives of a rm’s strategic plan, and thus ecient managers are likely to better capitalize CSP into the rm value. This study reveals that the discretionary nature of CSR activities helps to enhance (or impair) corporate nancial performance and is linked by the degree of managerial e- ciency. Specically, we nd that eciency works well to add value to the rm with product-related CSR (we view this as economic/strategic CSR), whereas eciency does not pay o  in environment-related CSR (we view this as altruistic/ Ine cient Group∶ TotalQ=− 0.096 +0.00\f PCSP (t \balue =0.52 ) E cient Group∶ TotalQ=0.107 +0.0\f5 PC\bP (t value =1.97 ) Ine cient Group∶ TotalQ=− 0.096 +0.0\f7 EC\bP (t value =1.85 ) E cient Group∶ TotalQ=0.108 −0.0\f0 ECSP (t \balue =−1.9\f ).

1 3 482 S. Y. Cho, C. Lee Table 5 Managerial eciency, change in CSP, and CFP regressions **, **, and * denote two-tailed signicance levels of 0.01, 0.05, and 0.10, respectively. The t-statistics reported in brackets are based on heter- oskedasticity-robust standard errors. All analyses control for industry and year xed e ects. Z-statistics are computed as follows:

Z= ( ̂HighME − ̂LowME ) √s2( ̂HighME )+s2( ̂LowhE ), where ̂HighME and ̂LowME are coeocient estimates from the two subsamples and s2() are the squared standard errors of the coe)cients In Panel A, the dependent variable of an ordered logistic regression model is change in CSP ( CSP) measured by change in total KLD-adjusted net score. The variable of CSP can take the following values: 0 (decrease in CSP), 1 (no change in CSP), or 2 (increase in CSP). Panel B reports the OLS regression results of Total Q on change in CSP using two subsamples partitioned on the median value of managerial e)ciency (HIGH ME vs. LOW ME). A two-digit SIC dummy is used to control for industry (xed eects. All variables are de(ned in Table1 Variable DEP=Ordered CSP Panel A: ordered logistic regression: managerial e)ciency and CSP ME 0.029*** [21.41] ROA 0.535*** [12.15] SIZE 0.255*** [344.69] LEV 0.222** [5.04] INST 0.158* [2.71] M_RET 2.549 [0.06] STD_RET 0.849*** [91.18] Industry eects Ye s Year eects Ye s Observations 11,037 Adj. R 2 (%) 5.02 VariableHIGH ME (1) LOW ME (2) Panel B: CSP and CFP conditioned on managerial eciency Intercept 256.481 *** [3.10]269.803 *** [3.24]282.656 *** [3.38] 279.652 *** [3.93] 293.918 *** [4.10] 303.375 *** [4.21] ΔCSP 1.124 *** [4.10]0.521 * [1.87] ΔPCSP 1.865 *** [3.17]1.611 * [1.88] ΔECSP 2.058 *** [3.21]2.191 *** [2.57] ROA 5.970 *** [6.38]6.002 *** [6.41]5.988 *** [6.39] − 4.488 *** [− 5.36]− 4.491 *** [− 5.37] − 4.532 *** [− 5.42] SIZE − 0.578 *** [− 9.64]− 0.555 *** [− 9.36]− 0.554 *** [− 9.44] − 0.422 *** [− 7.55]− 0.413 *** [− 7.77] − 0.420 *** [− 7.78] LEV − 6.011 *** [− 16.93]− 5.970 *** [− 16.80]− 5.957 *** [− 16.85] − 3.792 *** [− 12.78]− 3.784 *** [− 12.74] − 3.797 *** [− 12.78] INST − 0.081 [− 0.22]− 0.036 [− 0.10]− 0.048 [− 0.13] − 0.801 ** [− 2.24]− 0.773 ** [− 2.17] − 0.772 ** [− 2.16] GROWTH 3.744 *** [8.80]3.702 *** [8.70]3.681 *** [8.66] 2.260 *** [7.34]2.247 *** [7.31] 2.246 *** [7.31] STD_RET 0.056 [0.17]0.058 [0.17]0.098 [0.28] − 0.846 *** [− 2.59]− 0.841 [− 2.58] *** − 0.786 ** [− 2.40] Industry e ect Ye sYe sYe s Ye sYe s Ye s Year e ects Ye sYe sYe s Ye sYe s Ye s Observations 5,5185,5195,519 5,5195,519 5,519 Adj. R 2 (%) 14.2214.0714.04 11.1611.16 11.18 Di erence in coecient test (Z-statistics) ΔCSP (1.67)ΔPCSP (0.24) ΔECSP (− 0.12 ) 1 3 483 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance ethical CSR). These results suggest that more-ecient man- agers are more likely to engage in certain CSR activities that more directly contribute to corporate nancial performance.

