To Understate Or To Overstate?

“

”

To Understate Or To Overstate?

Understanding The Interplay Between Internal- and External Drivers for

Corporate Social Responsibility Decoupling

H.A. KLOMP – s2962659

University of Groningen

Faculty of Economics and Business

Words: 11,427

Group: 21

Supervisor: dr. P.J. (Philip) Steinberg

Co-assessor: P. (Pere) Arque-Castells, PhD

To Understate Or To Overstate?

Understanding The Interplay Between Internal- and External Drivers for

Corporate Social Responsibility Decoupling

ABSTRACT

Purpose: Decoupling of Corporate Social Responsibility (CSR) actions amongst firms has increased

tremendously over the past years. This thesis aims to determine which internal drivers lead a firm to either understate- (more substantive than symbolic actions) or overstate (more symbolic than substantive

actions) their CSR actions. This thesis uses the dynamic capabilities (DC) perspective as its theoretical backbone and operationalises firms’ reconfiguration capabilities by studying firms’ asset reconfiguration. Additionally, this thesis examines whether the quality of the institutional environment moderates the relationship between a firm’s reconfiguration behaviour and its CSR gap (overstating or understating CSR behaviour).

Design/Methodology/Approach: Using panel data of 305 U.S. firms over a timeframe of nine years

(2009-2016) this study tests the hypothesised effects of independent variables (growth asset

reconfiguration and retrenchment reconfiguration) and a moderator variable (quality of institutional environment measured by political risk) on a firm’s CSR decoupling behaviour (CSR gap). A random effects panel data regression model was used to perform the analysis.

Findings: The results of the thesis show no significant influence of a firm’s growth asset reconfiguration

activities and the firm’s CSR gap. Furthermore, no evidence was found that there is an interaction

between a firm’s growth asset reconfiguration and the institutional environment on a firm’s CSR gap. The results do show that higher retrenchment reconfiguration activity in firms, increases the likelihood of that firm choosing to understate its CSR behaviour. No evidence was found that there is an attenuating effect between a firm’s retrenchment reconfiguration and the institutional environment on a firm’s tendency to understate. However, the opposite effect appears to be present, which goes against expectations that arose from the theory.

Originality/Value: This thesis enhances the understanding of the interplay between internal drivers

(reconfiguration capabilities) and external drivers (institutional environment) on firms’ CSR behaviour, more specifically that of CSR decoupling.

KEYWORDS: decoupling, corporate social responsibility, CSR, symbolic, substantive, overstate,

3

TABLE OF CONTENTS

1. Introduction 4

2. Theoretical Framework 7

2.1. Corporate social responsibility 2.2. CSR decoupling 2.3. Dynamic capabilities 2.4. Institutional environment 2.5. Developing hypotheses 3. Methods 17 4. Results 22

4.1. Descriptive statistics and correlations 4.2. Regression results

4.3. Robustness checks

5. Discussion and conclusion 30

5.1. Theoretical implications 5.2. Practical implications

5.3. Limitations and future research

6. References 35

7. Appendix 45

Appendix A - Conceptual model

Appendix B - Internal and external CSR actions Appendix C - Political risk components

Appendix D - Calculation of total political risk U.S. 2009-2016 Appendix E - Hausman-tests

1. INTRODUCTION

Over recent years Corporate Social Responsibility (CSR) has been a subject attracting considerable attention from scholars and managers alike (e.g. Campbell, 2007; Margolis & Walsh, 2003).Adhering to stakeholders’, social and environmental expectations has become crucial for firm survival (Campbell, 2007). Next to being one of the necessities in a firm’s proverbial ‘survival kit’, a firm’s CSR efforts can also play a noteworthy role in advancing the Sustainable Development Agenda (Luetkenhorst, 2004) - a sustainability programme consisting of 17 goals, adopted by all United Nations Member States (UN General Assembly, 2015). Additionally, the new solutions necessary for addressing these global challenges can generate new opportunities for firms (Luetkenhorst, 2004). All factors contribute to a magnitude of scholars and managers reaching a general consensus on firms needing to accept CSR (e.g. Campbell, 2007; Deegan, 2002). Ample research has looked at CSR through the lenses of various theories, including agency theory, stakeholder theory, the resource-based view of the firm and institutional theory among others (Lindgreen & Swaen, 2010).

A firm’s internal pressures and the external pressures from a firm’s social context can send opposing signals (Dhaliwal et al., 2012; Meyer & Rowan, 1977). Whereas internal pressures call for efficiency and consistency, adherence to external societal expectations can result in inefficiency and inconsistency (Meyer & Rowan, 1977; Tashman et al., 2019). To alleviate this tension, firms can decide to separate the communicated programme from the firm’s core operations. A process commonly referred to as ‘decoupling’ (Campbell, 2007; Meyer & Rowan, 1977; Tashman et al., 2019). The two sides separated from one another are commonly referred to as ‘internal’ and ‘external actions’ (Hawn & Ioannou, 2016). Hawn and Ioannou (2016) find that firms either overstate- or understate CSR. The former meaning a firm shows more external- than internal actions and the latter meaning it shows more internal- than external actions. The authors argue that either side of this spectrum harms a firm’s market value. However, some authors argue that decoupling can come to enhance an organisation’s image (Oliver, 1991), whilst keeping the internal workings of the firm unchanged (Meyer & Rowan, 1977).

5 institutional theorems, respectively (Bansal, 2005). The institutional theorem offers an external view, showing how organisational change – and, therefore, a firm’s CSR behaviour - can often be characterised by firms aiming to adhere to institutionalised norms of their environment (Bansal, 2005; Dimaggio & Powell, 1983). However, theorists recognise limitations to this explanation, as companies within the same institutional field can still show differences in CSR behaviour (Colwell & Joshi, 2013). A firm’s internal characteristics can change how a firm responds to its external environment (Delmas & Burbano, 2011). An extensive review of the existing literature suggests that there is a need to understand the interplay between the two perspectives. Many scholars have analysed how internal- and external factors affect CSR decision-making in firms, however, there is no agreement on which one is dominant (Hahn & Kühnen, 2013; Liu & Anbumozhi, 2009; Rosati & Faria, 2018). Lyon & Montgomery (2015) argue that the interplay between the two can aide in explaining the differences in organisations’ CSR behaviour. To further the literature on CSR decoupling through an explanation of the interplay between internal- and external drivers of CSR behaviour, this thesis aims to answer the following research question:

“How do a firm’s reconfiguration capabilities influence its CSR behaviour (overstating vs. understating)? And how is this relationship influenced by the firm’s institutional environment?”

This thesis pays particular interest in the dynamic capabilities perspective, rooted in the resource-based view of the firm (RBV) to explain the firm’s internal drivers for CSR behaviour. The dynamic capabilities (DC) area, rooted in RBV theory, argues that when a firm finds itself in a dynamic environment, it is required to reconfigure its resources and knowledge accordingly (Teece et al., 1997). The DC theory intents to explain a firm’s competitive advantage and performance in highly dynamic markets (Eisenhardt & Martin, 2000; Teece et al., 1997; Teece, 2007). DCs can play an important role in understanding how firms work toward CSR (Amui et al., 2017). Teece et al. (1997) propose three organisational and managerial processes as the three core elements of DCs: 1) coordination and integration; 2) learning; and 3) reconfiguration and transformation. This thesis will focus on the latter and operationalises this factor by studying firms’ asset reconfiguration. The institutional environment of the firm is approached as the aggregated factor ‘political risk’, that measures an appropriation of the overall score of the institutional environment of the United States. This thesis studies its moderating effect on the link between a firm’s asset reconfiguration activities and its CSR behaviour.

expected to increase the likelihood of understating. The former being expected due to the increased need to retain legitimacy whilst experiencing growth. This effect is expected to be attenuated by a well-developed institutional environment. The latter being expected due to Ullmann’s (1985) suggestion that firms may understate their CSR behaviour to disguise costs that come with substantive CSR behaviour. This effect is also expected to be attenuated by a well-developed environment.

