Introduction

Corporate social responsibility (CSR) has become a highly institutionalised field of corporate activity in recent years (Shabana et al. 2017). According to common definitions, CSR activities involve social and environmental measures taken by corporations voluntarily and going beyond legal requirements (McWilliams and Siegel 2001). Critics have asked whether CSR was merely symbolic management or corporate ‘greenwashing’ (Bowen and Aragon-Correa 2014; Crilly et al. 2012). Moreover, corporate social irresponsibility (CSiR), i.e. “corporate actions that negatively affect an identifiable social stakeholder’s legitimate claims” (Strike et al. 2006), remains widespread. Governments and civil society actors have made efforts to regulate the CSR activities of firms, resulting in a complex web of CSR governance instruments.

A key public policy approach to CSR focuses on transparency by mandating the disclosure of non-financial information. For example, the European Union (EU) recently adopted a new Directive (2014/95/EU) making non-financial disclosure (NFD) mandatory for the largest European firms (Kinderman 2019); member states had to transpose this into their national legislations. On its own, business self-regulation may result in substantial information asymmetries (Hess 2007), making it impossible for stakeholders to determine whether managers are really acting in their best interest or not. As a result, stakeholders may undervalue some responsible actions and overvalue irresponsible ones (Lopatta et al. 2016).

Governments use regulation encouraging greater transparency to generate confidence and improve the information available to stakeholders about corporate social activities, in the hope that they, in turn, will effectively reward or punish firms through their market activities as investors, consumers, employees, and so on. Generally, NFD legislation specifies the information that corporations must disclose, but nonetheless grants businesses complete discretion regarding the nature of socially ‘responsible’ business practices. For example, it does not usually prescribe specific reporting formats nor does it demand verification by an external auditor. While mandatory non-financial disclosure requirements have become pervasive, surprisingly little research has been carried out on their effects on CSR activities (Shabana et al. 2017; Grewal et al. forthcoming; Ioannou and Serafeim 2017).

This paper thus asks the following research question: how does current policies regarding NFD influence the social components of firms’ CSR activities? We focus on the impact of mandatory NFD on the social dimension of responsible and irresponsible corporate activities (Lin-Hi and Muller 2013).

Following Fransen (2013) arguments about disaggregating different dimensions of CSR, we focus our analysis on one particular dimension (social issues). First, we developed hypotheses about the empirical effect of NFD regulation on firm-level CSR activities. Here, we drew on the concept of government-mandated NFD as a hybrid form of regulation (Steurer 2013). On the one hand, the introduction of mandatory rules regarding transparency may reduce business complacency regarding CSR and irresponsible business practices. On the other hand, these rules play a role in market practices since disclosure is only likely to change corporate actions if stakeholders reward or punish firms’ CSR activities. Consequently, transparency requirements may fail to produce substantial changes or unintentionally result in the adoption of CSR in a box-ticking fashion.

Second, we empirically tested our hypotheses about the impact of NFD regulation on firm-level CSR, using data on stock exchange-listed companies from 24 OECD countries. Our statistical analysis is based on a dataset drawn from the Asset4 environmental, social and governance (ESG) database containing information on corporate responsible and irresponsible activities between 2002 and 2014.

Our paper makes three main contributions. Empirically, our findings show that while NFD policies have increased the average level of CSR activity, it has had varying impacts on firms, depending on whether they previously led or lagged behind in CSR adoption, thereby reducing the variation in CSR activity among firms. In contrast, we found that NFD regulation had not influenced the level of corporate social irresponsibility (CSiR). Theoretically, our paper adds insights about regulatory trade-offs related to the effects of NFD regulation. While mandatory NFD has helped address some of the weaknesses of ‘pure’ business self-regulation, it has lacked powers of regulatory enforcement to prevent irresponsible corporate activities. The main implication of our results for public policy and business ethics is that mandatory NFD will only be effective in conjunction with greater external verification and stakeholder rights.

Finally, the limitations of the study will be discussed, and avenues for future research will be suggested.

Mandatory Non-financial Disclosure: Between Government and Self-Regulation

This section will propose a theoretical framework to understand the potential effects of mandatory NFD on corporate social responsibility. A growing literature has examined the role of government in institutionalizing CSR (Albareda et al. 2007; Knudsen et al. 2015; Scherer and Palazzo 2011; Dentchev et al. 2015). For example, Fox et al. (2002) outlined four roles of government: mandating (e.g. defining standards), facilitating (e.g. giving incentives), partnering with industry, and endorsing (e.g. through special awards). Hard regulation of CSR activities may consist of legislative (e.g. bans on child labour, equality acts) or economic instruments (e.g. taxes) that prescribe and enforce policies, with legal sanctions following their breach. Moreover, governments may influence CSR through soft regulation involving only indirect sanctions, as with labelling schemes or awards (e.g. The National German Sustainability Award or the Agreement on Sustainable Garment and Textile in the Netherlands).

In this context, mandatory non-financial disclosure regulation has emerged as a central instrument of government CSR policy. This type of regulation aims to promote transparency, which reduces information asymmetries between businesses and stakeholders (Hess 2007). Transparency may prompt changes in CSR activities, since corporations can benchmark themselves more easily in relation to competitors and promote discussion about best practices or industry standards (Russo-Spena et al. 2016). More importantly, stakeholders may become more effective in rewarding CSR or imposing sanctions on irresponsible corporate activities (Fernandez-Feijoo et al. 2014; Dhaliwal et al. 2011, 2012; Turban and Greening 1997; Brekke and Nyborg 2008).

