Despite a steady increase in the number of regulations across the world (KPMG et al. 2016), corporate sustainability reporting remained a largely discretionary activity in recent years. The main reason is that most regulation applied to only small fractions of companies (e.g., only the largest stock-listed companies, or companies belonging to one industry in a specific country), and prescribed only the provision of sustainability information, not what information must be provided (KPMG et al. 2016). Hence, one of the central questions for researchers in the field has been why companies engage in this largely voluntarily activity.

Accounting researchers typically consider economic arguments for voluntary disclosure (Healy and Palepu 2001). In the context of sustainability reporting, this implies that shareholders may value a company’s efforts to become more sustainable, but ̶ as outsiders ̶ lack the information to differentiate between below-average and above-average sustainability performers. Therefore, companies with above-average performance have an incentive to voluntarily engage in sustainability disclosure in order to signal this (Verrecchia 2001) and resolve the information asymmetry. Challenging the economic perspective that the strategic choices of an organisation can be explained by a rational analysis of costs and benefits alone, researchers have started using institutional theory as the theoretical viewpoint for researching corporate sustainability practices (e.g., Campbell 2006; Husted and Allen 2006), and more specifically sustainability reporting (e.g., Doh and Guay 2006; Larrinaga 2007). Institutional theory regards strategic choices as the results of the social contexts in which they are made, stressing the importance of cultural beliefs and rules, represented by other organizations and institutions.

In their seminal contribution to institutional theory, DiMaggio and Powell (1983) argue that organizational practices and attitudes are becoming more homogeneous through structuration of organizational fields by states and the professions, rather than competition and need for efficiency. This isomorphic change is caused by three mechanisms. The first, coercive isomorphism, is the result of (in)formal pressures on organizations to reflect societal expectations. These include legislation and other pressures from actors, such as NGOs, and the media (Bondy et al. 2012). Second, mimetic isomorphism describes the tendency of managers to imitate other organisations (business leaders). Uncertainty due to developments such as lack of government regulation, or negative publicity (Bondy et al. 2012) is a frequent cause. Third, normative isomorphism is the result of professionalization of the industry and illustrates the tendency to conform to the inherent establishment of norms and values of this professionalization. Companies tend to conform to these pressures in order to maintain or regain their legitimacy so as to safeguard their resources (DiMaggio and Powell 1983).

In recent years, empirical studies have provided some evidence for the influence of institutional factors on sustainability reporting. Higgins et al. (2018) suggest that sustainability reporting is disseminating in clusters, rather than steadily across all businesses. Clusters may refer to what Ortas et al. (2015) label ‘individual, organizational, national and transnational actors and agencies’.

At the national level, there are studies focusing on differences between country clusters or individual countries. Studies focusing on the difference in sustainability reporting between continental European and Anglo-Saxon countries, point at a comparatively lower reporting level for Continental-European countries in earlier years (Jackson and Apostolakou 2010), yet higher reporting levels in more recent years, as a result of coercive pressures from the legal stakeholder-oriented system (Gallego‐Álvarez and Quina‐Custodio 2017). Differences in corporate sustainability commitment are found between French and Spanish companies on the one hand, and Japanese on the other, even if all companies have adopted the UN Global Compact principles (Ortas et al. 2015). A study comparing sustainability reporting in France and Australia also shows differences; however, in high‐risk industries, coercive, mimetic and normative industry pressures overrule national institutional influences (Young and Marais 2012). Pedersen et al. (2013) find evidence for the prevalence of national (coercive), rather than individual organizational factors in the event of the launch of mandatory reporting rules in Denmark. Contrary to the established differences in the above studies, some other studies provide evidence that institutional pressures and the resulting reporting, between countries are rather similar due to normative pressures from global schemes and actors, such as GRI, WWF and Global Compact (Ali and Frynas 2017; De Villiers and Alexander 2014).