This thesis adds to the body of literature by empirically testing the resource-based- and institutional factors’ interplay and influence on firms’ CSR decisions. By researching the internal- and external drivers of a firm’s decision to decouple its CSR actions, both theoretical perspectives are enriched by providing empirical evidence. It contributes to research on organisations and their environments within the field of CSR, as it aides in understanding the differences in behaviour resulting from organisational aspects and the external environment the firm operates in. Several studies have pointed out the need for an integrated study, as most studies thus far have focused on either the resource-based orientation or the institutional orientation (Bansal, 2005). Furthermore, this thesis enriches the literature on overstating- and understating of CSR behaviour. The outcome of this thesis contributes to managerial implications regarding CSR decisions in firms, as it is important to explain why one firm engages in CSR practices, whilst the other does not.

The results show no significant influence of a firm’s growth asset reconfiguration activities and its tendency to overstate its CSR behaviour. Furthermore, the hypothesised interaction between a firm’s growth asset reconfiguration and the quality of the institutional environment on a firm’s CSR gap is also not supported. The hypothesised effect of retrenchment reconfiguration activity on the tendency to understate CSR behaviour finds support in the results. The opposite effect to that of the hypothesised attenuating effect between a firm’s retrenchment reconfiguration and the institutional environment on a firm’s tendency to understate is present.

7

2. THEORETICAL FRAMEWORK AND DEVELOPMENT OF HYPOTHESES

This thesis utilises deductive reasoning and conceptual thinking by gathering- and analysing extant literature on the research matter. Through the literature review, several hypotheses arise, from which a conceptual model is constructed. The conceptual model can be found under appendix A.

2.1. Corporate social responsibility

Corporate social responsibility (CSR) has reached the top of many research agendas over the past years (e.g. Campbell, 2007; Margolis & Walsh, 2003). It is important to bear in mind that a firm’s decision-making process regarding CSR influences the impact of business in society, as well as a firm’s performance. Discussions in both managerial- and theoretical corners argue that CSR is not merely the right thing to do, but it also leads to better firm performance (Margolis & Walsh, 2003). However, these arguments remain controversial, with several researchers pointing out a positive relation between CSR and organisational outcomes (e.g. Waddock & Graves, 1997; Ruf et al., 2001), whilst others point to mixed results on this relation (e.g. Bai & Chang, 2015). According to institutional theory, these differences can be explained by the contextual settings that surround these firms’ activities (Margolis & Walsh, 2003). Whilst, the resource-based approach to CSR focuses on a firm’s resources to explain firm heterogeneity and a firm’s competitive advantage. Hart (1995) is one of the first studies that applies the resource-based view theory to CSR, arguing that CSR can be a resource or capability that leads to a competitive advantage. McWilliams and Siegel (2011) concur and discuss how CSR practices can be a source for a firm’s sustainable competitive advantage.

2.2 CSR decoupling

As CSR plays a significant role in organisation policies, communication of the topic to the external environment becomes essential to succeed in CSR performance (Dhaliwal et al., 2012). Firms face a tension between the social context it operates in and the internal pressures to remain efficient (Meyer & Rowan, 1977). For a firm to adhere to the social context, it adopts structures and practices that reflect the decree of the institutional environment the firm operates in (Dimaggio & Powell, 1983). Legitimacy pressures and competing societal expectations can result in inefficient- and inconsistent processes (Meyer & Rowan, 1977). In a dynamic environment these forces change constantly, making it difficult for firms to respond appropriately. To avoid these conflicting forces, firms can separate the two. This process is commonly referred to as ‘decoupling’ (Campbell, 2007; Meyer & Rowan, 1977; Tashman et al., 2019).

The two sides separated from one another are referred to as ‘substantive’ and ‘symbolic’ actions (Meyer & Rowan, 1977), or as ‘internal’ and ‘external’ actions (Hawn & Ioannou, 2016). Symbolic CSR actions are communicated to the external environment through CSR disclosure and allow a firm to at least appear to fulfil the requests of the external environment (Meyer & Rowan, 1977; DiMaggio & Powell, 1983). When the communicated programme is separated from the firm’s core operations, the signalling benefits of CSR actions are obtained, without investing the scarce resources and time that would

otherwise be necessary in actually following up these actions (Carlos & Lewis, 2018). Hawn and Ioannou (2016) refer to substantive CSR actions as ‘real actions to develop organisational capabilities and to meet the expectations of those social actors upon which the organisation depends for critical resources’. As opposed to symbolic measures, substantive CSR actions do require investment of resources and time and can cause/require significant changes to the firm (Hawn & Ioannou, 2016).

9 accusations if stakeholders detect decoupling activities that did not happen in ‘good faith’ (Meyer & Rowan, 1977; Seele & Gatti, 2017). In the context of firms understating their CSR efforts, CSR practices are not actively communicated and external audiences could build the assumption that the firm is not engaging in CSR (Hawn & Ioannou, 2016), resulting in firms not reaping the benefits of their CSR engagements. This process is commonly referred to as ‘brownwashing’ (Kim & Lyon, 2015). This thesis uses the terms ‘overstating’ and ‘understating’ (as opposed to ‘greenwashing’ and ‘brownwashing’, respectively), as it also incorporates the social factor of CSR, rather than the mere environmental

concerns. To further the literature on CSR decoupling, this thesis examines the interplay between internal- and external drivers for CSR decoupling.

2.3. Dynamic capabilities

This paper uses the dynamic capabilities (DC) perspective (rooted in resource-based view literature) as its theoretical backbone to examine the aforementioned internal drivers. Researchers often consult the ‘Resource-Based View’ (RBV) of the firm when dissecting the process of creating a competitive advantage. The RBV of the firm is an influential framework that aides in understanding the firm’s internal environment workings in this process (Barney, 1991). The theory argues that a firm’s resources should be ‘1) valuable; 2) rare; 3) in-imitable; and 4) non-substitutable’ (Barney, 1991). The theory argues that developing firm-specific competences and capabilities aides in creating a firm-firm-specific competitive advantage and can be used in understanding a firm’s decisions regarding sustainability (Barney, 1991; Porter, 1991).

subsequently alter the existing resource base to align internal configuration with the external environment of the firm (Helfat et al., 2007; Teece, 2007).

Teece et al. (1997) argues there are three organisational- and managerial processes as core elements of dynamic capabilities: 1) coordination and integration; 2) learning; and 3) reconfiguration and transformation. The DCs of a firm can be disaggregated into three separate capabilities, it is important to note, however, that they are related to one another (Teece et al., 1997). Their processes are a subset of the terms referred to earlier: ‘sensing, seizing and managing (i.e. reconfiguring)’. In order to sense, seize and, eventually, reconfigure the firm when change occurs, it is required of a firm to handle its resources appropriately (Teece, 2007). Internal organisational routines (e.g. gathering and processing company- and customer information to be integrated into the production process) are attributed to the process of ‘coordination and integration’. The ‘learning’ process is considered to be more important than coordination and integration (Teece et al., 1997). This process enables firms to perform tasks more efficiently through repetition and experimentation, as well as enables new opportunities to be identified (Teece et al., 1997). The last process ‘reconfiguration and transformation’ is placed in high importance as well, as it is of value for firms to acquire the ability to sense the need for reconfiguration in highly dynamic environments (Teece et al., 1997; Amit & Schoemaker, 1993). Aragon-Correa and Sharma (2003) argue that the way a firm develops and manages its dynamic capabilities is pivotal to whether it will react passively- or proactively to changes in its market environment. Organisations that deal with CSR development have to deal with unpredictable changes (Hammervoll et al., 2012) and ambiguity (Aragon-Correa & Sharma, 2003) in complex environments, making DCs an adequate theory to apply in CSR. This thesis focuses on ‘the third leg of the dynamic capabilities triad’ (Helfat & Peteraf, 2014): reconfiguration of the firm.

