The usage of environmental management control systems and the score on the Transparency Benchmark.

(1)

The usage of environmental management control systems and the

score on the Transparency Benchmark.

MSc Business Administration Organizational and Management Control

June 20th₂₀₁₇

University of Groningen Faculty of Economics and Business Supervisor: Dr. J.S. Gusc

Co-assessor: Dr. H.J. van Elten

By Anmarie Germs

Student number: S2962802 Word count: 12,311

(2)

2

1. Abstract

Currently there are various methods for organizations to report on their environmental practices which leads to a lack of clarity and confusion among stakeholders. To provide clarity and facilitate a stakeholder dialogue, the Ministry of Economic Affairs of the Netherlands started the Transparency Benchmark. Through the changing concerns of the stakeholder towards more environmental concerns, companies are changing their control systems to be more open and externally focussed. The use of environmental management control systems (EMCS) concerns the environmental aspects of organizational performance. The response to the stakeholders’ concerns goes through external reporting. Therefore, it is expected that environmental management control systems influence the external reporting practices and through this the score on the Transparency Benchmark. This leads to the research question on the relation between the usage of environmental management control systems and the score on the Transparency Benchmark. The relation is expected to be moderated through the processing of information, since information systems are at the heart of any company’s control system. After collecting survey data through face-to-face meetings, a PLS analysis has been performed. The outcome of the analysis showed insignificant results for the concepts. However, significant relationships have been found with single variables, mainly concerning human resources, and single quality-oriented criteria of the transparency benchmark. This shows that the support of management and employees is important for EMCS to be successfully implemented. Future research should focus on the types of controls of Malmi and Brown (2008) and their relation to the score on the transparency benchmark since similarities are visible with the content-oriented criteria of the Transparency Benchmark.

Keywords: Environmental Management Control Systems (EMCS), Transparency Benchmark

(5)

5

2. Introduction

Companies are experiencing an increase in social pressure involving issues such as the economic, environmental ecology, and social well-being of employees and society (Chenhall, 2003). This is acknowledged by companies since they incresingly recognize the importance and seriousness of adopting corporate social responsibility (CSR) practices into their business. Perrini and Tencati (2006) indicate that various alternatives have been created to extend traditional methodologies. These methodologies focus merely on financial measurements, where alternatives integrate environmental information for decision making processes and reporting practices. Van Marrewijk (2003) states that, because there is no standard recipe to report on the environment each company should choose the method that is most suiting the companies aim, intentions, strategy, and circumstances. However, this results in various methods to disclose environmental information which leads to a “confusion and lack of clarity for the companies’ stakeholders” (Perrini and Tencati, 2006, p. 299). In addition, it has become more difficult to compare and benchmark the sustainable performance of companies (Siew, 2015). To provide clarity for stakeholders and facilitate a dialogue between companies and its stakeholders the Ministry of Economic Affairs of the Netherlands started the Transparency Benchmark in 2004. The Transparency Benchmark (TB)1_{is “an annual}

research on the content and quality of corporate social responsibility reports of Dutch companies”. This type of ranking make sense according to Daub (2007) because “they promote interest in the reports and contribute to a systematic improvement in the reports themselves” (p. 78).

Recent research on CSR has shifted from the principles of sustainable accounting and reporting towards strategies and management control systems (MCS) involving sustainability (Arjaliès and Mundy, 2013; De Villiers, Rouse, and Kerr, 2016; Gond, Grubnic, Herzig, and Moon, 2012; Henri and Journeault, 2010; Pondeville, Swaen, and De Rongé, 2013). Companies nowadays require suitable systems to measure and control their practices to evaluate whether they respond to the concerns of stakeholders in an effective way (Perrini and Tencati, 2006). By using MCS a company could shape employee’s practices and strategy, and if used correctly aid the company towards a more sustainable direction (Gond et al., 2012). This was already acknowledged by Durden (2008) who said managers would need MCS to regularly monitor whether the company is operating according to their CSR and stakeholder goals. According to Chenhall (2003) “if the MCS are found to be useful then they are likely to

(6)

6

be used and provide satisfaction to individuals, who then presumably can approach their tasks with enhanced information” (p. 132). Pondeville et al. (2013) indicate that information is the raw input of management control and information systems are at the heart of any company’s control system. Information systems collect, process, and store information for control, decision-making, and coordination (Laudon & Laudon, 2002).

Previous researches have been focussing on formal and informal MCS and the influence of these control systems on the perceived performance or environmental disclosure. In this research the focus will be on environmental management control systems (EMCS) and the transparency towards stakeholders in the Netherlands through the TB score. Looking at the TB scores of for example pharmaceuticals, the industry average is 30 points out of the 200 points that can be earned. Are these companies using EMCS to a small extend and consequently have less information available to report to their external stakeholders? Therefore, this research will be examining the relation between the usage of EMCS and the score in the transparency benchmark. This aim leads to the following research question:

What is the relation between the usage of environmental management control systems and the score on the Transparency Benchmark?