At the organisational level, empirical studies mainly focus on differences in sustainability reporters over time. Bhimani et al. (2016) report differences in sustainability reporting between early and late adopters, due to respectively normative/coercive, and subsequent mimetic forces. Similarly, Shabana et al. (2017), find evidence for the prevalence of coercive isomorphism in the first stage (‘defensive reporting’), normative isomorphism in the second stage (‘proactive reporting’) and mimetic isomorphism in the third stage (‘imitative diffusion’) of sustainability reporting adoption. Nikolaeva and Bicho (2011) find that GRI adoption is mainly driven by mimetic pressures, suggesting the importance of competitive and media pressures, and reputational factors.

A third strand of empirical studies, rather than focusing on national or organizational factors, concentrates on the specific institutional contexts in which managers operate, known as ‘fields’. Higgins et al. (2018) describe fields as: ‘(…) groups and individuals that interact and, by so doing, collectively shape norms and expectations within that context’. Fields include contexts such as industries, strategies or sustainability issues. Empirical studies on industry contexts find that companies within the same industry differ in their reporting (Herremans et al. 2016), and that sustainability practices converge over time within an industry, and across industries only when adopted by the industry leaders (Ioannou and Serafeim 2019b). Higgins et al. (2018) analyse the interaction patterns of the actors in what they call the (sustainability) ‘issue-based field’ and conclude that the field attracts sustainability-committed companies on which pressure for sustainability action is exerted, whereas non-reporters have almost no interactions with this field. Bebbington et al. (2009) find that sustainable development reporting is the result of normative, rather than coercive pressures, from interactions between different institutions in a field of companies following a differentiation strategy with regards to sustainable development reporting.

As early as the 1970s, the UN have attempted to establish binding international rules governing the activities of firms (Ruggie 2007). The vision of the UN on the role of businesses in human rights was materialized with the Guiding Principles of Business and Human Rights in 2011. Reaffirming the role of the state as protector of human rights, it also stipulates the role of businesses as one that should respect human rights, independent of the state’s obligations (McPhail and Ferguson 2016; Ruggie 2008). Despite improved acknowledgement of human rights (risks) in corporate reporting in recent years, most reports remain silent on policies, or outcomes thereof (KPMG 2017).

Empirical studies on human rights reporting are scarce. Scant research shows that human rights disclosures are overall low for samples of the top 50 Australian financial service companies for the year 2009/2010 (Islam 2011), the largest 201 GRI-reporters in the world (Cubilla-Montilla et al. 2019), and the top 100 Australian listed companies in the context of ‘modern slavery’ disclosures (Christ et al. 2019). However, for a sample of multinational garment retail companies, it is shown that disclosure of specific human rights standards has increased significantly over time as a result of the development of standards by the ILO, stressing the importance of the role of International Governmental Organizations (Islam and McPhail 2011).

This study investigates how sustainability reporting has developed over the years, focussing on the institutional context, including The Directive, in which this development has taken place. Mandatory reporting rules are considered one of many institutional factors that mutually constitute one another (Witt and Miska 2018). We seek to discover trends in the development of disclosure and assess the potential impact of the recent EU legislation. With this aim in mind, the study combines a longitudinal design with a mixed methods approach. First, we analyse the institutional context in which sample companies operate, listing important events that may have influenced reporting over the years. We focus on one particular sustainability topic, human rights, since The Directive is the first reporting regulation in which human rights are mentioned explicitly. Moreover, in The Netherlands there was no legislation on human rights reporting before (Ministerie van Binnenlandse Zaken en Koninkrijksrelaties 2013). This enables the investigation of a potential effect of The Directive. Moreover, this focus enables more profound, deeper analysis of the complex institutional environment of sustainability reporting. Second, we content analyse the annual and sustainability reports of a sample of companies over the years 2002 to 2018 for inclusion of human rights topics. The resulting data are analysed for potential trends, and subsequently used to estimate the marginal effect of The Directive on the basis of a pre-post analysis.

Our sample consists of all financial services companies listed on the three most important Dutch stock market indexes: the AEX, AMX and AScX, resulting in eight banks and insurance companies. All these companies are public interest entities (PIE), which are directly affected by The Directive.