Ultimately, our ndings suggest that managerial eciency is a moderator to consider in future studies on the CSP and CFP relation along with other moderators. Furthermore, we can o er the potential di erential e ects of this moderator depending on the types of CSR activities. These results are robust to the several di erent specications of the tests using both regressions (level and change models) and post hoc probing analyses.

This study contributes to the literature in several ways.

First, we introduce managerial eciency in CSR studies for the CSP–CFP relation. This can be an answer to the call for a study by Waddock and Graves (1997b) that argues cer - tain managerial characteristics could ll the gap between CSP and CFP. Second, the evidence of di erential e ects of CSP with the varying degree of managerial eciency on CFP suggests that managerial discretion impacts not only the determination of CSR investment but also the choice of par - ticular types of CSR. By demonstrating an example of such variation, this study sheds light on CSR literature by show - ing the importance of the types of CSR and matched e- ciency rule. In other words, we demonstrate that eciency rule is not a panacea for all cases of CSR decisions. Thus, managerial eciency can play a role of ex ante indicator of value-creating (or value-decreasing) CSP. Our study is subject to some caveats. First, although several recent studies show the validity of the measure of managerial eciency from di erent angles, measurement errors can still exist. Furthermore, despite the use of Total Q as the proxy for rm value capturing overall stakeholders’ wealth, other proxies for the nancial performance measure or nonnancial performance measure should be developed and used in the CSP–CFP relation studies because the long- term nonnancial e ect of a specic dimension of CSP on corporate nancial performance is not fully captured in the traditional measure of rm value. Also, several control variables adopted in our study cannot fully control for the e ect of macroeconomic factors that may constrain strategic choices of CSR activities, even among ecient managers.

Therefore, our conclusions should be interpreted with cau- tion. Finally, it is beyond the scope of this study to identify all the realms of CSR activities and to classify them into the right spectrum of underlying incentives (e.g., strategic, ethical, or altruistic CSR).

Despite the limitations of this study, our ndings can pro- vide insights for future research. One can study the strength of a tie between the social responsibility committee and the compensation committee within a board of directors. If there is a strong tie between the two committees, the rm is more likely to incorporate CSP in management compensation con- tracts, resulting in a di erential e ect on CSP and the rela- tion between CSP and CFP. Also, this study can be extended into the global setting, allowing for the e ects of cultural diversication and di erent social values on altruistic CSR because altruistic CSR is advocated to be the social norm in some countries. In addition, we look forward to studies measuring CSR activities using actual spending of CSR investment rather than the KLD CSR performance measure as a proxy of CSP. Such studies can provide more direct evidence on how managers eciently spend on CSR and whether the relationship between actual CSR spending and CFP can be varied. Compliance with Ethical Standards Ethical Approval This article does not contain any studies with human participants or animals performed by any of the authors. CSP –Total QP CSP –Total QE CSP –Total Q Fig. 2 Graphical post hoc probing: the role of managerial eciency.

This gure presents regression lines for the e ect of product-related CSP on CFP moderated by managerial eciency. The two partitioned groups are based on samples with high and low managerial eciency.

For detailed procedures to plot the regression lines, see Jose (2013, https://psychology.victoria.ac.nz/modgraph/) 1 3 484 S. Y. Cho, C. Lee Appendix: KLD Rating KLD STATS is an annual snapshot of CSP rating data with a binary (i.e., 1 or 0) summary of strength (positive) and concern (negative) ratings of a rm’s CSR. The following are the qualitative issue areas in KLD data.

KLD dimensionStrengths (1 or 0)Concerns (1 or 0) Community Charitable givingInvestment contro- versies Innovative giving Negative economic impact Non-US charitable giving Indigenous people relations Support for educa- tion Tax disputes Support for housing Other concern Indigenous peoples relations Volunteer programs Other strength Corporate govern- ance Limited compensa- tion award High compensation Ownership strength Ownership concern Transparency strength Accounting concern Political accountabil- ity strength Transparency concern Other strength Political accountabil- ity concern Other concern Diversity CEO diversityControversies Promotion Nonrepresentation Board of directors Other concern Work/life benets Women and minority contracting Employment of disabled Gay and lesbian policies Other strength Employee relations Union relationsUnion relations No-layo  policy Health and safety concern Cash prot sharing Workforce reductions Employee involve- ment Retirement benets concern Retirement benets strength Other concern Health and safety strength Other strength KLD dimension Strengths (1 or 0)Concerns (1 or 0) Environment Benecial products and services Hazardous waste Pollution prevention Regulatory problems Recycling Ozone-depleting chemicals Clean energy Substantial emissions Communications Agricultural chemicals PP&E environmental maintenance Other concern Management system Other strength Product QualityProduct safety R&D/innovation Marketing/contracting concern Benets to economi- cally Antitrust Disadvantaged Other concern Other strength Human rights Positive record in South Africa South Africa Indigenous peoples relations Northern Ireland strength Burma concern Labor rights strength Mexico controversies Other strength Labor rights concern Indigenous peoples relations Concern Other concern References Attig, N., & Cleary, S. (2015). Managerial practices and corporate social responsibility. Journal of Business Ethics, 131, 121–136.