The research question was analysed after collecting data from 34 Dutch companies who received a score on the TB of 2016. The usage of EMCS was analysed through both formal and informal EMCS. In addition, the effect of the processing of information on the relationship is studied. The outcome of the Partial Least Square (PLS) analysis shows no significant results indicating that there is no relation between the usage of EMCS and the score on the TB. Even though, this paper contributes to the field of sustainable disclosure and reporting tools in the Netherlands. According to Siew (2015) a lot needs to be done to improve corporate reporting tools and the understanding of the tools by the users. This paper helps practitioners to understand the influence of the TB and provide future research suggestions to investigate the effect of (E)MCS.

(7)

7

3. Concept and Theoretical background

3.1 Transparency benchmark

A key concept of this research is the Transparency Benchmark (TB), specifically the score on the benchmark. Although this is not a recognized scientific theory, the goals, the standardized procedures for implementation, improvement, reviews, and auditing controls make this concept into a grounded theory. A grounded theory is “the relation between concepts which emerged from the population by constant comparing and then are related to each other by a theoretical code” (Glaser, 2016). Since 2004 data has been gathered through the TB about the participating companies and their external reporting practices. The practices of these companies are assessed using scores on different criteria and in the end combined into a theoretical code.

The TB is an initiative from the Ministry of Economic Affairs examining external reporting about corporate social responsibility (CSR) aspects of large companies within the Netherlands. The company with the best performance on CSR reporting will receive the Crystal prize. Handing out an award makes the TB a method of evaluation of corporate sustainability reporting (Daub, 2007). The analysed reports are not only financial reports, but also include corporate responsibility, sustainability, and social reports (Participant Protocol Transparency benchmark, 2016). The fundamental idea behind the TB is that it will encourage companies to be more transparent about their policies and results regarding CSR, which in turn will facilitate a stakeholder dialogue (Van den Berg, Hanemaayer, Wever, & Monod de Froideville, 2013). Besides this, the TB also gives the opportunity to discover trends on both business and sector level. The audit conducted by B&A Groep in 2013 showed an increase of 36% in the TB score of the companies within the target group since the introduction in 2004 (Van den Berg et al., 2013). This increase is supporting the fundamental idea behind the TB to encourage companies to be more transparent. Through the publication of the scores companies are stimulated to receive a higher score next time by being even more transparent.

(8)

8

den Berg et al., 2013). Yet, using the TB as a monitoring instrument is only applicable for large and high scoring companies since stimulation is not essential anymore. The audit by Van den Berg et al. (2013) showed that the TB did not affect the transparency of the lower scoring companies or the zero scoring companies. Therefore, the TB should still be used as a stimulating instrument to increase the transparency of low to zero scoring companies, which was also set on the agenda after the audit in 2013. All in all, the TB could help to monitor the transparency of large companies and should improve stimulating low scoring companies to be more transparent and increase the scores on the benchmark.

3.1.1 The future of TB

In the near future, impacts on the TB are the policy letter “Corporate Social Responsibility pays off”, the increase in reporting with the GRI guidelines, and the adjustments in the EU legislation (Van den Berg et al., 2013). First, the policy letter will pursue on the stimulating strategy of the TB. Second, the TB should be adjusted in a way that it matches the GRI guidelines because the use of these guidelines is already a standard in high scoring companies (Van den Berg et al., 2013). Finally, with the adjustments in the EU legislation, which is mainly focused on large companies, the TB should be altered to give understanding about the results and risks of CSR. This would lead to a shift in the TB towards a monitoring instrument. After the assessment in 2013 the TB has been altered in 2014 putting it in line with the GRI guidelines and integrated reporting. The TB also excluded the medium and small enterprises. In addition, the Dutch government agreed on strengthening the stimulating strategy towards transparency and support the EU legislation. This means that the TB should, as explained, be adjusted with the new EU legislation and the stimulating strategy will be pursued. As a monitoring instrument the TB evaluates the CSR policies and practices of the companies from an external perspective. Internally these companies should control their policies and practices to pursue the EU legislation. A way for companies to perform internal control is through management control systems.

3.2 Management Control Systems

(9)

9

experiencing more uncertain environments there is a need for more open, externally focused, and non-financial types of MCS (Chenhall, 2003). Build upon the definition of MCS of Simons (1995) the definition of environmental management control systems (EMCS) is defined by Pondeville et al. (2013, p. 318) as “a package of formal, information-based routines and procedures that managers use to maintain or alter patterns in organizational activities, specifically concerning the environmental aspects of organizational performance”. In addition, it is important to acknowledge the relevance of informal control. The combination of both formal measurements and informal control steer the MCS towards CSR goals (Durden, 2008). Together with the formal controls within a company the informal control systems ensure the involvement and participation of both managers and employees to discuss environmental issues (Pondeville et al., 2013). Integrating sustainability into MCS would facilitate the implementation of a sustainability strategy which helps to achieve corporate sustainability performance (Wijethilake, 2017).