Steurer (2013) conceptualised mandatory NFD as a hybrid form of regulation: “Although these disclosure regulations rely on binding and sanctioned laws, I regard them as hybrids because they unfold their steering potential only in combination with civil regulation (mainly market pressure via consumer decisions) and/or (pre-emptive) business self-regulation.”Footnote 1 Mandatory NFD has a binding character and may impose sanctions, such as a fine on board members or the dissolution of firms (e.g. in Denmark) that failed to disclose (Fox et al. 2002). However, firms do retain more or less complete discretion about their actual activities, for NFD rules do not prescribe any specific standards regarding policy adoption or outcomes. For example, the French Grenelle II Act provides no legal sanctions for non-compliance, but firms must provide information about their CSR activities at a stakeholder’s demand. Consequently, firms do not face any government sanctions if they fail to adopt particular CSR-related policies (Reid and Toffel 2009; Hess 2007). This hybrid character has been widely recognised in political science and legal studies (Scheltema 2014; Perritt Jr. 2001).

The effects of mandatory NFD on firm-level CSR activities constitute an interesting issue for organizational research. These effects may involve increased adoption of CSR-related policies and practices, as well as the prevention of, or reduction in ethically irresponsible corporate activities (Lin-Hi and Muller 2013; Strike et al. 2006; Mena et al. 2016; Bowen and Aragon-Correa 2014; Bartley and Egels-Zandén 2016; Xiaoping et al. 2016). While there exists strong theoretical justification in support of the concept of a hybrid form of regulation, empirical research has scarcely developed. Uddin et al. (2016) have called for empirical studies in a comparative perspective (see also Schneiberg and Bartley 2008). Documenting empirical effects is also important for understanding the potential effectiveness of mandatory NFD as a regulatory instrument with regard to ethically critical issues.

Regulatory Trade-Offs: Towards a Theoretical Framework

To understand the potential effects of mandatory NFD, we explored how it combined specific aspects of classic or hard regulation by government with softer aspects of ‘pure’ business self-regulation (Steurer 2013; Hess 2008, p. 450). Our framework contrasts these two ideal types (in the sense of Max Weber’s), comparing them across three dimensions: ambit, content and enforcement. These dimensions suggest possible trade-offs. Government regulation may be more stringent around minimum standards but suffer problems of rigidity as regards content if a ‘one-size-fits-all’ approach is followed. Conversely, business self-regulation may be more flexible in supporting best practices, but tolerate greater complacency towards firms’ strategic non-compliance.

In the case of hard regulation by government, all corporations fall within a mandatory ambit, since legislation typically creates uniform requirements for all firms within a jurisdiction or a legally defined category. In terms of regulatory content, hard regulation tends to be rule-based and to specify particular behaviours, either positively (‘must do’) or negatively (‘must not do’). Governments, as ‘CSR organizers’ (Rasche et al. 2013), create these ‘rules of the game’ through the political process, driven by public interest concerns and, ideally, in support of democratically informed legitimate interests. Hard regulation is enforced directly by state agencies, often through legal or administrative supervision.

In the case of business self-regulation, what falls within the ambit is voluntary, whereby firms control their own engagement with CSR. Rather than stipulating specific rules, the content of self-regulation is centred on broad principles that guide corporate behaviour, often articulated as social norms or perceived ‘best practices’ with an aspirational quality.Footnote 2 These principles may emerge through the decentralised adoption of practices by firms themselves (Matten and Moon 2008) or thanks to the more coordinated efforts of business associations (Bowen 2017).Footnote 3 Self-regulation is ‘enforced’ only in the sense that market actors may reward or sanction CSR through unilateral actions or coordinated campaigns. For example, investors may monitor compliance and either sanction non-compliance through lower share prices or accept non-compliance as being justified under certain circumstances (MacNeil and Li 2006). Here, the costs and benefits of regulatory compliance are internalised by firms and their stakeholders. Whereas the former type of regulation is vulnerable to state failure (e.g. regulatory capture, corruption, lack of capacity), the latter is exposed to market failure.

Both harder regulation by government and softer self-regulation by business imply potential trade-offs, since they have different objectives. We would expect classical government regulation to be more stringent in the sense of more specific content, and involving public enforcement—thus with a strong potential for promoting minimum standards around CSR. By establishing rules clearly about what corporations must not do and providing state enforcement, government regulation may be effective in preventing irresponsible activities (Lin-Hi and Muller 2013; Locke et al. 2013).

However, greater stringency may come at the cost of flexibility by promoting a ‘one-size-fits-all’ approach. First, rules may be relatively unresponsive to the specific business circumstances of firms and the local constellation of stakeholder interests. Second, the political process may be slow to respond to changes in the business environment, and may also be subjected to strong veto points and lobbying (Abbott and Snidal 2000). Consequently, government regulation is often viewed as being rigid and imposing requirements that lack relevance for particular firms—thereby costly for business.

In contrast, we would expect self-regulation to offer greater flexibility across these three dimensions—thus with a strong potential for fostering best practices. Its voluntary nature and the use of broad principles provide substantial scope for adjusting rules to the specific needs of individual companies (Gregory 2002). This flexibility may help focus firms on the ‘win–win’ aspects of CSR, which create tangible benefits for business through improved relations with stakeholders (Dhaliwal et al. 2011, 2014). Likewise, self-regulation gives leeway to experiment with the development and diffusion of best practices. The ‘private’ nature of self-regulation also means that CSR practices (or even codes and standards) can be quickly reformulated by business itself in response to changing circumstances.

However, greater flexibility may imply greater scope for business complacency. Since firms may fail to adopt CSR owing to market failures or weak stakeholder pressure, business self-regulation may fail to uphold minimum standards (Adams 2004). Likewise, the content may lack stringency in the sense of being ‘softer’ towards business interests, since it lacks the democratic legitimacy gained through the political process or institutionalised stakeholder involvement (Shamir 2008). Table 1 provides a summary overview of these trade-offs.