11 March, 1993). Capabilities and bundles of resources grow over time, creating path dependencies. Firms often remain within the boundaries of capabilities that have previously ensured a reliable and efficient result. It creates a reluctancy toward change, even when the environmental condition the firm operates in changes (Hannan & Freeman, 1984; Levinthal & March, 1993). When the firm attempts to address the dynamic environment, it might get stuck to its core rigidities, which comes at a cost of the firm’s long term success (Barnett & Pontikes, 2008; Hannan & Freeman, 1984; Kelly & Amburgey, 1991, Uotila et al., 2009). The capabilities trap is notable in CSR management as the link between CSR activity and economic performance is not always evident (Margolis et al., 2007).

Acquisition, resource deployment, and asset divestiture are all elements of the dynamic process in which firms can change their resource base – internally and externally (Capron et al., 2001). The traditional approach to asset reconfiguration distinguishes between growth and retrenchment reconfiguration (Chakrabarti et al., 2011). Growth asset reconfiguration is associated with improving the firm’s competitive position, whilst retrenchment reconfiguration improves the firm’s internal efficiency (Anand and Singh, 1997). Growth asset reconfiguration can take different forms, which can mainly be attributed to sourcing assets externally- and internally. External sourcing can include purchasing or acquiring assets, as well as collaboration with partners in developing resources (Rosenkopf & Nerkar, 2001). Lack of experience and difficulties in integration can disable a firm to reap the benefits of external sourcing (Haleblian & Finkelstein, 1999; Capron et al., 2001). Whilst internal sourcing is often more facile for firms, as internally developed path dependencies are more easily understood (Karim & Mitchell, 2000). Asset retrenchment offers opportunities in ‘reducing assets as a means of mitigating conditions for a financial downturn’ (Robbins & Pearce, 1992). Extant literature tends to focus on the negative side of divestiture, arguing that it is a sign of acquisition failure. However, some scholars show that divestiture can also be a sign of a firm gaining scale efficiency by reducing their excess capacity (Capron et al., 2001). By divesting assets, firms can reduce their excess capacity and increase efficiency (Anand & Singh, 1997).

2.4. Institutional environment

it increases its survival capabilities (Kondra & Hinings, 1998). The institutional environment of a firm can directly influence a firm’s decision-making process regarding resource reconfiguration. Additionally, it influences the eventual benefits and/or detriments that come with its reconfiguration decisions (Chakrabarti et al., 2011). The quality of the institutional environment can create restrictions for a firm’s strategic management in the sense that when underdeveloped, a firm would have to deal with higher transactions costs and a constrained environment (Hoskisson et al., 2000). The topic of weak institutions being detrimental for firms is surrounded by a cloud of controversy, as some research suggests that firms can devise structures and strategies to fill institutional voids, reaping benefits in the process (Hoskisson et al., 2000; Jackson & Apostolakou, 2010; Oliver, 1991). It is clear that extant literature paints a picture of both challenges (Bruton et al., 2003; Hoskisson et al., 2000) as well as benefits (Hoskisson et al., 2000; Jackson & Apostolakou, 2010; Oliver, 1991) for a firm facing an environment with weak market-oriented institutions.

13

2. 5. Developing hypotheses

Growth reconfiguration in regards to overstating CSR behaviour. I posit that growing firms tend to

overstate their CSR behaviour due to 1) increasing external pressures as they expand; as well as 2) the increased need to retain legitimacy. Mitchell et al. (1997) and Kim and Lyon (2015) argue that firms that experience growth, gain more attention from their environment. Bowen (2014) indicates that as institutional pressures increase, substantive actions of firms become limited as they will continuously try to adhere to the stakeholder pressures through symbolic actions. With firm growth arises the need to retain legitimacy (Meyer & Rowan, 1977), giving a firm incentives to overstate CSR behaviour as a green image will help in this regard. With growth often comes increased exposure to external actors, including those of regulatory nature (Decker, 2003; DiMaggio & Powell, 1983; Zucker 1987). Decker (2003) and Innes and Sam (2008) find that regulatory bodies treat firms that showcase environmentally-friendly behaviour more favourably as opposed to firms that do not. The arguments for the increasing pressures and the increased need to retain legitimacy result in the following hypothesis:

H1A. The higher the growth reconfiguration activity of the firm, the higher the likelihood of that

firm choosing to overstate its CSR behaviour.

Attenuating/moderating effect of institutional environment. Dhaliwal et al. (2012) argue that in a developed

H1B. The effect of growth reconfiguration on a firm overstating its CSR behaviour is attenuated

by the quality of the institutional environment the firm operates in.

Retrenchment reconfiguration in regards to understating CSR behaviour. One of the incentives for firms

to understate their CSR behaviour is associated with the associated costs of substantive CSR behaviour (Ullmann, 1985). Various constantly changing pressures from the external environment make it difficult for firms to adhere to every need presented by its stakeholders. Establishing organisational changing routines for CSR behaviour, requires a firm to overcome existing path dependencies to avoid the capabilities trap. Firms can easily get caught in a rut due to these path dependencies, as they are deeply rooted in the firm’s day-to-day activities and, therefore, difficult to change (Helfat & Peteraf, 2003; Winter, 2003; Leaonard-Barton, 1992). Breaking these impediments needs a lot of effort and continuous attention. Moreover, when comparing symbolic CSR actions to substantive ones, it becomes evident that substantive actions often require investments (time and money) as well as persistence. Substantive CSR actions call for change processes that affect the organisational structure and sometimes even its culture (Eccles et al., 2014; Hawn & Ioannou, 2016). Ullmann (1985) suggests that firms may underreport their CSR behaviour to disguise the associated costs. Asset retrenchment offers opportunities in ‘reducing assets as a means of mitigating conditions for a financial downturn’ (Robbins & Pearce, 1992). In the case of financial downturn, salience of shareholders increases (Kim & Lyon, 2015), as low profitability increases market scrutiny (Morrow et al., 2007). Shareholders may suspect that the firm is pursuing these CSR actions at the expense of more profit-oriented programmes (Ullmann, 1985). As established before, the link between CSR and firm performance is not always clear. Researchers have pointed out before that whilst CSR may be detrimental for short-term firm performance, it might be beneficial in the long run (Margolis & Walsh, 2003). This increases a firm’s exposure to uncertainty and risks, especially with respect to the return on their investments. I expect that the increased risks force firms to weigh their CSR disclosure more carefully. This results in the following hypothesis:

H2A. The higher the retrenchment reconfiguration activity of the firm, the higher the likelihood

of that firm choosing to understate its CSR behaviour.