(10)

10 3.3 Formal Management Control Systems

As explained before MCS consist of “a package of formal, information-based routines and procedures that managers use to maintain or alter patterns in organizational activities” (Pondeville et al., 2013, p. 318). Formal MCS are, according to Langfield-Smith (1997), the most visible and objective aspects of the control system. Formal controls such as limits of authority and power, are written down and know by management and employees (Ouchi, 1979). Examples of these aspects presented by Gond et al. (2012, p. 208) are “strategic planning, budgeting, financial measurement systems, non-financial measurement systems, hybrid measurement systems, project management, and evaluation and reward”. Formal measurements towards sustainability require explicit measures and a regular monitoring approach that reflects the CSR aspects (Norris and O’Dwyer, 2004). Traditional MCS focussing merely on financial measures are perceived as weak in addressing sustainable and social issues regarding to the inclusion of the interests of a wide range of stakeholders, since traditional MCS only tend to focus on the shareholders’ interests (Gond et al., 2012). According to De Villiers et al. (2016) it is easier to operationalize sustainability objectives when these are incorporated into the MCS of a company. EMCS provides managers and employees with a tool to match expected results with decision making, as well as to integrate sustainable goals and performance into the planning and reward systems (Gond et al., 2012; Pondeville et al., 2013). Examples of EMCS, according to Gond et al. (2012), are “environmental budgeting, environmental/sustainability performance evaluation systems, eco-control or sustainability balanced scorecard” (p. 209). EMCS facilitate the discipline to evaluate the aspects covered by sustainability reporting through objective measures (De Villiers et al., 2016). When sustainability reporting is perceived as part of EMCS it ensures that sustainability becomes part of the MCS and the facilitation of management decisions and focus (De Villiers et al., 2016). So, when a company integrates sustainable aspects into their formal MCS it is expected that these sustainable practices are reflected in the companies external reporting, which would result in a higher score on the TB. This results in the first hypothesis of this research.

H1: The use of formal environmental management control systems has a positive relation on the score on the transparency benchmark.

(11)

11

that the two types of MCS should be mutually reinforcing and that both types on its own are unlikely to be sufficient and could affect the achievement of social responsibility goals. An organization is not purely focussing on one type of control but will include features of each of the different types of control (Ouchi, 1979). However, the features of, for example, formal control may be used in a higher extent than informal control at the same moment in time.

3.4 Informal Management Control Systems

In contrast to formal MCS, “an informal system consists of shared values, beliefs, and traditions that guide the behaviour of employees” (Norris and O’Dwyer, 2004, p. 177). Cravens, Lassk, Low, Marshall, and Moncrief (2004) explains that “informal control includes high levels of professional and cultural control” (p. 241). An informal system can aid directing the focus of the company’s employees towards strategic and operating objectives (Durden, 2008). The effectiveness of MCS will vary based on the responses of employees involved (Merchant and Van der Stede, 2012). While formal budgets are perceived as useful in assisting planning, informal communications provide more information in a flexible way (Chenhall, 2003). The integration of sustainability into management control and strategy should be dealt with as something people do instead of looking at management control, either traditional or sustainable, as something a company has (Gond et al., 2012). Pondeville et al. (2013) add that informal systems are implemented to ensure support by both managers and employees in order to solve environmental issues. Because of the effect of informal control systems on employees, the human resource management policies of a company appear to be critical for the successful implementation of sustainability strategies (Pondeville et al., 2013). According to Cravens et al. (2004) the informal controls encourage commitment through the shared values and beliefs visible in the companies’ environment. Informal control recognizes how the formal aspects, monitoring and measurement, are imbedded as part of the company culture (Norris and O’Dwyer, 2004). The perception of sustainability by employees influences the actions of the employees and their sustainable practices and reporting. Therefore, it is expected that there is a relation between the use of informal EMCS and the score on the TB, which results in the second hypothesis.

(12)

12 3.5 Processing of information

The assessment of Van den Berg et al. (2013) showed that there is a clear relation between the scores on the TB and the time spend on the information processing, which probably explicates the importance of transparency according to the company. Due to the position of information in MCS it is important to consider the effect of information processing within a company. “Information processing capacity comprises an organization’s ability to perform information processing activities such as the collection, processing and use of information” (Schick, Gordon, and Haka, 1990, p.203). There are factors that affect the information processing practice. One of these factors is the human aspect within processing information. A limitation is the human ability to process information in an environment with an ever-increasing quantity and variety of information (Macintosh, 1994; Schick et al., 1990). The need to process information is often linked to (task) uncertainty (Macintosh, 1994; Williams and Seaman, 2002) and uncertainty is reduced through information processing (Schick et al., 1990). Macintosh (1994) indicates that high task uncertainty requires more information which needs to be processed in order to get the job done. To manage task uncertainty companies could set rules, goals, and hierarchies that specify prescribed behaviour (Macintosh, 1994). These are, as indicated in the previous section, parts of the formal MCS. Besides managing task uncertainty, well-developed information systems can also facilitate the development and implementation of both formal MCS and informal MCS (Pondeville et al., 2013). This has also to do with the evolution of MCS over the years, as Chenhall (2003) indicates, because the focus changed from more formal financial information to the need for a much wider scope of information. As the need for information changed, the MCS were adjusted accordingly to process the information. Examples of the broader scope are “external information related to markets, customers, competitors, non-financial information related to production processes, and predictive information” (Chenhall, 2003, p. 129).