The choice for The Netherlands as the case for our study is driven by the fact that developments regarding non-financial information disclosure in this country started in a rather early stage compared to most other EU-countries (Kolk et al. 2001). Financial services are an interesting sector in the light of institutional contexts of human rights disclosure, as they are increasingly held accountable for human rights infringements of their clients (DNB 2019). As a result, financial institutions in the Netherlands have been a popular target of NGO campaigns. Two initiatives, in which human rights play an important role, are in line with these developments: In 2009 the first edition of the ‘Fair Finance Guide’, an influential online portal informing consumers about the (lack of) sustainability of investments of banks and (in later editions) insurance companies, was launched. In 2019 the OECD published a guidance tool for the financial services sector to implement the OECD Guidelines for Multinationals (OECD 2019).

For each company, we collected the annual reports from 2002 to 2018. Table 1 shows the eight companies in in our sample in alphabetical order. Column (1) and (2) show the financial years for which annual reports were included. For all companies, we included reports for 18 financial years, with the exception of NIBC, which did not publish a digital annual report for 2002. The preferred language of reports is English: Only when no English report was available, a Dutch report was included. Sustainability reports were included for the years before companies converted to integrated reporting. Column (3) shows the year in which the company first switched to integrated reporting. Column (4) shows the number of sustainability reports included in the sample.

Column (5) shows the mean number of human rights mentions in the annual reports. Column (6) shows the main number of mentions when including sustainability reports. Since sustainability reports generally include more words on human rights than the annual report, we will use different panels to compare the two in our main analysis.

Table 2 provides an overview of the main variables, divided in two panels. Panel 1 refers to a sample with only annual and integrated reports, whereas Panel 2 refers to the whole sample, thus including sustainability reports. Our main dependent variable is the number of times the phrase ‘human rights’ or the Dutch equivalent ‘mensenrechten’ is used in the annual report, expressed as a ratio per 10,000 words, to control for variation in the size of annual reports. The statistics of this measure are shown in Row (3) and (7). The use of this variable is based on the assumption that the more the term is applied in annual reports, the more emphasis an organisation places on human rights reporting. As a complementary measure, we also use an index developed by Pencle and Mălăescu (2016), shown in Row (4) and (8). This validated index is a dictionary containing almost 300 words and phrases that are strongly associated with human rights reporting. Pencle and Mălăescu (2016) developed it by analysing the texts of Initial Public Offerings. As this index does not contain the term ‘human rights,’ it is complementary to our human rights word count. A disadvantage of this measure is that only English reports can be included, resulting into a smaller sample.

Table 3 shows the result of our analysis of the institutional context of human rights reporting by Dutch financial institutions between 2002 and 2018. We searched for the major events and developments over the past three decades that may have pressured the companies in our sample, making use of the following sources: official government information sources¹ and Dutch media (via the Nexis database). These factors are displayed in chronological order. The table shows examples of coercive (such as: growth of SRI, legislation, GRI, OECD (Matten and Moon 2018; Shabana et al. 2017)), mimetic (e.g., business coalitions for CSR, Global Compact (Matten and Moon 2018)) and normative (e.g., initiatives by professional associations (Matten and Moon 2018)) pressures.

What Table 3 clearly shows is that EU Directive is only one of many factors that could have influenced human rights reporting. It also shows that the institutional context, in which human rights reporting decisions are taken, is a complex of constantly changing (inter)national institutional forces, represented by institutions, including companies, that exercise power via negotiation, persuasion, and confrontation. Moreover, analysing the historical developments that occurred before The Directive was implemented, we cannot exclude that both The Directive and human rights reporting are the result of a changing institutional context.

Based on the increasing pressures on financial institutions over the years as shown in Table 3, one would expect to observe a constant increase in the reporting on human rights. In this subsection, we will explore potential trends.