Table 1 Ideal types of hard government regulation and business self-regulation
Full size table

Nevertheless, mandatory NFD regulation may potentially combine the advantages of business self-regulation and government regulation, thereby lessening regulatory trade-offs. Indeed, mandatory NFD essentially uses instruments of government regulation to enhance transparency around corporate social responsibility, thereby making self-regulation more effective (Lepoutre et al. 2007; Steurer 2013; Gond et al. 2011). By adopting mandatory NFD, the state uses government rules in a market-enabling fashion. The objective here is to reduce information asymmetries around CSR (Cui et al. 2018) and create greater transparency among firms through more widespread, standardised reporting formats (Slager et al. 2012). Transparency may thereby improve the socio-material conditions for the mobilisation of stakeholders (Gond and Nyberg 2016) and make their engagement in enforcing minimum standards around CSR more effective. Conversely, state-mandated disclosure may provide top management with external legitimacy to act upon non-financial criteria in their decision-making processes (Avetisyan and Ferrary 2013; Giamporcaro and Gond 2016; Igalens and Gond 2005; Neumann et al. 2011). Here, the state plays a more catalytic than coercive role (Reinecke and Ansari 2016).

In theory, this policy approach will also maintain flexibility for firms in terms of the specific content that they adopt. Mandatory NFD does not prescribe any specific CSR activities (Antal and Sobczak 2007). Nor does it institutionalise any stakeholder rights that would create enforceable claims around specific social responsibilities. Nonetheless, neo-institutional theory suggests that transparency may lead to mimetic forms of isomorphism, as firms adopt practices similar to those of their competitors (DiMaggio and Powell 1983), leading to more homogeneous and, potentially, rigid forms of CSR (Russo-Spena et al. 2016; Chatterji et al. 2016).

In sum, government regulation and business self-regulation may interact empirically in a variety of complex ways. In this study, we focused on the analytical features of these two ideal types to better conceptualise the hybrid character of NFD regulation. Here, governments regulate disclosure of CSR activities that also remain governed by business self-regulation (Hess 2014; Parker 2007).

How Does Mandatory Disclosure Shape CSR Activities?

Next, we will develop hypotheses related to the effects of country-level NFD regulation on firm-level CSR activities. We will draw on the regulatory trade-offs framework presented above to explore whether mandatory NFD may result in greater stringency around minimum CSR standards (similar to harder kinds of government regulation), and whether or not this has any consequences for the flexibility of CSR activities (normally associated with self-regulation).

First, concerning the ambit of regulation, mandatory NFD is likely to increase CSR adoption by firms. Several studies have suggested that firms substantially increased their CSR activities in response to NFD regulation (Young and Marais 2012; Albertini 2014; Chelli et al. 2014). Since firms will have to disclose certain kinds of activities or will compare their disclosed activities with peer firms, CSR adoption will tend to increase the scope of CSR activities across more issues (Chen et al. 2018). Therefore, we posit a baseline positive influence of NFD regulation on CSR:

H1a

NFD regulation will lead to an increase in the average level of CSR activity.

The above hypothesis does not have any clear implications regarding regulatory trade-offs involving stringency. Thus, we argue that disclosure may increase stringency by having varying effects on firms depending on their prior level of CSR activity. Faced with NFD regulation, firms are unlikely to maintain very low levels of engagement with CSR, since increased transparency will bring them under greater scrutiny relative to competitor firms (Brunner and Ostermaier 2017). By spotlighting the gap in their activities, transparency may induce behavioural changes, in particular among firms with low levels of CSR relative to their peers (Chatterji and Toffel 2010). On the other hand, firms with higher CSR levels will be less likely to modify their CSR engagement as a result of mandatory disclosure (Giannarakis et al. 2017). Thus, NFD regulation may make it difficult for firms with a very low sense of responsibility to ignore CSR issues as they did under self-regulation, thereby increasing stringency or reducing complacency around CSR. Hence, we posit that:

H1b

NFD regulation will lead to a larger increase in the level of CSR activity among firms in the bottom 20% as regards CSR activity than among those in the top 20% (increasing stringency/reducing complacency).

NFD regulation is also likely to increase the stringency of regulatory enforcement. To evaluate this dimension, we focused on a widely acknowledged minimum standard for CSR, namely the reduction in irresponsible activities (Lin-Hi and Muller 2013; Armstrong and Green 2013). While corporations should seek to avoid controversial activities or at least buffer their negative social consequences, market sanctions for irresponsible behaviour depend on stakeholder mobilisation, are prone to market failures, and are often surprisingly weak (Frynas 2010; Jackson et al. 2014).

Increased transparency may lead to less irresponsibility for at least two reasons. First, disclosure requirements may raise awareness and scrutiny of socially relevant issues within the firm, and increase efforts to prevent or end controversial activities. Second, outside the firm public awareness and sensitivity to socially irresponsible practices may be raised. By having a greater amount of information and degree of comparability, market intermediaries, such as NGOs, may become more effective in compiling CSR-related information and may scrutinise the behaviour of companies to a greater extent (Fernandez-Feijoo et al. 2014). The visibility of controversial or outright negative corporate actions is a necessary condition for stakeholders to threaten effective sanctions on firms engaging in such behaviour (Surroca et al. 2013). As a consequence of increased stringency, we argue that NFD regulation is likely to lower the level of irresponsible activities, thereby reducing complacency among firms within regimes of pure business self-regulation. Hence, we posit that:

H2

NFD regulation will lead to fewer irresponsible activities by companies (increasing stringency/reducing complacency).