(13)

13

support planning and control (Perego and Hartmann, 2009). The model of Pondeville et al. (2013) showed that an environmental information system characterises a response to pressure from regulatory stakeholders. For a while several researches have indicated the reasons for disclosing information as a result of stakeholder pressure. For example, Fiol (1995) explained that the reason for most companies to publish corporate information is to show a favourable image to external stakeholders. Conformingly, Gond et al. (2012) highlight that external reporting may serve as a ‘masks’ hiding practices which sole purpose is to reconstruct an eroded legitimacy. A reason for companies to do this is because poor sustainable performance might impact the company’s share price through unwanted publicity (Jones, 2010). More specifically looking at the TB, the increase in transparency is due to; the increasing significance and publicity of CSR, companies perceiving sustainability as an important part of their branding, changes in the Dutch law regarding reports on non-financial information, the increased value for stock listed companies of the Dow Jones Sustainability Index, and the interference of NGO’s regarding the sustainability of companies (Van den Berg et al., 2013).

A case presented by Macintosh (1994) about Apollo computers showed how processing of information influenced the efficiency of the company. The case showed that not enough information was communicated to the top managers, which resulted in top management setting inappropriate operating rules. According to Macintosh (1994) the fundamental idea is “to get the right amount of information into the decision-making process at the right time” (p. 59). Then again, according to Eppler and Mengis (2004) the performance towards decision making of an individual differs with the amount of information the person is exposed to. Showed by an inverted U-curve, the performance of information processing regarding decision making is positive until a certain point and declines rapidly when more information needs to be processed, also known as information overload (Chewning and Harrell, 1990). Conversely, this is not the case with information processing for reporting. With the self-assessment by companies for the TB, the high scoring companies have more information available to fill in the assessment because of elaborated CSR policies (Van den Berg et al., 2013). So, when companies have more information available for reporting, more information can be processed in order to be transparent.

(14)

14

information available when needed there is a positive effect on their disclosure. Therefore, adding up all the aspects, it is expected that the processing of information positively moderates the relation between EMCS and the score on the TB resulting in the third and fourth hypotheses.

H3: The processing of information positively moderates the relationship between the use of formal environmental management control systems and the score on the transparency benchmark.

H4: The processing of information positively moderates the relationship between the use of informal environmental management control systems and the score on the transparency benchmark.

3.6 Conceptual Model

Following, figure 1 shows the conceptual model of this research indicating the relationships which will be examined.

(15)

15

4. Methodology

To measure the relation between formal and informal environmental management control systems (EMCS) and the score on the TB, and the moderating effect of the processing of information, data will be collected through a quantitative survey combined with interview questions. The data collected with the survey is a contribution to the data base of a larger research project of the Accounting department of the University of Groningen. The survey used to collect the data has been developed at an earlier research project at the University of Groningen. A Partial Least Square (PLS) analysis on the quantitative data will be performed using SmartPLS. PLS will be used to predict where relationships might or might not exist (Chin and Newsted, 1999). The steps taken in order to perform the research are explained more into detail in the next sections.

4.1 Population and sample selection

The population of this research are national and multinational companies that have participated in the TB of 2016, in total 483 companies. The companies in the TB are organized into six categories (Participant Protocol Transparency benchmark, 2016):

1. Public interest entities with more than 500 employees 2. Companies listed on the Amsterdam stock exchange

3. Companies with operations in the Netherlands with an accountable turnover and/or number of employees

4. State-owned organizations

5. Universities and university medical centres

6. Large companies (with more than 250 employees) that voluntarily joined the TB this year or in the past

The reason that only large companies are included in the benchmark is a result of the obligation to include socially responsible aspects in the annual reports of which companies with less than 250 employees are exempted (Van den Berg et al., 2013).

(16)

16

Therefore, the zero scoring companies are excluded from the population because external stakeholders could not evaluate the perceived quality of the reports. Resulting in a population of in total 252 companies.

4.2 Sample selection

In order to investigate whether there is a relation between the use of EMCS and the score on the TB the sample should include both companies scoring high and low on the benchmark. Therefore, theoretical sampling is used instead of random sampling. However, there is no selection based upon the industry the companies are operating in to give a boarder view of the Dutch market. Since the current data base consists of a majority of high scoring companies, the focus is on low scoring companies which will be added to the data set. The low scoring companies are companies with a score below 91. According to Cohen (1992), a research with three independent concepts in the measurement and structural model requires a sample size of 30 with a statistical power of 80% for detecting R2 values of at least 0.25 (10% significance level) (Hair, Hult, Ringle, and Sarstedt, 2017).