Attenuating/moderating effect of institutional environment. I posit that in a developed institutional

15 structures that enable them access to decision makers (Schneiberg & Soule, 2005). I posit that the urgent calls for short-term profits is diminished by the calls for corporate socially responsible behaviour. Additionally, in a developed institutional environment regulatory powers show a higher capacity to enforce the regulatory rules of CSR (North, 1990; Ostrom, 1990). The salience of these external pressures increases in a well-developed institutional environment and results in the following hypothesis:

H2B. The effect of retrenchment reconfiguration on a firm understating its CSR behaviour is

attenuated by the quality of the institutional environment the firm operates in.

3. METHODOLOGY

This research is of a theory-testing nature, as it empirically tests the influence of institutional factors on the dynamic capabilities theory. It follows the proposed approach for theory testing by Viswesvaran & Ones (1995). After important constructs and relationships are determined in the literature review, measures are identified for each of the constructs. After collection of the data, the covariation across the measures is tested. Finally, the model is empirically tested to ensure it fits the collected data (Viswesvaran & Ones, 1995).

Data sources

To test the proposed hypotheses, this research will make use of longitudinal firm-level secondary data of firms based in the U.S. (2010-2016). The data on the dependent variable (firm’s CSR actions) is gathered through the Thomson Reuters database ASSET4. The data on financial and business information used to calculate the reconfiguration activity of a firm is gathered through Thomson Reuters ASSET4 and missing data is added through Bureau van Dijk’s database Orbis. Control variables are gathered from both of the aforementioned sources. Political risk measures are gathered from The International Country Risk Guide (ICRG), provided by The PRS Group. The initial size of the sample consists of 2,573 U.S. based firms. After gathering the CSR data from Thomson Reuter’s ASSET 4, the final sample included 305 firms.

Measurements Dependent variables

Substantive and symbolic CSR actions. Data on the substantive and symbolic actions of firms was gathered

through the Thomson Reuters database ASSET 4. For measuring the substantive and symbolic CSR actions, Hawn and Ioannou’s (2016) approach was followed. Their list of internal and external CSR actions can be found under appendix B. An assumption that could have been made in the process of data gathering is that a missing value would mean that a firm did not perform this CSR action. However, as the literature review pointed out, a firm may choose not to report its CSR behaviour. This assumption could, therefore, not be made. This research uses listwise deletion to reach a dataset that is as complete as possible. After cleaning up the data, a sample of 305 firms remains.

Data on external actions of the firm consisted of binary data. To arrive at a score (in percentage), the sum of the 20 datapoints as divided by the number of datapoints and multiplied by 100. Several datapoints on the internal actions of the firm consisted of percentages: 1) Board Gender Diversity; 2) Audit

Committee Non Executive Members; 3) Nomination Committee Non Executive Members; and 4) Independent Board Members. Industry averages were calculated (based on the two-digit SIC codes that

17 average were allocated a ‘1’, whilst ones that performed below the industry average were allocated a ‘0’. After which a sum of the datapoints was calculated, after which this sum was divided by the number of datapoints and multiplied by 100, to arrive at a score (in percentage) for the internal actions.

To arrive at a score of substantive and symbolic CSR actions, formulas 1 and 2 are used, following Hawn and Ioannou’s (2016) approach.

𝑆𝑢𝑏𝑠𝑡𝑎𝑛𝑡𝑖𝑣𝑒 𝐶𝑆𝑅 𝐴𝑐𝑡𝑖𝑜𝑛𝑠𝑡 = 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑙 𝐶𝑆𝑅 𝐴𝑐𝑡𝑖𝑜𝑛𝑠𝑡 𝐿𝑜𝑔𝑔𝑒𝑑 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠𝑡 Formulae 1 𝑆𝑦𝑚𝑏𝑜𝑙𝑖𝑐 𝐶𝑆𝑅 𝐴𝑐𝑡𝑖𝑜𝑛𝑠𝑡 = 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝐸𝑥𝑡𝑒𝑟𝑛𝑎𝑙 𝐶𝑆𝑅 𝐴𝑐𝑡𝑖𝑜𝑛𝑠𝑡 𝐿𝑜𝑔𝑔𝑒𝑑 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠𝑡 Formulae 2

CSR gap. To arrive at a score for the CSR gap, formula 3 is used. The CSR gap is measured by subtracting

the substantive CSR actions from the symbolic CSR actions and dividing this by the logged total assets. A natural logarithm of the total assets is used to normalise the data. When this outcome is negative, the firm tends to understate its CSR behaviour, whereas if the outcome is positive, the firm tends to overstate its CSR behaviour (Hawn & Ioannou, 2016).

𝐶𝑆𝑅 𝐺𝑎𝑝𝑡 =

𝑆𝑦𝑚𝑏𝑜𝑙𝑖𝑐 𝐶𝑆𝑅 𝐴𝑐𝑡𝑖𝑜𝑛𝑠𝑡− 𝑆𝑢𝑏𝑠𝑡𝑎𝑛𝑡𝑖𝑣𝑒 𝐶𝑆𝑅 𝐴𝑐𝑡𝑖𝑜𝑛𝑠𝑡

𝐿𝑜𝑔𝑔𝑒𝑑 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠𝑡

Formulae 3

Independent variables

Growth reconfiguration. Growth reconfiguration includes both external growth and internal development

in a given year t over total assets in a given year t. Retrenchment reconfiguration. Disposal of fixed assets over total assets. This results in formulas 4 & 5 following Chakrabarti et al.’s approach (2011). There is a risk of omitted variable bias resulting from the selective consideration of reconfiguration in the overall ‘triad’ of dynamic capabilities that usually incorporate: 1) coordination and integration; 2) learning; and 3) reconfiguration. This risk is further discussed in the last section of this thesis.

𝐺𝑟𝑜𝑤𝑡ℎ 𝑟𝑒𝑐𝑜𝑛𝑓𝑖𝑔𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑡=

𝑇𝑜𝑡𝑎𝑙 𝑔𝑟𝑜𝑤𝑡ℎ𝑡

𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑡

𝑅𝑒𝑡𝑟𝑒𝑛𝑐ℎ𝑚𝑒𝑛𝑡 𝑟𝑒𝑐𝑜𝑛𝑓𝑖𝑔𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑡 =

𝐷𝑖𝑠𝑝𝑜𝑠𝑎𝑙 𝑜𝑓 𝑓𝑖𝑥𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠𝑡

𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑡

Formulae 5

Moderator

Political risk. Many researchers argue that political conditions are among the most important factors

influencing the external environment of the firm, and, in turn, corporate disclosures (e.g. Williams, 1999). This thesis builds on the assumptions of ‘coercive isomorphism’, which states that governments - through regulatory bodies - can impose rules on corporations to comply with certain CSR pressures (Henriques & Sadorsky, 1999; Porter & van der Linde, 1995). Political risk measures are gathered from The International Country Risk Guide (ICRG), provided by PRS Group. ICRG gathers 22 variables in three subcategories of risk: political, financial, and economic. For each of the subcategories, a separate index is created. Next to the regulatory environment, the Political Risk Rating consists of 12 variables that cover both political and social factors. The Political Risk Rating ensures consistency over time (and between countries) through a series of pre-set questions for each risk component (PRS Group, 2019). This consistency makes the variable more suitable for time-series analysis as opposed to – for instance – the Environmental Performance Index.

ICRG measures political risk through 12 components. The PRS Group assigns weights to each ‘risk component’ based on its importance and ascribes risk points. This number of points can range from zero to a number depending on the fixed weight of the component given in the overall political risk assessment.

Appendix A offers an overview of the risk components, weights, and sequence that are used to produce the

political risk rating. It is important to note that several components are constructed through more sub-components. For example, Government Stability is made up out of the subcomponents Government Unity,

Legislative Strength, and Popular Support. Whilst Socioeconomic Conditions is made up out of Unemployment, Consumer Confidence, and Poverty.