So far, we have considered that the modest increase in stringency associated with NFD regulation may help reduce the downside of self-regulation, i.e. high complacency of firms. However, our conceptualisation of trade-offs implies that stronger government regulation might involve a corresponding reduction in flexibility and an increase in rigidity. On the one hand, the hybrid character of NFD regulation relies largely on market-driven enforcement in relation to CSR adoption, thus we might not expect any substantial reduction in flexibility—thereby preserving the benefits associated with pure self-regulation and creating a ‘best of both worlds’. On the other hand, even a very modest dose of government regulation may be expected to reduce the flexibility of firms’ engagement with CSR, owing to coercive or mimetic institutional pressures.

Regarding the latter point, as CSR activities become more transparent and thereby may be compared among peers, firms may respond by imitating specific types of activities (Russo-Spena et al. 2016; Chatterji et al. 2016). If firms are uncertain about what is expected of them, a typical response will be ‘mimetic isomorphism’, where firms seek legitimacy by becoming more similar to each other in their activity profile (DiMaggio and Powell 1983). But imitation also implies a greater degree of rigidity. Rather than voluntarily adopting those CSR activities that are most strategically salient to their situation, some firms are more likely to adopt a standard package of ‘content’ around CSR, often on the advice of market intermediaries, such as CSR consultants or audit firms that promote industry-standard solutions (Fortanier et al. 2011). As a result, standardisation of CSR may be associated with lower flexibility and a box-ticking approach that avoids genuine stakeholder engagement (Bondy et al. 2008). Hence, we posit that:

H3a

NFD regulation will be associated with a greater similarity of CSR activities among firms within the same country (reducing flexibility/increasing rigidity)

Along similar lines, NFD regulation may lead to a greater standardisation of CSR activities and thereby narrow the gap between ‘best practices’ and the activities of the average firm (Shabana et al. 2017). While a higher average level of CSR engagement may be normatively desirable, paradoxically, greater institutionalisation of CSR may erode the business case for the voluntary adoption of CSR. For the business case implies the freedom to engage selectively with CSR activities that help to differentiate a firm’s CSR profile from competing firms, so that stakeholders may reward these activities through greater loyalty and longer-term investment (Lohmeyer 2017; Thijssens et al. 2015). However, transparency may narrow the gap between best and average practices, making it harder for firms to adaptably differentiate their CSR profiles, and lead to a levelling off in CSR engagement (Holder-Webb et al. 2009). Hence, we posit that:

H3b

NFD regulation will lead to a narrowing gap between the CSR activities of firms with a previously high CSR activity level and the level of the average firm in the same country (reducing flexibility/increasing rigidity).

Method and Data

To understand how NFD rules influence the CSR activities of firms, we conducted statistical analyses that compared firms in three ‘early adopter’ countries (France, the UK and Denmark) with other OECD countries, as well as taking into account the changes within these countries over time.

France established the first standardised set of social reporting indicators in 1977 by requiring all companies with 300 employees or more to report on 130 indicators for employment-related activities (Antal and Sobczak 2007; Sobczak and Coelho Martins 2010). The 2001 Law on New Economic Regulations (NRE) (Loi Nouvelle Régulation Economique) later mandated all publicly listed French companies to report non-financial information related to their social and environmental impacts. The Grenelle I Act (3 August 2009) and the Grenelle II Act (12 July 2010) widened the ambit of companies subjected to CSR reporting, added more topics for disclosure using a ‘comply or explain’ approach, and subjected NFD to verification by an accredited independent third party.

The UK followed suit by requiring a ‘business review’ within the 2006 Companies Act, which compelled company directors to report on the impact of operations on employees, the community, and the environment. A Regulation passed in 2013 required a Strategic Report containing specific disclosures about human rights, community issues, gender diversity, and greenhouse gas emissions

Soon after, in 2008 Denmark mandated NFD for companies. A key government objective was to improve the international branding and competitiveness of Danish firms through the promotion of international CSR standards. In 2012, provisions were extended to include human rights and climate impact and, in 2015, anti-corruption and bribery, social and employee-related aspects, and diversity on the board of directors were added. The new 2015 issues reflected the requirements of the 2013 European Union’s NFD Directive. Tables 8, 9, and 10 in Appendix present a chronological overview of NFD legislation in France, the UK and Denmark.

These three countries were first among the OECD to introduce NFD regulation with an explicit focus on the social dimension. In this context, mandatory disclosure is defined as regulation concerning the disclosure of social activities that applies to a broad segment of privately owned stock exchange-listed corporations. We excluded cases where regulation applied only to state-owned companies, as in Sweden since 2007. We also excluded disclosure regulation limited to hazardous environmental risks, as found in the Netherlands or the US. Spain is an ambiguous case, since its 2011 government regulation only recommends (rather than mandates) disclosure (González and Vílchez 2015). Similar requirements were also introduced by Norway in 2013, but this only covers 1 year during our observation period.

In the EU, Directive 2014/95/EU on non-financial disclosure required all member states to transpose the Directive into their national legislation before December 2016. This development will change the picture of CSR disclosure and makes our analysis timely

Our statistical analysis makes use of the Thomson Reuters Asset4 ESG database, which measures firm-level corporate responsibility activities. This database has already been validated in the CSR literature (Eccles et al. 2014; Cheng et al. 2014; Hawn and Ioannou 2016; Rathert 2016). Recently, it has increasingly been used in studies published in the Journal of Business Ethics (Al-Shaer and Zaman 2017; Aouadi and Marsat 2018; Benlemlih et al. 2018). Asset4 functions as a financial intermediary providing investment information related to environmental, social, and governance (ESG) issues; it systematically collects information from company reports, company websites, and other sources, such as newspapers and nongovernmental organizations. In addition, firm-level financial data were obtained through Thomson Reuters DataStream.