4.3 Respondents

From each of the companies one business unit (BU) manager will participate in the research. The choice of BU managers is based upon their key role of being the link between lower level employees and top management of an organization, in which BU managers acknowledge and use EMCS (Reimer, van Doorn, and Heyden, 2016). The respondents need to have knowledge about the use of EMCS to participate in the research (Remenyi, Williams, Money, and Swartz, 1998). Therefore, it requires for BU manager participating in the research to be responsible for one or more areas; full profit and loss, revenue, cost, and/or (not-for-profits) budget. These areas of responsibilities are most often observed by BU managers who are responsible for more than 25 employees. Contact with these BU managers is made through LinkedIn and email followed by phone calls to discuss the details and set an appointment of an hour.

4.4 Data collection

(17)

17

participant not being fully honest but providing the most desirable answer resulting in an increased response bias. Furthermore, to lower the response bias minimal information is provided to the participant beforehand and during the data collection. The participant will be informed about the research and the main topics, but, for example, the relations that will be research will not be explained to the participant. Through this the participant will not be able to give pre-informed, favourable answers.

In total six appointments are scheduled by the researcher to collect data as a contribution to the larger data base. Before an appointment two surveys will be printed, one for the participant and one for the researcher. While the participant is filling in the survey, the researcher will available to assist the participant when more information is needed if a question might be unclear. It will take approximately 40 minutes to fill in the survey. In addition to the survey, some in depth interview questions will be asked to the six respondents who the researcher will meet. The interviews will be recorded and transcribed afterwards. The use of an interview helps to increase the reliability of this research. Van Aken, Berends, and Van der Bij (2012) explain that surveys and interviews have different advantages, and combining them yields complementary results. In addition to the complementary effect on the results, these interview questions will help to explain or illustrate the findings of the survey (Remenyi et al., 1998). The survey and interview questions can be found in Appendices A and C. The data is collected within a time frame of eight months, from October 2016 to May 2017.

4.5 Operationalisation of the concepts: measurement

The concepts of this research are: formal EMCS, informal EMCS, processing of information, and the score on the TB. All concepts are treated as reflective in this research meaning that the direction of causality, between the variables and concept, points from the concept to the variables (Riou, 2016). In the reflective measurement model “measures represent the effects or manifestations of an underlying construct” (Hair et al., 2017, p. 58).

(18)

18 4.5.1 Formal EMCS

The concept is measured using nine items based upon the measurements created by Pondiville et al. (2013, p. 323). The nine items include:

FEMCS 1. Sustainability objectives

FEMCS 2. Sustainability performance indicators and rewards FEMCS 3. Sustainability performance indicators and promotion FEMCS 4. Comparison of results and objectives

FEMCS 5. Sustainability information on employees FEMCS 6. Sustainability criteria in investment decisions FEMCS 7. Sustainability rules and procedures

FEMCS 8. Sustainability functions

FEMCS 9. Procedures external sustainability communication 4.5.2 Informal EMCS

This concept is measured using eight items which are based upon the measures created by Pondiville et al. (2013, p. 323). The eight items are:

IEMCS 1. Employee suggestions to the sustainability of the primary process

IEMCS 2. Employee suggestions to the sustainability of the primary product/service IEMCS 3. Employee suggestions field of sustainability

IEMCS 4. BU manager involvement sustainability management process IEMCS 5. Sustainability discussed during meetings

IEMCS 6. Freedom to manage sustainability issues IEMCS 7. Work teams manage sustainability problems

IEMCS 8. Cooperation between different BU on sustainability issues 4.5.3 Processing of information

The moderating concept is measured using in total six items. Out of these six items four items are based upon the measures of Pondiville et al. (2013, p. 323). The other two items of this survey are developed at an earlier research project at the University of Groningen. All six items are as followed:

PI 1. Sustainability information regarding possible future events PI 2. Sustainability information on broad external factors

PI 3. Non-economic information

PI 4. Quantification likelihood future sustainability-related events PI 5. Sustainability information provided frequently

(19)

19 4.5.4 Score on the TB

Finally, the score on the TB is the dependent concept gathered from a secondary data source. The scores used for this research will be the results of the most recent year, namely 2016, and will be retrieved from the TB website on May 9, 2016. On behalf of the ministry, the TB is currently performed by EY and several activities are outsourced to MVO Nederland2_{. The score on the TB is a result of the self-assessment done by the participating}

company, which is than analysed by EY. After the analysis of EY a panel of experts is assessing the twenty best scoring companies. Finally, the ministry will officially adopt the scores, ranging from 1 to 200, into the benchmark. The scores are divided into eight criteria, of which the first three are content focused and the other five are focused on quality (Criteria Transparency Benchmark 2017, n.d.). More information about these criteria are available in appendix B. The eight criteria and their maximum score are:

TB 1. Company and business model (33) TB 2. Policy and results (34)

TB 3. Management approach (33) TB 4. Relevance (20) TB 5. Clearness (20) TB 6. Reliability (20) TB 7. Responsiveness (20) TB 8. Coherence (20) Total score (200) 4.5.5 Control variable