Formula 7 summarises the calculation of political risk. Where PRUSt equals the political risk of the

U.S. at time t, RiskUSt is the index score for the risk component of the U.S. at time t, and Point equals the maximum amount of points the component can score.

19 represent the unscaled political risk. This is then rescaled by multiplying with 100

12 to receive the total

political risk per year.

Control variables

To control for other variables that may influence the decoupling of CSR, control variables are implemented. Other researchers have found that firm size, industry type and financial performance are important control variables in similar research (Graafland & Smid, 2016). I, therefore, use firm size, industry, financial

performance as control variables. Additionally, I control for R&D intensity, sales growth, and financial leverage.

Firm size. Firm size has been found to be an influence on a firm’s CSR decoupling behaviour

(Shevchenko et al., 2016; Wickert et al., 2016). Larger firms become more visible, which attracts attention from the external environment. This may lead firms to report their CSR behaviour more favourably. To control for firm size, the number of employees was obtained from the Thomson Reuters database. A natural logarithm of the variable was used to normalise the data.

R&D intensity. R&D intensity has been found to influence CSR outputs. Other researchers have

previously used R&D intensity as a proxy for intangibles (Hawn and Ioannou, 2016). The data was gathered from Thomson Reuters’ Eikon database. To account for missing values, the industry average was calculated- and used for the firms that did not report R&D expenditures. This was based on SIC industry codes, gathered from Bureau van Dijk’s database ‘Orbis’. R&D intensity was calculated as R&D expenses over total sales. A natural logarithm of the variable was used to normalise the data.

Sales growth. Previous research has taken sales growth as a variable to account for a firm’s

capability to invest in substantive CSR actions (Hussain et al., 2018). The percentage of change in revenue was gathered from the Thomson Reuters’ database ‘Eikon’. A natural logarithm of the variable was used to normalise the data.

Financial performance. Margolis & Walsh (2003) argue that a firm’s CSR behaviour is (indirectly)

related to its performance. To account for this factor return on assets (ROA) was gathered from Thomson Reuters’ database ‘Eikon’. A natural logarithm of the variable was used to normalise the data.

Financial leverage. Financial leverage, measured as the total liabilities of the firm over its total

assets, was gathered from Thomson Reuters’ database Eikon. Financial leverage has been found to have an effect on the use of the assets with regards to CSR performance (Tashman et al., 2019).

Industry. As the industry a firm operates in may influence a firm’s CSR behaviour (Delmas &

Industry SIC Freq. Percent Cum.

Mining 1000-1999 208 8.5 8.5

Manufacturing 2000-3999 752 30.8 39.3

Transportation, communications, electric, gas and sanitary service

4000-4999 296 12.1 51.5

Wholesale trade 5000-5199 72 3.0 54.4

Retail trade 5200-5999 200 8.2 62.6

Finance, insurance and real estate 6000-6799 616 25.2 87.9

Services 7000-8999 296 12.1 100

Total 1449 100.00

Table 1 – Overview of industries Analysis

To test the hypotheses, a panel data regression analysis is performed. A Hausman test is performed for each model in STATA to see whether a fixed-effects or random-effects model is appropriate. Firstly, multicollinearity is tested by conducting a pairwise correlation test and an estimation of VIF-values. Model 1 tests Panel Data Analysis to determine the influence of the control variables on the CSR gap, over time. Model 2 includes the independent variables. Model 3 and 4 include the interaction effect of the institutional environment. Lastly, model 5 includes all the aforementioned variables and control variables. After testing the main models, I test for robustness to see whether the results hold under slightly different circumstances.

Models:

1. 𝐶𝑆𝑅 𝐺𝐴𝑃𝑖𝑡 = 𝛽0 + 𝛽1 ∙ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑖𝑡+ 𝛽2 ∙ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑑𝑢𝑚𝑚𝑖𝑒𝑠𝑖𝑡 + ∈

Where CSR GAP stands for the CSR gap of a firm; Control variables stand for firm size measured by the natural logarithm of the number of employees; sales growth measured by the natural logarithm of change in revenue; firm performance measured by the logged return on assets; the natural logarithm of the firm’s financial leverage; the logged R&D intensity. Industry dummies stand for the attributed industries, the dummy variables take “1” when the firm operates in the industry and “0” when it does not. i stands for firm and t stands for year. ∈ is the error term. 2. 𝐶𝑆𝑅 𝐺𝐴𝑃𝑖𝑡 = 𝛽0 + 𝛽1 ∙ 𝐺𝑅𝑂𝑖𝑡+ 𝛽2 ∙ 𝑅𝐸𝑇𝑅𝑖𝑡+ 𝛽3 ∙ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑖𝑡+ 𝛽4 ∙

21 Where GRO stands for growth asset reconfiguration and RETR for retrenchment reconfiguration.. 3. 𝐶𝑆𝑅 𝐺𝐴𝑃𝑖𝑡 = 𝛽0 + 𝛽1 ∙ 𝐺𝑅𝑂𝑖𝑡+ 𝛽2 ∙ 𝑅𝐸𝑇𝑅𝑖𝑡+ 𝛽3 ∙ 𝐺𝑅𝑂 ∗ 𝑃𝑂𝐿𝑖𝑡+ 𝛽4 ∙ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑖𝑡+

𝛽5 ∙ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑑𝑢𝑚𝑚𝑖𝑒𝑠𝑖𝑡 + ∈

Where GRO * POL is the interaction between growth asset reconfiguration and political risk. 4. 𝐶𝑆𝑅 𝐺𝐴𝑃𝑖𝑡 = 𝛽0 + 𝛽1 ∙ 𝐺𝑅𝑂𝑖𝑡+ 𝛽2 ∙ 𝑅𝐸𝑇𝑅𝑖𝑡+ 𝛽3 ∙ 𝑅𝐸𝑇𝑅 ∗ 𝑃𝑂𝐿𝑖𝑡 + 𝛽4 ∙ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑖𝑡+

𝛽5 ∙ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑑𝑢𝑚𝑚𝑖𝑒𝑠𝑖𝑡 + ∈

Where RETR * POL is the interaction between retrenchment reconfiguration and political risk 5. 𝐶𝑆𝑅 𝐺𝐴𝑃𝑖𝑡 = 𝛽0 + 𝛽1 ∙ 𝐺𝑅𝑂𝑖𝑡+ 𝛽2 ∙ 𝑅𝐸𝑇𝑅𝑖𝑡+ 𝛽3 ∙ 𝐺𝑅𝑂 ∗ 𝑃𝑂𝐿𝑖𝑡+ 𝛽4 ∙ 𝑅𝐸𝑇𝑅 ∗ 𝑃𝑂𝐿𝑖𝑡+

𝛽5 ∙ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑖𝑡+ 𝛽6 ∙ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑑𝑢𝑚𝑚𝑖𝑒𝑠𝑖𝑡 + ∈