Our sample of firms consisted of stock exchange-listed corporations during the period 2002–2014. This covers the time before and after the implementation of non-financial disclosure rules in Denmark and the UK. However, the very early introduction of policies in France during the 1980s predates the existence of any source of comparable international CSR data.

We reported our results for 24 OECD countries that had a sample size of 10 firms or more per year during the period of observation. Moreover, and to make the countries more comparable, we restricted our sample to companies responsible for 90% of total market capitalisation per year in countries with more than 100 companies, namely Australia, Canada, Japan, the UK, and the US. Table 2 shows the countries and average annual number of firms included in our analysis. The annual number of companies used in the analysis varied slightly from year to year owing to missing values. In total, our sample contains 19,709 firm-year observations.

Table 2 Average annual number of companies per country
Full size table

For our analysis, and as suggested by Fransen (2013), we examined CSR activities in a disaggregated fashion, focusing solely on those related to social issues. We devised a CSR measurement based on 36 items published by Asset4 that measure the presence or absence of policies related to social issues, as well as whether or not companies have taken specific action to implement these policies.

We included corporate activities across seven sub-dimensions of the social pillar: product responsibility, human rights, community, employment quality, health and safety, training and development, and diversity (see Table 3). The CSR indicator was generated by combining the discrete data items in each sub-dimension and calculating a simple average of ‘yes’ activities. While each sub-dimension was measured by a different number of indicators, we gave each dimension equal weight in the aggregate score and captured the potential variety of CSR activities.

Table 3 Composition of the CSR Index and CSiR Index, based on the Thomson Reuters Asset4 ESG database
Full size table

Moreover, we created a second index for CSiR based on a coding of corporate controversies and scandals, as well as estimated regulatory penalties and fines across a similar range of issues (see Table 3). For example, irresponsible activities related to health and safety include the total number of injuries and fatalities or if the company is under the spotlight of the media because of a controversy linked to workforce health and safety. The CSiR index combines discrete ‘yes’ and ‘no’ items with continuous variables that were ranked between 1 and 0 relative to the whole Asset4 database between 2002 and 2014. Whereas CSR data are largely based on self-reporting by companies, the Asset4 reports companies’ association with irresponsible activities and negative events as reflected in global media, NGO reports or reported regulatory infringements and lawsuits. As such the CSiR measure includes both activities conflicting with law as well as activities where public expressions of concern have been documented.

Consequently, nearly 50% of firms scored zero as regards CSiR, reflecting a potential underreporting bias. The resulting variables for CSR and CSiR ranged from 0 to 100, expressing the percentage of measured activities in relation to the possible maximum. Missing data was excluded from the analysis.

We tested H1a and H1b using the CSR index, and H2 using the CSiR index. We also calculated one additional dependent variable. To test H3a and H3b, we created a measure of variation in the CSR activities between firms. A ratio of the spread of CSR activities reflects the distance between each firm and the median CSR activity level in each country in each year.

The main independent variable was a country-level indicator of NFD policies developed by using information published by the Hauser Institute for Civil Society. The institute has collected cross-national information on NFD regulation going back to the 1970s. We used this to build a binary NFD variable indicating whether a country had implemented mandatory non-financial disclosure requirements (1) or not (0). As mentioned earlier, we focused explicitly on the impact of NFD legislation on the social activities of a wide range of stock exchange-listed companies.

In our analyses, we included several firm-level control variables (Jackson and Apostolakou 2010; Koos 2012; Padgett and Galan 2010). First, we expected that CSR activities would increase with firm size, measured here by the logged number of employees. Second, financial performance might positively influence CSR through the availability of resources, so we measured financial performance according to return on assets (pre-tax income as a percentage of total assets, winsorized values in the 1st and 99th percentile). We also included a ratio of total debt to total assets to capture financial constraints on CSR activities owing to indebtedness. Third, CSR might also be influenced by the degree of innovation, in view of greater future business opportunities or the need to justify new products in the eyes of customers. Thus, we included the level of R&D spending as a percentage of total assets.

To acknowledge possible reverse causality of firm characteristics, these variables were lagged by 1 year. In general, the coefficients of the control variables were consistent with previous studies. Moreover, we included industry fixed effects based on a super-sector industry classification benchmark (ICB) to control for sectoral differences that affect CSR activities and year dummies to account for the observed trend towards increasing levels of CSR. Finally, we added country dummies to control for wider institutional differences between countries. All firm-level control variables were obtained from financial data collected by Thomson Reuters DataStream.

A key consideration in our study was whether our CSR index would be suitable to capture the effects of NFD regulation. Broad requirements for disclosure do not necessarily lead to the adoption of specific policies or implementation activities. However, a company might previously have adopted a CSR-related policy, but only begin to provide explicit information to the public about this after disclosure has become mandatory. In this sense, the measurement would capture the shift from an implicit to a more explicit form of CSR, but not necessarily a change in corporate activities. To control for this type of confounding, we controlled for firm-level adoption of Global Reporting Initiative (GRI) standards as an indicator of the quality of reporting. Consequently, we argue that it is unrealistic to reduce differences in our measured CSR index to differences in reporting, but we have interpreted these to reflect changes in actual CSR activities.

To model the impact of NFD regulation between the years 2002 and 2014, we adopted different general linear squares (GLS) random effects models; these enabled us to test our hypotheses using robust standard errors clustered by company. Random effects estimations are appropriate since our key explanatory variable for NFD regulation does not vary much over time (Wooldridge 2010). All estimations were carried out using STATA 12.1 and 15.0.