Finally, the effect of a control variable will be measured. As Hair et al. (2013) advice, if the effect of a control variable is significant the finding needs to be used to carefully draw conclusion. Out of the two primary means of a control variable, the focus in this study is on statistical control in which the control variable is included in the primary analysis (Becker, 2005). According to Spector and Brannick (2011) the assumed effect of a control variable is to create a bias in the observed relationships. Therefore, a control variable is included in this research to rule out the assumed effect of the variable. The control variable in this research is the size of the organization. The size of the organization will be measured based on the number of employees and a logarithm of this number is used to measure the effect. A reason to include the organization size as a control variable is because the size measured through the

(20)

20

number of employees can have a positive effect on the adaptation and extend of CSR reporting (Hahn and Kühnen, 2013). It could be assumed that larger companies have a bigger impact on society and more resources to produce high quality external reports.

4.6 Consistency and Validity

The evaluation of these concepts relies on three characteristics, namely internal consistency, convergent validity, and discriminant validity (Hair et al., 2017; Riou, 2016). Since the moderator concept is also a reflective measurement model, it should be treated similar to the dependent and independent concepts by using the same criteria (Hair et al., 2017). First, internal consistency is reflecting the reliability of the measurement through the inter-correlations between the items (Hair et al., 2017; Riou, 2016). The traditional criterion to measure the internal consistency according to Riou (2016) is the Cronbach’s Alpha (α). On the other hand, Hair et al. (2017) suggest, in addition to the Cronbach’s Alpha, measuring the different outer loadings of the items, which is called composite reliability. Composite reliability is similarly interpreted as the Cronbach’s Alpha, requiring a value between 0.70 and 0.90 (Hair et al., 2017). In addition, the Cronbach’s Alpha should represent the lower bound of the evaluation (> 0.70) since the composite reliability could overestimate the internal consistency (Hair et al., 2017). When items are removed from the concept, this should be done carefully. Hair et al. (2017) advices to delete outer loadings between 0.40 and 0.70 when this leads to an increase in the composite reliability resulting in a required value, while loadings below 0.40 are always deleted.

Secondly, the convergent validity is about the level to which an item relates to the other items within the same concept (Riou, 2016). In order to evaluate the convergent validity the average variance extracted (AVE) is calculated (Hair et al., 2017). An AVE value of 0.5 or higher means that, on average, the concept explains more than half of the variance of its items (Hair et al., 2017; Riou, 2016).

(21)

21 4.7 Analysis

Once the construct measures are reliable and valid the structural model will be analysed. A Partial Least Square (PLS) analysis will be performed using SmartPLS. The choice for a PLS analysis is based on the fact that the sample size is small, normally distributed data is not required, it places minimum requirement on the measurement level, and has the possibility to incorporate reflective measurement models (Hair et al., 2017; Pondeville et al., 2013). PLS estimates the path coefficients and helps to explain the variance (R-square) in the dependent concept, the score on the TB, when examining the model (Hair et al., 2017). “The path coefficients have standardized values approximately between -1 and +1” (Hair et al., 2017, p. 184).

In addition, bootstrap resampling, with 500 resamples, will be used to test the significance of the estimated values for the path relationships in the structural model (Chin, 1998). The standard procedure of computing resampling statistics without compensating for sign changes will be followed. Bootstrapping makes it possible to calculate the empirical t-values for all path coefficients (Hair et al., 2017). A path coefficient is statistically significant at a certain significance level when the empirical t-value is larger than the critical value (Hair et al., 2017). The critical value used in this research is 1.65 with a two-tailed test and a significance level of 10%.

Even though the concept and moderator are measured reflectively, a two stage approach is used to determine whether the moderator has a significant effect on the relationship (Hair et al., 2017). The effects will be estimated and evaluated in two analyses to avoid a common mistake of confounding the direct and simple effects (Hair, Ringle, & Sarstedt, 2013). First the direct effects in the PLS path model are evaluated, secondly the moderator analysis is performed including the product term and its interaction effect.

4.7.1 Level of analysis

(22)

22

influence of corporate level management control is much higher compared to the influence of the local business unit conditions. Looking at multinational firms, significant influences are observed from the parent company on business unit MCS resulting in similarities compared to other business units of the company despite differences in location (Van der Stede, 2003). Two reasons could explain the effect of similar MCS within the different business units of a company. First, Pratt, Mohrweis, and Beaulieu (1993) found in their sample that firms recruit and hire people who have similar values to those working in the company. Secondly, this could be the effect of coercive, mimetic and normative isomorphism (DiMaggio and Powell, 1983).

(23)

23

5. Results

The results of this research are presented starting with the analysis of the variables and concepts, followed by the descriptive statistics of this research and the PLS analysis. In order to gather the data for analysis some struggles were encountered. Since the focus was on low scoring companies, the willingness to participate was lower compared to high scoring companies. “People who are more interested in environmental issues and place a higher value on environmental quality may be more inclined to respond to the survey” (McNeill and Roberge, 2002, p.57). Due to these struggles the average TB score of the sample, including 34 companies, is 123.97. The median of the TB score is 157, indicating that the majority of the sample scored above average. Compared to the original intended sample, including companies with a score below 91, the average score is perceived as high.