Aupperle, K., Carroll, A., & Hateld, J. D. (1985). An empirical exami- nation of the relationship between corporate social responsibility and protability. Academy of Management Journal, 28, 446–463.

Bagnoli, M., & Watts, S. (2003). Selling to socially responsible con- sumers: Competition and the private provision of public goods.

Journal of Economics and Management Strategy, 12, 419–445.

Baik, B., Farber, D., & Lee, S. (2011). CEO ability and management earnings forecasts. Contemporary Accounting Research, 28, 1645–1668.

Barnett, M. L. (2007). Stakeholder inuence capacity and the variabil- ity of nancial returns to corporate social responsibility. Academy of Management Review, 32, 794–816.

Barnett, M. L., & Salomon, R. M. (2006). Beyond dichotomy: The curvilinear relationship between social responsibility and financial performance. Strategic Management Journal, 27, 111101–111122.

Baron, D. P. (2001). Private politics, corporate social responsibility, and integrated strategy. Journal of Economics and Management Strategy, 10, 1–75.

1 3 485 Managerial E ciency, Corporate Social Performance, and Corporate Financial Performance Baron, D. P., & Diermeier, D. (2007). Introduction to special issue on nonmarket strategy and social responsibility. Journal of Econom- ics and Management Strategy, 16, 539–545.

Benabou, R., & Tirole, J. (2010). Individual and corporate social responsibility. Economica, 77, 1–19.

Bhattacharya, C. B., & Sen, S. (2003). Consumer-company identica- tion: A framework for understanding consumers’ relationships with companies. Journal of Marketing, 67, 76–88.

Carroll, A. (1991). The pyramid of corporate social responsibility: Toward the moral management of organizational stakeholders.

Business Horizons, 34, 39–48.

Carroll, A. (2016). Carroll’s pyramid of CSR: Taking another look. International Journal of Corporate Social Responsibility, 1(1), 3–8.

Carroll, A., & Shabana, K. (2010). The business case for corporate social responsibility: A review of concepts, research and practice.

International Journal of Management Reviews, 12, 85–105.

Chakravarthy, B. (1986). Measuring strategic performance. Strategic Management Journal, 7, 437–458.

Cho, S. Y., Lee, C., & Pfei er, R. J. (2013). Corporate social respon- sibility performance and information asymmetry. Journal of Accounting and Public Policy, 32, 71–83.

Clogg, C., Petkova, E., & Haritou, A. (1995). Statistical methods for comparing regression coecients between models. American Journal of Sociology, 100(5), 1261–1293.

Demerjian, P., Lev, B., & McVay, S. (2012). Quantifying managerial ability: A new measure and validity tests. Management Science, 58, 1229–1248.

Demerjian, P., Lewis, M., Lev, B., & McVay, S. (2013). Managerial ability and earnings quality. The Accounting Review, 88, 463–498.

Deng, X., Kang, J., & Low, B. S. (2013). Corporate social responsibil- ity and stakeholder value maximization: Evidence from mergers.

Journal of Financial Economics, 110, 87–109.

Godfrey, P. (2005). The relationship between corporate philanthropy and shareholder wealth: A risk management perspective. The Academy of Management Review, 30, 777–798.

Hambrick, D. C. (2007). Upper echelons theory: An update. Academy of Management Review, 32, 334–343.

Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: The organi- zation as a reection of its top managers. Academy of Management Review, 9(2), 193–206.

Hillman, A. J., & Keim, G. D. (2001). Stakeholder value, stakeholder management, and social issues: What’s the bottom line? Strategic Management Journal, 22, 125–139.

Holmbeck, G. N. (2002). Post-hoc probing of signicant moderational and mediational e ects in studies of pediatric populations. Jour - nal of Pediatric Psychology, 27, 87–96.

Hong, B., Li, Z., & Minor, D. (2016). Corporate governance and execu- tive compensation for corporate social responsibility. Journal of Business Ethics, 136, 199–213.

Jayachandran, S., Kalaignanam, K., & Elert, M. (2013). Product and environmental social performance. Strategic Management Jour - nal, 34(10), 1255–1264.

Jensen, M. (2002). Value maximization, stakeholder theory, and the corporate objective function. Business Ethics Quarterly, 12, 235–256.

Johnston, J. S. (2005). Signaling social responsibility: On the law and economics of market incentives for corporate environmen- tal performance. University of Pennsylvania, Institute for Law & Economics Research Paper 05-16. Available at SSRN: http://ssrn.

com/abstract=725103 or https://doi.org/10.2139/ssrn.725103.