The Effects of Mandatory Disclosure Regulation on Firm-Level CSR Activities

This section presents the results of our statistical analysis of how NFD policies disclosure regulation has influenced CSR activities carried out by firms around social issues. We explored the research question by examining our five hypotheses: H1a to H3b.

Table 4 shows the descriptive results for all variables used in our models. The low correlations between our firm-level control variables suggest that all may be included in the estimation models. The correlations between NFD regulation and our various dependent variables hint to different types of relationships; these will be further examined in the following analyses. Moreover, CSiR and CSR variations were correlated with Year, suggesting a strong time trend in the data.

Table 4 Descriptive results and correlations
Full size table

To test our five hypotheses, we applied several different regression models. First, we explored hypothesis H1a about the effect of NFD regulation on CSR activities in general, assuming that NFD regulation would lead to an increase in the average level of CSR activity. Table 5 reports the results of a random effects model. Whereby Model 1 displayed the effects of the control variables, Model 2 supported H1a with a significant positive average effect of NFD regulation on firm-level CSR activities. The results reflect the positive effect of NFD in countries with an NFD policy in comparison to countries without one. This supports the political reasoning in countries such as Denmark: it was intended to use NFD as a benchmark for Danish companies and their CSR behaviour worldwide.

Table 5 Results of random effects model of CSR activities at firm level, all firms as well as top and bottom 20% of firms
Full size table

Moreover, the results also revealed an interesting relationship between different forms of governance. NFD regulation increased firm activities by an average of seven points on a scale between 0 (zero activities) and 100 (full range of activities). In comparison, the firm-level adoption of GRI Sustainability Reporting Standards, which has commonly been known as a measurement of self-regulation in the field of CSR for many years (Levy et al. 2010), influenced CSR activities twice as much—by twelve points. There is an interesting discrepancy in the effect sizes of a purely ‘voluntary’ adoption of a reporting standard compared with mandatory disclosure regulation.

Addressing the question of trade-offs more closely, we then dealt with H1b by examining whether NFD regulation would lead to the largest increase in CSR activities among firms with a previously low level of CSR activity in comparison to those with a high level. Table 5 presents the results of Models 3 and 4, which estimated the impact of NFD regulation separately for those firms in the top and bottom twenty per cent within each country. We assumed that NFD regulation would lead to a greater increase in CSR among firms in the bottom twenty per cent group. The analysis shows that NFD regulation had a significant positive effect on both groups, but the effect size was indeed almost twice as large for the bottom twenty as for the top twenty. This illustrates the fact that NFD encouraged firms with low CSR to increase their activities, thus suggesting an effect comparable to ‘minimum standards’ in reducing the complacency of firms placed under pure self-regulation and leading to a greater degree of stringency.

Hypothesis H2 concerned the link between NFD regulation and CSiR, and assumed that NFD regulation would lead to a decrease in irresponsible activities by companies. In general, and as reported in Table 4 (correlation between CSiR and Year), CSiR increased across the overall OECD sample over time. Table 6 shows the results of our random effects modelling of the effect of NFD on CSiR activities. Model 1 shows the effects of control variables, Model 2 reports the impact of NFD regulation while controlling for size, performance, R&D, industry, and the adoption of GRI standards. Our analysis finds no support for a significant relationship between NFD regulation and firm-level CSiR activities. Despite the assumption that NFD regulation would discourage irresponsible activities, we could not find any support for this or for the assumed enforcement and effectiveness of market forces.

Table 6 Results of random effects model of CSiR activities at firm level, overall sample as well as UK and Denmark (shown separately)
Full size table

Moreover, looking at the UK and Danish sub-sample separately—to compare these countries before and after the implementation of NFD legislationFootnote 4—the results produced by Models 3 and 4 suggest that NFD regulation was associated with significantly higher levels of CSiR in these countries, although the model for Denmark is less consistent because it included fewer cases and some sectors were omitted. Overall, Table 6 indicates that disclosure requirements had no direct impact on increased efforts to prevent or end controversial activities. Nor did they have any indirect effects, i.e. through information provision affecting the awareness and sensitivity of market intermediaries and prompting them to more effectively monitor and sanction firms.

Hypothesis H3a suggested that NFD regulation would be associated with greater similarity of CSR activities among firms within the same country. As reported in Table 4 (correlation between Variation in CSR and Year), national variation among firms decreased across the overall OECD sample over time. Low or declining variation suggests a trend towards isomorphism and growing institutionalisation of CSR activities. Table 7 shows the results of our statistical estimation using a random effects model. The results of Model 2 show that NFD regulation had a significant negative effect even after controlling for firm-level characteristics.

Table 7 Results of random effects model concerning the ratio of firm-level CSR activity to the median firm, for the whole sample as well as for the top 20%
Full size table

Finally, H3b posited that NFD regulation would lead to a decrease in the level of CSR activity among firms with a previously high level. Table 7 presents the results of a random effects model estimating the influence of NFD regulation on the relative advantage of top companies in relation to the median company in each year. To model the relative advantage of top companies, we made calculations only for a sub-sample: the top twenty per cent in each country in each year. Model 3 shows the effects of control variables, Model 4 reports the impact of NFD regulation while controlling for size, performance, R&D, industry, and the adoption of GRI standards. The analysis shows that NFD regulation has had a significant negative effect on differentiation over time. If a country adopted NFD regulation, this did decrease differentiation among well performing firms over time.