4. RESULTS 4.1. Descriptive statistics and correlations

Regression analyses lean on the assumption that its variables are normally distributed so outliers do not drive the results of the analysis. For several variables this was not the case, therefore, some were converted to their natural logarithm (Baltagi, 2008). Table 2 and table 3 show the descriptive statistics and the correlation matrix, respectively. As using a natural logarithm of variables can cause multicollinearity issues, it is important to check whether correlation between predictors is present. Table 3 shows that in this thesis, correlation between independent variables is low, with the exceptions of the correlation between growth

asset reconfiguration * political risk and growth asset reconfiguration (0.95), and retrenchment reconfiguration * political risk and retrenchment reconfiguration (0.86). High correlation values in this

field of research is rather common and should, therefore, not cause too many issues. To check whether this causes multicollinearity, the variance inflation factors (VIF) were gathered. Table 4 shows multicollinearity between the interaction variable growth asset reconfiguration * political risk and the independent variable

growth asset reconfiguration as its value exceeds 10 (Hair et al., 2010). Issues that normally arise from

multicollinearity, however, may not arise in this case. Disatnik & Sivan (2014) argue that high correlation between an interaction term and an independent variable (or control variable) on which the interaction term is built is common and it does not have to impact the coefficients in regression analyses. To validate this, close attention is paid to model 4 that shows a regression without the interaction term of growth asset

reconfiguration * political risk. An additional VIF test was run - excluding the aforementioned interaction

23

(1) (2) (3) (4) (5)

Variables (for all firms: 305) N mean sd min max

CSR Gap 2,440 -1.378 0.919 -3.666 2.269

Growth asset reconfiguration 2,440 0.0838 0.121 0 0.888

Retrenchment reconfiguration 2,440 0.0458 0.182 0 3.330

Growth asset reconfiguration * Quality of inst. env (political risk)

2,440 6.9751 10.05 0 73.57

Retrenchment reconfiguration * Quality of inst. env (political risk)

2,440 3.8306 15.27 0 280.88

Firm size (log.) 2,440 9.104 1.716 3.332 14.60

Sales growth (log.) 2,440 17.71 3.132 13.07 25.20

Financial performance (log.) 2,440 1.524 1.016 -4.758 4.357

Financial leverage (log.) 2,440 0.479 0.163 0.0250 2.749

R&D intensity (log.) 2,440 0.165 0.241 0.000154 3.964

Manufacturing industry (dummy) 2,440 0.308 0.462 0 1

Mining industry (dummy) 2,440 0.0852 0.279 0 1

Transportation industry (dummy) 2,440 0.121 0.327 0 1

Wholesale industry (dummy) 2,440 0.0295 0.169 0 1

Retail industry (dummy) 2,440 0.0820 0.274 0 1

Finance industry (dummy) 2,440 0.252 0.435 0 1

Services industry (dummy) 2,440 0.121 0.327 0 1

25

Variable VIF 1/VIF

Growth asset reconfiguration 12.28 0.081466

Growth asset reconfiguration * Political risk 11.56 0.086528

Manufacturing industry (dummy) 8.20 0.121893

Finance industry (dummy) 7.82 0.127800

Services industry (dummy) 4.69 0.213210

Transportation industry (dummy) 4.67 0.214236

Retrenchment reconfiguration 4.07 0.245430

Retrenchment reconfiguration * Political risk 3.98 0.251449

Mining industry (dummy) 3.77 0.264999

Retail industry (dummy) 3.58 0.279581

Firm size (log.) 1.55 0.646419

Financial performance (log.) 1.35 0.738215

Sales growth (log.) 1.27 0.786520

R&D Intensity (log.) 1.25 0.799773

Financial leverage (log.) 1.19 0.837811

Political risk (quality of inst. environment) 1.03 0.969549

Mean VIF 4.52

Table 4 – Variable Inflation Factors

Variable VIF 1/VIF

Manufacturing industry (dummy) 8.17 0.122325

Finance industry (dummy) 7.81 0.128037

Services industry (dummy) 4.68 0.213780

Transportation industry (dummy) 4.67 0.214294

Retrenchment reconfiguration 4.03 0.247913

Retrenchment reconfiguration * Political risk 3.98 0.251449

Mining industry (dummy) 3.77 0.265420

Retail industry (dummy) 3.56 0.281115

Firm size (log.) 1.55 0.646770

Financial performance (log.) 1.31 0.761276

R&D Intensity (log.) 1.25 0.800193

Sales growth (log.) 1.23 0.810139

Financial leverage (log.) 1.19 0.837833

Growth asset reconfiguration 1.16 0.859862

Political risk (quality of inst. environment) 1.01 0.987906

Mean VIF 3.29

4.2. Regression results

After preparing the data for panel data analysis, a Hausman-test is performed on every model to see whether a fixed-effects model or a random-effects model is appropriate. Appendix E offers an overview of the Hausman-tests per model. To determine whether a panel effect or a random-effects model should be used, a Breusch-Pagan Lagrange Multiplier test was performed for every model. Appendix F offers an overview of the Breusch-Pagan LM tests. Both tests indicate that a random-effects model is the most appropriate for this thesis, meaning the study is focused on the effects between firms. Whereas in a fixed-effects model it would check within the variants, within the same firm over the years.

Model 1 includes only the effect of the control variables on the CSR gap. Model 2 includes the

independent variables growth asset reconfiguration and retrenchment reconfiguration. In model 3 the interaction effect growth asset reconfiguration * political risk is added. Model 4 excludes the aforementioned interaction effect, but includes the interaction effect retrenchment reconfiguration *

political risk. Model 5 includes both aforementioned interaction effects.

The regression results show both a positive- and negative – but insignificant – effect of growth

asset reconfiguration on the CSR gap of firms. With an estimated coefficient ranging from -0.47 to 0.14.

The hypothesised effect of growth asset reconfiguration on CSR gap was positive (overstating). Since there is no significant correlation in any of the models, hypothesis 1A is rejected. Based on this context, it seems growth asset reconfiguration does not influence a firm’s decision to overstate(- or understate) its CSR behaviour. The results do show a rather high robust standard error in models 3 and 5, indicating that the estimate of the coefficient is not very precise. A common explanation for this is a small sample size, however, this is not the case in this thesis. Multicollinearity issues may partially explain this, as this was not ruled out completely in the VIF tests.

The effects of retrenchment reconfiguration on the CSR gap are insignificant in model 1, 2 and 3. In model 4 and 5, the standard coefficient moves to a significant -0.88 and -0.71, respectively, on the 10% significance level. The results do show a high robust standard error for this variable as well (0.61 and 0.60), indicating that the estimate of the coefficient is not precise. The R2 _{of the models are at 15.69% and 15.66%,} respectively, the values show that the model does show explanatory value. Multicollinearity issues were also ruled out with the VIF tests concerning this variable. The hypothesised effect of retrenchment reconfiguration on CSR gap was negative (understating). As the results show a significant negative effect in these models at the 10% significance level, hypothesis 2A is not rejected. And it can carefully be concluded (due to the high standard errors) that the influence of retrenchment reconfiguration on the CSR gap of firms is significant at the 10% significance level. This conclusion is tested through robustness checks in model 6 and model 7 shown in appendix G.

27 coefficient of -0.02 in all three models where this factor was included. The effect was significant at the 1% significance level for model 3 and model 5, and at the 5% significance level for model 4. This goes against expectations that arose from the literature. Garcia-Sanchez et al. (2016) argue that the stronger the institutional environment, the higher the quality of CSR disclosure becomes. A possible explanation could be that other parts of the institutional landscape that were not covered by the political risk index cancelled out the positive effects of the political landscape. The political risk index incorporates the political landscape and socioeconomic conditions, however, factors such as financial- and economic risk may have been omitted.

The interaction effect of growth asset reconfiguration and political risk, however, does not seem to show a significant effect. The estimated coefficient ranges from 0.02 to 0.03. This results in a rejection of hypothesis 1B that states it would have an attenuating effect.

The interaction effect retrenchment reconfiguration * political risk does show a significant effect at the 10% significance level. With an estimated coefficient of 0.06 in model 4 and model 5. However,

hypothesis 2B states that an attenuating effect would be expected. The hypothesis can therefore be rejected

with the opposite effect being present.