Figure 1 explores the extent to which the sample companies reported on human rights over the years. It plots the proportion of firms that mention human rights at least once in their annual or sustainability report against the subsequent reporting years. Panel 1 shows the trend for annual and integrated reports only, Panel 2 also includes the sustainability reports. The figure clearly shows an increasing trend in human rights reporting over time, starting with 29% of the companies mentioning human rights in 2003 increasing to 100% in 2015 and beyond. The dashed vertical lines on the x-axis represent the year The Directive was announced, and its inclusion in the Dutch Civil Code, respectively. If The Directive were an exogenous shock, one would expect a cut in the time trend here. We do not observe that in this graph.

Figure 2 and 3 explore whether the extensiveness of human rights reporting has increased over time. Both figures show separate curves for Panel 1, which includes annual and integrated reports only, and Panel 2, which also includes sustainability reports. Figure 2 shows that the number of times ‘human rights’ was mentioned in the annual reports, increased over time. Overall, however, mentions of human rights are scarce: Even at the peak in 2016, less than 2 words per 10,000 words in an annual report were devoted to human rights. The trend is different when including the sustainability reports. Panel 2 indicates that most of the reporting on human rights was included in sustainability reports. After 2014, most companies stopped publishing sustainability reports and switched to integrated reporting, which significantly affected the total annual reporting on human rights. We also observe an increase in human rights reporting in Panel 1 after 2014, indicating that some of the human rights reporting switched to the integrated report. This might also be influenced by The Directive, which compels companies to report on human rights in their annual report.

Figure 3 shows the alternative measure for the extensiveness of human rights reporting we use the Pencle and Mălăescu index. It shows a similar trend as Figure 2. There is an upward time trend visible that behaves less erratically than the human rights count (except around 2008), but show otherwise the same pattern, including the sharp decrease in overall human rights reporting (Panel 2) after 2014.

Now that we have a better understanding of the institutional pressures and the increasing trend of reporting on human rights over the past two decades, we are able to explore the marginal effect of the new EU Directive. We estimate the effect by comparing the means in annual reporting before and after the announcement date. Table 4 shows the results of this analysis. Panel 1 shows the average reporting on human rights in the annual reports in our sample before and after the announcement; Panel 2 does the same, but includes sustainability reports. Column (1) and (2) include data from 2013 to 2016 ̶ two years before and after the announcement, Column (3) and (4) include data from 2012 to 2017 ̶ three years before and after the announcement, Column (5) and (6) include data from 2011 to 2018 ̶ four years before and after the announcement.

Table 4 shows some interesting developments after the announcement. First, the proportion of reports that mention human rights increases significantly, whether the sustainability reports are included or not. After 2014, all companies in our sample mention human rights in their annual reporting. Second, the human rights count increases in the annual reports, but overall ̶ including the sustainability reports – there is no significant change. We also find no significant changes in the human rights index scores.

Nevertheless, the analysis in Table 4 merely shows the changes before and after a certain cut-off point in time. We have established that in the Dutch financial sector, a variety of intuitional forces have driven reporting standards over the past two decades, resulting in a steady increase in trend in human rights reporting. If The Directive is regarded as an exogenous shock in human rights reporting, it should affect human rights reporting independently of this trend. Hence, a more accurate estimation of the marginal effect of The Directive can be obtained with an OLS-estimation controlling for the trends over time:

HR_it = α + β₁Directive_t + β₂Year + ε_it

In this equation HR_it is natural logarithm of the ‘human rights’ count of company i in annual report of year t; and A_t a dummy variable equal to 1 if The Directive was first implemented for administrative year t or 0 otherwise. Variable Year represents the year of the annual report and parameter β₂ the trend in human rights reporting over time.

The results of this analysis are shown in Table 5. Column (1) shows the estimates of the effect on the announcement of The Directive, Column (2) shows the effect of the implementation of The Directive and Column (3) shows the estimates of the full model including firm-fixed effect. Columns (4)–(6) follow the same structure but include the sustainability reports.

The estimates in Table 5 confirm our other analyses. There is a significant time trend in human rights reporting over the years, regardless of whether sustainability reports are included. However, when we control for this time trend, we find no significant effect of either the announcement or the implementation of The Directive.