Confirming the business case for the ‘voluntary’ adoption of CSR, the result indicates a closing of the gap between ‘best practices’ around CSR and the activities of the average firm. Thus, a narrowing of the gap between best and average practices entails diminished differentiation and thus a reduction in competitive advantage. Opponents of ‘hard’ regulation have argued that regulated CSR might make it harder for firms to differentiate their CSR profiles and would lead to a levelling off of CSR engagement and declining flexibility; in fact, we found that this trend was equally valid for ‘soft’ NFD regulation. The results, therefore, do not support the advantage claimed by NFD regulation in comparison to hard regulation.

To test the robustness of our results, we used a lower threshold for categorising NFD requirements based on other studies; these cite the Netherlands (Dentchev et al. 2015), the US, Sweden or Norway as positive instances of NFD policy. The Netherlands and the US have adopted NFD rules for environmental issues that fall outside the scope of this study. However, Sweden and Norway have imposed the NFD regulation of social activities on state-owned companies. Although this policy does not cover all listed companies, its application to state-owned companies may produce mimetic effects on privately owned corporations. To take mimesis into account, calculations in our random effects panel model also took the NFD regulation in Norway and Sweden into consideration. The results remained robust.

Discussion

As CSR becomes institutionalised around the world, a major debate has emerged about the most effective way to regulate CSR activities and the role of government (Kinderman 2012; Knudsen and Moon 2017). This paper contributes to the growing literature on public regulation of CSR by examining its impact on firms. In particular, we focus on mandatory NFD as a central instrument of government policy aiming to increase the transparency of firms’ CSR activities. We conceptualise the effects of mandatory NFD in terms of the stringency vs complacency of firm-level CSR activities, as well as in terms of the flexibility vs rigidity of these practices. Our paper makes three main contributions.

First, our paper makes an empirical contribution by examining the effects of NFD regulation on firm-level CSR activities across OECD countries. We find that mandatory NFD has led to an increase in these activities and that it has had the largest impact on those firms with previously low levels of CSR (consistent with H1a and H1b). However, we find no evidence that mandatory NFD leads to a corresponding decrease in irresponsible actions, i.e. CSiR (no support for H2).

Our focus on the social dimension of CSR complements previous research by others that showed that greater disclosure might be linked to a more negative environmental performance (Aragón-Correa et al. 2016; see also Clarkson et al. 2011). Our paper thus suggests that disclosure could possibly be used as a strategy to legitimate irresponsible activities. Finally, we find that an NFD policy leads to more homogeneous CSR activities within a country and reduces the gap between ‘best practice’ firms and ‘typical’ firms (consistent with H3a and H3b).

Second, our paper makes a theoretical contribution to the literature on the role of government in CSR (Gond et al. 2011) by conceptualising its impact at firm level. To this end, we devised a framework for understanding how government regulation and business self-regulation might shape CSR. However, the hybrid character of mandatory NFD suggests that it can function as coercive state regulation, but can also offer market-based regulation operating through interactions between firms and their stakeholders. Specifically, we examined whether mandatory NFD might help to increase the stringency of CSR practices (as with government regulation) while maintaining the high degree of flexibility associated with business self-regulation.

However, our empirical results offer only partial support for the idea that NFD regulation may help overcome such trade-offs. On the basis of our evidence, it is unclear whether or not mandatory NFD will increase CSR stringency. While we find an increase in CSR adoption by ‘laggard’ firms—which points to greater stringency—we also find that there is no corresponding reduction in CSiR (Lin-Hi and Muller 2013).

Our research has, therefore, clear implications for the NFD Directive 2014/95/EU, which national governments have recently transposed into law across the EU. Our findings suggest that this Directive will lead to increased CSR activity without necessarily reducing levels of CSiR. Given that preventing irresponsible activities is often seen as a bottom line for CSR (Lin-Hi and Muller 2013), the contribution of mandatory NFD to establishing minimum standards seems doubtful. However, we do not rule out the possibility that the Directive’s provisions regarding human rights due diligence and mitigating adverse impacts could have a real impact and help to reduce CSiR.

Our results also show declining variance among CSR practices, which points to growing rigidity as a result of mandatory NFD. This finding implies that even ‘soft’ forms of government regulation may trigger other types of isomorphic processes, as suggested by neo-institutional theory. Future research might examine how transparency changes processes of commensuration (Espeland and Stevens 1998) around CSR and how different actors use information and benchmarking in mobilising their CSR-related activities (Beunza and Ferraro 2018).

Third, our paper also contributes to the literature on public policy and business ethics by exploring implications for the effectiveness of mandatory NFD, as explained below. To this end, we argue that classical government regulation and ‘pure’ business self-regulation generate trade-offs related to ambit, content and enforcement. Using the concept developed by Steurer (2013), we interpret mandatory NFD as a hybrid form of regulation with a potential for reducing trade-offs by improving the quality of information to stakeholders—thereby strengthening their capacity to enforce their interests.

However, our empirical results show some important limitations of this interpretation and suggest the need to problematise links between information disclosure and CSR more generally. Transparency in the form of information disclosure may be a necessary condition for greater accountability, but is certainly not a sufficient condition (Roberts 2009; Michelon et al. 2015). Our empirical results suggest that during the period covered by this paper (until 2014), mandatory NFD was unable to improve effectiveness by reducing regulatory trade-offs.

We conclude that NFD in its current form comes across as a somewhat ‘toothless’ tiger. Non-financial disclosure by companies highlights only the positive aspects of CSR but does not address the impacts of potentially negative behaviour (CSiR). A recent Danish study of 279 large Danish firms supports this finding. The study found that while 81% of these firms reported on environmental, social, human rights, and anti-corruption activities, only 14% reported about significant negative possible impacts of the firm in those four CSR areas (Carve Consulting 2016).