(Model 1) (Model 2) (Model 3) (Model 4) (Model 5) VARIABLES

Growth asset reconfiguration 0.14 -0.21 0.12 -0.47

(0.10) (0.71) (0.10) (0.87)

Retrenchment reconfiguration 0.07 0.07 -0.88+ -0.71+

(0.05) (0.05) (0.61) (0.60)

Growth asset reconfiguration * political risk 0.02 0.03

(0.07) (0.07)

Retrenchment reconfiguration * political risk 0.06+ 0.06+ (0.04) (0.04) Quality of inst. environment (political risk) -0.02* -0.02** -0.02* (0.01) (0.01) (0.01) Firm size (log.) -0.14*** -0.14*** -0.14*** -0.14*** -0.14***

(0.03) (0.03) (0.03) (0.03) (0.03)

Sales growth (log.) -0.00 -0.00 -0.00 -0.00 -0.00

(0.00) (0.00) (0.00) (0.00) (0.00) Financial performance (log.) -0.00 -0.00 -0.00 -0.00 -0.00

(0.02) (0.02) (0.02) (0.02) (0.02) Financial leverage (log.) -0.19 -0.20 -0.18 -0.19 -0.19

(0.13) (0.13) (0.13) (0.13) (0.13) R&D Intensity (log.) 0.21* 0.20* 0.21* 0.20* 0.20* (0.10) (0.10) (0.10) (0.10) (0.10) Manufacturing industry (dummy) -0.46** -0.46** -0.46** -0.46** -0.46**

(0.15) (0.15) (0.15) (0.15) (0.15)

Mining industry (dummy) -0.30 -0.30 -0.30 -0.30 -0.30

(0.23) (0.23) (0.23) (0.23) (0.23) Transportation industry (dummy) 0.02 0.02 0.03 0.02 0.02

(0.17) (0.17) (0.17) (0.17) (0.17)

Wholesale industry (dummy) 0.02 0.02 0.03 0.02 0.02

(0.19) (0.19) (0.19) (0.19) (0.19)

Retail industry (dummy) 0.14 0.15 0.15 0.15 0.15

(0.18) (0.18) (0.18) (0.18) (0.18) Finance industry (dummy) -0.13 -0.12 -0.12 -0.12 -0.12

(0.16) (0.16) (0.16) (0.16) (0.16) Constant 0.19 0.18 1.70* 1.82** 2.04* (0.30) (0.30) (0.79) (0.63) (0.84) Observations 2,440 2,440 2,440 2,440 2,440 Number of firms 305 305 305 305 305 R2 _{0.1547 0.1564 0.1563 0.1569 0.1566} Hausman 0.3016 0.4217 0.3491 0.4031 0.3480

Robust standard errors in parentheses *** p<0.001, ** p<0.01, * p<0.05, + p<0.1

29

4.3. Robustness checks

To test whether these results hold under different circumstances, two additional analyses were performed with lagged variables (see appendix G). The lagged variables also rid the models of endogeneity issues that may have otherwise occurred in the previous models. Model 6 shows the analysis using a one-year (t+1) lag. In this case, the direct effect of political risk on the dependent variable does not change much, however, it does change from being significant at the 1% significance level to the 0.1% level. The effect of R&D intensity becomes insignificant. The effect of sales growth becomes significant at the 5% significance level. What is most notable is that the effects of retrenchment reconfiguration and the interaction effect retrenchment reconfiguration * political risk become insignificant as well. Model 7 shows the analysis using a two-year (t+2) lag. In this case, the same can be said on the direct effect of

political risk and the effect of R&D intensity. Equally notable is that the effects of retrenchment reconfiguration and the interaction effect retrenchment reconfiguration * political risk also become

5. DISCUSSION AND CONCLUSION

This thesis combines insights from dynamic capabilities theory and institutional theory to further the literature on drivers for CSR decoupling in firms. It looks at the direct effects of firm-specific reconfiguration capabilities on CSR decoupling and the moderating effect of the institutional environment that the firm operates in. The research question to be answered in this thesis is phrased as follows: “How

do a firm’s reconfiguration capabilities influence its CSR behaviour (overstating vs. understating)? And how is this relationship influenced by the firm’s institutional environment?” A panel dataset of CSR

behaviour, reconfiguration capabilities and other variables was gathered from 305 U.S.-based firms ranging a period of eight years (2009-2016). This thesis aims to investigate: 1) the effect of a firm’s growth reconfiguration activity on a firm’s CSR gap; 2) the expected attenuating effect of the institutional environment on the aforementioned relationship; 3) the effect of a firm’s retrenchment reconfiguration activity on a firm’s CSR decoupling behaviour; and 4) the expected attenuating effect of the institutional environment on the aforementioned relationship.

The results show no significant differences in a firm’s CSR gap based on its growth reconfiguration activity (H1A). Whilst it does show a significant difference in the CSR gap based on the firm’s retrenchment reconfiguration activity at the 10% significance level (H2A). Acceptance of these results do have a sceptic undertone, as the standard errors were rather high and the results do not hold under the robustness checks. The expected attenuating effect of the institutional environment on the relationship between the firm’s growth reconfiguration activity and its CSR gap shows no significant difference (H1B). Whilst the moderating effect of the institutional environment on the relationship between the firm’s retrenchment reconfiguration and its CSR gap shows a significant result at the 10% significance level. However, this effect was expected to be attenuating, whilst the results show a strengthening relationship. Therefore, H2B was also rejected. The thesis also finds that the direct effect of the institutional environment on the CSR gap was small but significant. The fact that the effect is close to 0 is in line with expectations that arose from the theory to a certain extent, as a well-developed institutional environment was expected to close firms’ CSR gaps. However, whilst theory implies the expected effect would be around 0, the outcome shows a small – but negative - effect, implying the more developed the institutional environment, the more a firm’s tendency to understate its CSR behaviour increases.

31 not capable of reaping the benefits thereof due to the absence of coordination and integration skills and/or learning skills. Additionally, as noted by previous literature, although dynamic capabilities are essential in generating competitive advantage, it must still be combined with an adequate strategy in order to become effective (Shuen et al., 2014). If a firm shows negative effects on these omitted variables, be it the other two legs of the ‘dynamic-capabilities-triad’ or the associated strategy, it may cancel out the initial effects of its reconfiguration activities. This is underlined by the fact that the control variable R&D intensity (used as a proxy for intangibles) shows a significant effect on a firm’s CSR gap. This thesis operationalises a firm’s reconfiguration capabilities by studying its asset reconfiguration, omitting the intangible variables such as intangible assets, capabilities, and the firm’s strategy. Another possible explanation for the fact that the first hypothesis could not be supported, is that when a firm experiences growth, it binds a part of its resource base on aspects that are important in its specific growth strategy that do not support CSR. This would, in turn, cancel out the effect that would otherwise be seen in its CSR behaviour.

This thesis shows that a firm’s tendency to understate its CSR actions is influenced by a firm’s retrenchment reconfiguration activity (hypothesis 2A), this was supported at the 10% significance level. The explanation for which is twofold: 1) substantive CSR actions require investments (time and money) as well as persistence (Hawn & Ioannou, 2016); and 2) the salience of shareholders increases when a firm experiences financial distress (Ullmann, 1985).

This thesis shows no support for hypothesis 1B that states that the quality of the institutional environment attenuates the effect of the relationship between growth reconfiguration and a firm’s tendency to overstate its CSR actions. A possible explanation for which is that the variable that measures the institutional environment (political risk) may omit certain factors that would otherwise cancel out the political risk score and change the overall score for the institutional environment. As the ICRG score usually incorporates financial and economic risk ratings as well (PRS Group, 2019). Extant literature has shown that these factors also play a role in firms’ CSR behaviour (Campbell, 2007).