Jose, P. E. (2013). ModGraph-I: A programme to compute cell means for the graphical display of moderational analyses: The internet version, Version 3.0. Victoria University of Wellington, Welling- ton, New Zealand. Retrieved from https://psychology.victoria.

ac.nz/modgraph/ Kang, C., Germann, F., & Grewal, R. (2016). Washing away your sins? Corporate social responsibility, corporate social irresponsibility, and rm performance. Journal of Marketing, 80, 59–79.

Kaplan, S., & Zingales, L. (1997). Do investment-cash ow sensitivi- ties provide useful measures of nancial constraints? Quarterly Journal of Economics, 112, 159–216.

Kiron, D., Kruschwitz, N., Haanaes, K., & Velken, I. V. S. (2012). Sus- tainability nears a tipping point. MIT Sloan Management Review, 53(2), 69–74.

Koch, J. V., & Cebula, R. J. (1994). In search of excellent management. Journal of Management Studies, 31, 681–699.

Lev, B., Petrovits, C., & Radhakrishnan, S. (2010). Is doing good for you? How corporate charitable contributions enhance revenue growth. Strategic Management Journal, 31, 182–200.

Lopez, M. V., Garcia, A., & Rodriguez, L. (2007). Sustainable develop- ment and corporate performance: A study based on the Dow Jones Sustainability Index. Journal of Business Ethics, 75(3), 285–300.

Lubin, D. A., & Esty, D. C. (2010). The sustainability imperative. Har - vard Business Review, 88(5), 42–50.

Mackey, A., Mackey, T., & Barney, J. (2007). Corporate social respon- sibility and rm performance: Investor preferences and corporate strategies. The Academy of Management Review, 32, 817–835.

Malmendier, U., & Tate, G. (2005). CEO overcondence and corporate investment. The Journal of Finance, 60(6), 2661–2700.

Manescu, C. (2009). Is corporate social responsibility viewed as a risk factor? Evidence from an asset pricing analysis. Unpublished working paper, University of Gothenburg, Gothenburg, Sweden.

Manner, M. (2010). The impact of CEO characteristics on corporate social performance. Journal of Business Ethics, 93, 53–72.

Margolis, J., H. Elfenbein., & J. Walsh. (2009). Does it pay to be good … and does it matter? A meta-analysis of the relationship between corporate social and nancial performance. Unpublished working paper, Harvard University, Cambridge, MA.

McGuire, J., Dow, S., & Argheyd, K. (2003). CEO incentives and corporate social performance. Journal of Business Ethics, 45(4), 341–359.

McWilliams, A., & Siegel, D. (2001). Corporate social responsibil- ity: A theory of the rm perspective. Academy of Management Review, 26, 117–127.

Oh, W. Y., Chang, Y. K., & Cheng, Z. (2016). When CEO career hori- zon problems matter for corporate social responsibility: The mod- erating roles of industry-level discretion and blockholder owner - ship. Journal of Business Ethics, 133, 279–291.

Osagie, E. R., Wesselink, R., Blok, V., Lans, T., & Mulder, M. (2016). Individual competencies for corporate social responsibility: A literature and practice perspective. Journal of Business Ethics, 135, 233–252.

Peters, R., & Taylor, L. A. (2017). Intangible capital and investment-q relation. Journal of Financial Economics, 123, 251–272.

Servaes, H., & Tamayo, A. (2013). The impact of corporate social responsibility on rm value: The role of customer awareness.

Management Science, 59, 1045–1061.

Shiu, Y., & Yang, S. (2017). Does engagement in corporate social responsibility provide strategic insurance-like e ects? Strategic Management Journal, 38, 455–470.

Slater, D. J., & Dixon-Fowler, H. R. (2009). CEO international experi- ence and corporate social performance. Journal of Business Eth- ics, 89, 473–489.

Thomas, A., & Simerly, R. (1994). The chief executive ocer and corporate social performance: An interdisciplinary examination.

Journal of Business Ethics, 13, 959–968.

1 3 486 S. Y. Cho, C. Lee Waddock, S. A., & Graves, S. B. (1997a). The corporate social perfor- mance–nancial performance link. Strategic Management Jour - nal, 18, 303–319.

Waddock, S. A., & Graves, S. B. (1997b). Quality of management and quality of stakeholder relations: Are they synonymous? Business and Society, 30, 250–279. Waldman, D. A., Siegel, D. S., & Javidan, M. (2006). Components of CEO transformational leadership and corporate social responsibil- ity. Journal of Management Studies, 43, 1703–1725.

Williams, R. J., & Barrett, J. D. (2000). Corporate philanthropy, crimi- nal activity and rm reputation: Is there a link? Journal of Busi- ness Ethics, 26(4), 341–350.

1 3 Reproduced with permission of copyright owner. Further reproductionprohibited without permission.