Furthermore, the NFD approach does not encourage firms to identify new issues that are gaining prominence in the public discourse regarding what it means to be socially irresponsible, such as the tax transparency of Starbucks or privacy issues at Facebook (Economist 2018). One way for companies to address emerging, potentially damaging social issues is for them to engage with a diverse range of key stakeholders.

Policy Implications

One implication of our study is that policy-makers need to consider whether existing NFD requirements will lead to the provision of information of sufficient quality. To enhance quality, policy makers could require companies to undertake external assurance of the information disclosed. The importance of rigorous independent verification processes, designed to reassure stakeholders about the credibility, completeness and materiality of the social information reported, has been stressed in various previous studies—including works on political CSR (Adams 2004; Wickert 2016; Zorio et al. 2013). Other recommendations to corporations as regards improving the quality of CSR reporting include accurate disclosure, full disclosure and enhanced quality of diversity of assurance provider (Sethi et al. 2017). This is important since the political debate around mandatory NFD regulation in the pioneering cases of France, the UK and Denmark shows a clear rejection of stronger government regulation of CSR in favour of a modified approach to self-regulation. A key premise in these countries is that stakeholder issues should be handled through market-based governance; yet governments did acknowledge the need to improve the prevalence and quality of the information flow to stakeholders, particularly investors, to help them better assess the opportunities and risks linked to their future investments.

A second implication is that NFD regulation will only increase stringency in combination with strong stakeholder rights. Stakeholders need to have enough resources to evaluate the information disclosed and the power to sanction firms that fail to meet their expectations. For instance, stakeholders can be given a voice in corporate decision-making through co-determination at board level or by improving shareholder engagement.

In sum, our results suggest the need to problematise the widespread assumption that firm-level CSR is separate from, and may effectively substitute for government regulation—rather, they are closely interdependent in ways not yet sufficiently understood (see also Wickert 2016; Mäkinen and Kourula 2012).

A third implication is the need to better understand the role of flexibility. Despite the fact that NFD regulation leaves much at the discretion of firms, our empirical results show a surprising degree of conformity when it comes to CSR adoption. While self-regulation can, in principle, be expected to yield flexibility and promote a diversity of firm-level practices, our results suggest that NFD regulation may have isomorphic effects on firms. This finding is consistent with studies suggesting that CSR disclosure practices have become increasingly homogenous and reveal a box-ticking mentality (Pedersen et al. 2013).

In considering these implications, we note that here effectiveness is viewed in connection with minimum standards and flexibility, as outlined in Table 1. Hence, we assume that effective regulation will address both of these objectives to some extent. Ultimately, this is an ethical question for business and public policy as to exactly where the balance should lie.

With regard to business ethics, our findings suggest the following. Prima facie, the fact that mandatory NFD increases firms’ average level of CSR engagement is normatively desirable. However, the fact that it does not decrease CSiR muddies the waters. If we grant that CSR promotion is less beneficial than CSiR reduction (Clark and Grantham 2012), then mandatory NFD has ambivalent ethical implications. Hence reducing CSiR will be a key priority in improving the ethical benefits of non-financial disclosure.

Nevertheless, we do not exclude the possibility that the EU’s NFD Directive 2014/95/EU (CSR Europe & Global Reporting Initiative 2017) will help to promote stringency, reduce CSiR, and enhance the ethical benefits of non-financial disclosure and corporate social responsibility.

Limitations and Future Research

Whilst we consider our methodological approach and empirical data to be particularly well-suited to this study’s research objectives, their limitations should be acknowledged. First, until recently, relatively few advanced industrialised countries had adopted mandatory NFD although, increasingly, developing and emerging market economies, such as Kenya and China, are following their example. Our statistical model combined the influence of regulatory changes within countries and the presence/absence of NFD regulation between countries. We relied largely on a yes/no indicator rather than a finer-grained measurement of specific NFD requirements. For example, France has far more specific requirements than the UK and mandates a greater role of third-party audits. Henriques (2010) found that only half of FTSE 100 companies disclosed detailed quantitative information and argues that the poor quality of reporting suggests that “the Business Review does not appear to be serving the purpose for which it was intended.” But even in France, NFD guidelines and standards have been criticised as being too vague to support quality disclosure (Dhooge 2004).

As more countries adopt mandatory NFD regulation within the European Union, we argue that there is scope for more detailed work on its effects on CSR in the EU; we are mindful of how much more effort will be necessary to gain a fuller understanding. Additional, finer-grained research could be a natural extension of our study, helping to explore the similarities and differences in other EU countries.

Second, despite the high credibility enjoyed by Thomson Reuters’ Asset4 ESG data among investors and scholars, the criteria chosen to measure the social performance of firms may not be optimal. Thus, available data sources for CSR and CSiR research should be handled with caution. Our CSR score combines indicators about the presence or absence of activities. If NFD regulation leads to the adoption of more policies and related implementation efforts, it does not necessarily mean that outcomes will improve.

Indeed, much literature suggests that CSR is little more than symbolic management (Perez-Batres et al. 2012). CSiR data reflects reporting by newspapers and NGOs on critical or irresponsible events. It may be that increased disclosure and transparency will simply make these more visible, thereby changing perceptions of irresponsibility rather than underlying business practices. Future studies may address this potential recursive effect and explore long-term dynamics.

However, even if we interpret CSiR cautiously as reflecting only stakeholders’ perceptions, the fact remains that disclosure does not reduce such negative perceptions even over long periods of time (such as that in our study). If companies were free to choose their own reporting methods, they would be able to highlight positive aspects of their activities and effectively gloss over more negative facts.