5.1 Analysis of concepts

First, the evaluation of the concepts on the internal consistency, convergent validity, and discriminant validity are presented. Table 2 on page 25 shows the results of the internal consistency and convergent validity analysis. Looking at the formal EMCS, two out of the nine variables were removed based upon their outer loading which was below 0.70. By removing these variables, the composite reliability and AVE increased above the threshold (0.70 and 0.50 respectively). The removed variables are:

1. Sustainability criteria in investment decisions

2. Sustainability performance indicators and promotion

Even though there are four other variables with an outer loading below 0.70, these were not removed because the outer loadings are close to 0.70 and the required composite reliability has been reached.

Secondly, the evaluation of the informal EMCS resulted in the removal of one variable, namely the freedom to manage sustainability issues. The variable has been removed because the outer loading was below 0.40, which are variables that always need to be deleted. The composite reliability of informal EMCS (.909) is a bit above the required value of 0.90. Although values above 0.90 are not desirable, no changes will be made since the composite reliability of informal EMCS is close to the upper bound.

(24)

24

below 0.70. Therefore, it has been decided to continue the calculations with one variable for processing of information, namely sustainability information on broad external factors.

Looking at the variables of the TB score all the outer loadings are high. In addition, the Cronbach’s Alpha, composite reliability, and AVE are also very high. Such a high composite reliability is not desirable as it indicates that all the measures are measuring the same phenomenon (Hair et al., 2017). However, deleting measures does not result in a decrease of the values below 0.90.

As a final step the discriminant validity is assessed using the Fornell-Larcker criterion. Table 1 shows the outcome of the assessment. The value of the same concept, indicated in bold, shows a higher value than the greatest value with another concept, which shows that all variables are valid measures of unique concepts.

Formal EMCS Informal EMCS Processing of information TB score

Formal EMCS 0.710

Informal EMCS 0.643 0.767

Processing of information 0.596 0.189 1.000

TB score 0.221 0.291 0.271 0.935

N=34

(25)

25

Concepts and items IL* α* CR* AVE*

Formal EMCS .834 .876 .504

FEMCS 1. Sustainability objectives .638

FEMCS 2. Sustainability performance indicators and rewards .621

FEMCS 4. Comparison of results and objectives .692

FEMCS 5. Sustainability information on employees .740

FEMCS 7. Sustainability rules and procedures .841

FEMCS 8. Sustainability functions .698

FEMCS 9. Procedures external sustainability communication .716

Informal EMCS .892 .909 .588

IEMCS 1. Employee suggestion to sustainability of the primary process .710 IEMCS 2. Employee suggestion to sustainability of the primary

product/service

.734

IEMCS 3. Employee suggestion field of sustainability .830 IEMCS 4. BU manager involvement sustainability management process .723

IEMCS 5. Sustainability discussed during meetings .767

IEMCS 7. Work teams manage sustainability problems .781

IEMCS 8. Cooperation between different BU on sustainability issues .812

Processing of information 1.00 1.00 1.00

PI 2. Sustainability information on broad external factors 1.00

TB score .979 .982 .873

TB 1. Company & business model .938

TB 2. Policy and results .952

TB 3. Management approach .953 TB 4. Relevance .910 TB 5. Clearness .936 TB 6. Reliability .868 TB 7. Responsiveness .968 TB 8. Coherence .949

* IL= Indicator Loading, α= Cronbach’s Alpha, CR= Composite Reliability, AVE= Average Variance Extracted

(26)

26 5.2 Descriptive statistics

The descriptive statistics will give a better image of the sample of this research. First, table 3 shows the respondent characteristics.

Mean Max Min St. Deviation

# employees BU 386 5000 12 886.14

# employees organization 11,829 104,500 350 23,002.65

Years as BU manager 6 25 0.50 5.71

Years in the organization 14 38 0.67 10.06

N=34

Table 3, Descriptive respondent characteristics

Following, table 4 shows the responsibility areas carried by the BU managers. As mentioned before the areas of responsibility are based upon the requisite that managers have knowledge about the EMCS used. A bit more than half of the sample has full profit and loss responsibility (56%).