5.1. Theoretical implications

This thesis addresses the need for more research on the drivers of firms’ CSR behaviour, and in particular that of CSR decoupling. Whilst the institutional theorem offers an external perspective on the subject, researchers have argued that a firm’s internal characteristics change the firm’s response to this environment (Delmas & Burbano, 2011). This thesis addresses said gap and offers an insight on the interplay between the two perspectives.

Firstly, it provides partial support for the notion that firms operating under financial distress (showing forms of retrenchment reconfiguration) tend to understate their CSR behaviour. This is in line with expectations that arose from the literature. Ullmann (1985) suggests that firms may underreport their CSR behaviour to disguise the associated costs. It is expected that this is underlined by the increased salience of shareholder needs when a firm experiences financial downturn (Kim & Lyon, 2015). Furthermore, the empirical findings did not lend support to the posited effect of growth asset reconfiguration on a firm’s CSR gap. As such, it extends the debate about the internal firm-specific drivers on a firms’ decisions regarding CSR decoupling.

Secondly, the findings of this thesis extend the debate about the workings of the firm’s institutional environment and CSR decoupling. The expected attenuating effect of the institutional environment on the relationship between growth asset reconfiguration and a firm’s tendency to overstate its CSR actions cannot be supported by the empirical findings. Whilst the expected attenuating effect of the institutional environment on the relationship between retrenchment reconfiguration and a firm’s tendency to understate its CSR actions cannot be supported by the empirical findings – and even shows the opposite effect. Moreover, the direct effect of political risk on firms’ decoupling behaviour was small but significant. Whilst the fact that it is close to 0 goes hand in hand with expectations, the negative effect was unexpected. Lastly, this thesis finds a negative mean value of the CSR gap, meaning the average firm in the sample understates its CSR behaviour, exhibiting more substantive actions as opposed to symbolic actions. This goes against the notion that usually prevails in literature on overstating CSR behaviour (in particular that of greenwashing literature), that usually states that a firm chooses to overstate its CSR behaviour (read: greenwash) (e.g. Bowen, 2014). This finding is in line with the findings of Hawn and Ioannou (2016) who also found that understating CSR behaviour was prevalent in their sample. It exhibits the need for a closer look at the salient alarming view of greenwashing/overstating.

5.2. Practical implications

33 value, underlining the importance for firms to heed this finding. Next to the importance for firms, external stakeholders and regulatory bodies can also find importance in these findings, as the notion that firms usually overstate their CSR actions is prevalent. One could argue, however, that firms understating their CSR would be a preferrable outcome (as opposed to firms overstating their CSR behaviour). The direct effects from substantive CSR actions seem beneficial. However, it must be noted that as understating also harms a firm’s market value, the long-term effects are, therefore, not yet foreseeable.

Secondly, the results of this thesis imply that when a firm is operating under financial distress (read: practicing retrenchment reconfiguration) the tendency of that firm to understate its CSR behaviour increases. Shareholders pressuring firms for decisions based on profit-oriented programmes should heed this finding. As established before, the link between CSR actions and firm performance is not always clear. Previous research has pointed out that whilst it may be detrimental for firm performance in the short run, it may be beneficial in the long run (Margolis & Walsh, 2003).

Thirdly, dynamic capabilities are considered as replicable routines (Winter, 2003). Firms can benefit from the replicability of dynamic capabilities due to commonalities to develop stable routines for the development of CSR strategies and the implementation thereof. Taking notice of the results from this thesis can aide in developing the right dynamic capabilities to minimise the firm’s CSR gap. As overstating or understating practices are detrimental to a firm’s market value due to a perceived lack of transparency and accountability by stakeholders (Hawn & Ioannou, 2016).

5.3. Limitations and future research

This thesis offers theoretical implications and practical implications. It is important to note, however, that there are still limitations which can give direction for future research on the interplay between internal- and external drivers on CSR decoupling.

Firstly, the factor ‘reconfiguration capabilities’ was operationalised by studying firms’ asset reconfiguration. CSR behaviour, however, also relies heavily on intangible assets. Ideally, one would use the instrumental variables to arrive at a conclusion. This thesis, however, adopts parsimony over completeness to shed an initial light on the hypothesised relationships. The measures of this thesis could, for instance, be improved on in future research by supplementing the objective data on the mere tangible assets with a large-scale survey that also incorporates the intangible firm-specific drivers for CSR decoupling.

incorporate: 1) coordination and integration; 2) learning; and 3) reconfiguration. Ideally, one would use the instrumental variables to arrive at a conclusion. Future research could examine the interplay between the other parts of the dynamic capabilities triad. As several researchers argue that they are interconnected (Wu et al., 2012).

Thirdly, as the study relies on a sample that consists of U.S.-based firms only, generalisability of the results is limited. Future research could take cross-country data into account to research the effects in different environments. The generalisability of the results is also influenced by the fact that the sample only includes large firms, future research could look at the interplay between internal- and external drivers in smaller firms to see whether the results hold in this context as well.

Fourth, as the measure for the institutional environment is mostly focused on the political measures (though, incorporating some socioeconomic conditions), it may have omitted other factors that play a role in the external environment the firm operates in. Future research could include the financial and economic risk ratings, which may incorporate different external pressures and could show different results. Additionally, future research could take industry-level institutionalism into consideration as well.

Fifth, measuring symbolic and substantive CSR behaviour over time is difficult. As symbolic actions require fewer resources to implement (Carlos & Lewis, 2018), whereas substantive actions require investments (time and money) as well as persistence (Eccles et. al., 2014; Hawn & Ioannou, 2016). One can account for this by measuring the variables at different points in time. However, as firms face external pressures that change continuously, it becomes exceedingly difficult to measure the effects.

35

7. REFERENCES

Amit, R., & Schoemaker, P. J. (1993). Strategic assets and organizational rent. Strategic management jou

rnal, 14(1), 33-46.

Amui, L. B. L., Jabbour, C. J. C., de Sousa Jabbour, A. B. L., & Kannan, D. (2017). Sustainability as a dy namic organizational capability: a systematic review and a future agenda toward a sustainable transition. J

ournal of Cleaner Production, 142, 308-322.

Anand, J., & Singh, H. (1997). Asset redeployment, acquisitions and corporate strategy in declining industries. Strategic Management Journal, 18(S1), 99-118.

Aragón-Correa, J. A., & Sharma, S. (2003). A contingent resource-based view of proactive corporate environmental strategy. Academy of management review, 28(1), 71-88.

Bai, X., & Chang, J. (2015). Corporate social responsibility and firm performance: The mediating role of marketing competence and the moderating role of market environment. Asia Pacific Journal of

Management, 32(2), 505-530.

Baltagi, B. (2008). Econometric analysis of panel data. John Wiley & Sons.

Bansal, P. (2005). Evolving sustainably: A longitudinal study of corporate sustainable development. Strategic management journal, 26(3), 197-218.

Barnett, W. P., & Pontikes, E. G. (2008). The Red Queen, success bias, and organizational inertia. Management Science, 54(7), 1237-1251.

Barney, J. (1991). Firm resources and sustained competitive advantage. Journal of management, 17(1), 99-120.

Bourgeois III, L. J. (1981). On the measurement of organizational slack. Academy of Management

review, 6(1), 29-39.

Bowen, F. (2014). After greenwashing: Symbolic corporate environmentalism and society. Cambridge University Press.