Area of responsibility N Frequency Percentage

Full profit and loss responsibility 34 19 56%

Revenue responsibility 34 4 12%

Cost responsibility 34 7 20%

(Not-for-profits) budget responsibility 34 4 12%

Table 4, Responsibility areas

(27)

27

Mean Min* Max St. Deviation

TB score 123.97 10 195 64.24

TB 1. Company & business model 23.206 0 33 10.437

TB 2. Policy & results 20.500 2 34 10.421

TB 3. Management approach 20.765 -1 33 11.680 TB 4. Relevance 12.147 -2 20 7.105 TB 5. Clearness 13.676 2 20 6.659 TB 6. Reliability 8.294 0 20 7.290 TB 7. Responsiveness 11.824 0 20 7.857 TB 8. Coherence 13.559 0 20 7.030 Formal EMCS 3.731 1 7 1.796

FEMCS 1. Sustainability objectives 3.559 1 7 1.802

FEMCS 2. Sustainability performance indicators and rewards 2.824 1 7 1.599 FEMCS 4. Comparison of results and objectives 3.971 1 7 1.774 FEMCS 5. Sustainability information on employees 3.676 1 6 1.710 FEMCS 7. Sustainability rules and procedures 4.382 1 7 1.627

FEMCS 8. Sustainability functions 3.706 1 7 1.807

FEMCS 9. Procedures external sustainability communication 4.000 1 7 1.815

Informal EMCS 4.130 1 7 1.650

IEMCS 1. Employee suggestion to sustainability of the primary process 4.676 2 7 1.470 IEMCS 2. Employee suggestion to sustainability of the primary product/service 4.176 1 7 1.635 IEMCS 3. Employee suggestion field of sustainability 4.265 1 7 1.578 IEMCS 4. BU manager involvement sustainability management process 4.206 1 6 1.549 IEMCS 5. Sustainability discussed during meetings 4.265 1 7 1.633 IEMCS 7. Work teams manage sustainability problems 3.500 1 7 1.576 IEMCS 8. Cooperation between different BU on sustainability issues 3.824 1 7 1.806

Processing of information 4.294 2 7 1.426

PI 2. Sustainability information on broad external factors 4.294 2 7 1.426

*Points are be deducted when no information is given on certain topics N=34

Table 5, Descriptive statistics concepts and variables

5.3 Hypotheses tests

(28)

28

presented path coefficients are not significant using a 10% significance level. Therefore, hypotheses one to four are rejected based on the data.

Coefficient t-value Conclusion

H1 Formal EMCS  TB Score -0.209 0.655 Not significant

H2 Informal EMCS  TB Score 0.348 1.178 Not significant

H3 Moderating effect formal EMCS  TB Score 0.097 0.386 Not significant

H4 Moderating effect informal EMCS  TB Score 0.079 0.328 Not significant

Processing of information  TB Score 0.344 1.512 Not significant

Table 6, Hypotheses test results including moderating effect

Figure 2, Model including path coefficients

(Dashed line = negative relation, solid line = positive relation)

(29)

29

All correlations show a weak variance on the TB criteria. Chin (1998) describes that the R-square, indicating the explained variance, should be considered substantial (.67), moderate (.33), or weak (.19). Furthermore, significant path coefficients should be interpreted relative to each other, in which they should be at least 0.20 but when above 0.30 the path coefficient should be considered (Chin, 1998; Hair et al., 2017). Significant correlations are visible with the TB criteria reliability, clearness, and coherence, that are part of the quality-oriented framework of the TB. The content- and quality-quality-oriented frameworks of the TB are presented in appendix B.

Coefficient R-square t-value

Formal EMCS

FEMCS 1. Sustainability objectives  TB 6. Reliability 0.310 0.096 1.987*

FEMCS 2. Sustainability performance indicators and rewards  TB 6. Reliability 0.363 0.132 2.338* FEMCS 3. Sustainability performance indicators and promotion  TB 6. Reliability 0.346 0.119 2.237* FEMCS 5. Sustainability information on employees  TB 6. Reliability 0.340 0.116 2.315* Informal EMCS

IEMCS 7. Work teams manage sustainability problems  TB 6. Reliability 0.427 0.183 2.692 ** IEMCS 8. Cooperation different BU on sustainability issues  TB 5. Clearness 0.367 0.135 2.385* IEMCS 8. Cooperation different BU on sustainability issues  TB 6. Reliability 0.381 0.145 2.396* IEMCS 8. Cooperation different BU on sustainability issues  TB 8. Coherence 0.334 0.112 2.027* Processing of information

PI 2. Sustainability information on broad external factors  TB 5. Clearness 0.366 0.134 2.466* PI 2. Sustainability information on broad external factors  TB 6. Reliability 0.325 0.105 2.165*

* t-value > 1.96 (5% significance level) ** t-value > 2.57 (1% significance level)

Table 7, Path coefficients single correlation analysis

(30)

30

has an individual effect on the clearness and reliability score, of which the clearness score has a higher path coefficient.

5.4 Control variable

Finally, the effect of the control variable on the TB score has been analysed. This is done by adding the control variable, size of the organization, to the model including all concepts and the moderating effect. Figure 3 shows the results of the analysis including the control variable. The path coefficient (0.398) and the t-value (2.541) of the control variable show a significant effect at a significance level of 5%. Adding the control variable did not change the significance of the other path coefficients, which remain insignificant. Since the effect of the control variable is significant, it needs to be considered when drawing

conclusions.

Figure 3, Model including path coefficients and control variable * 5% significance level

(31)

31 5.5 Level of analysis

As explained in chapter 4.7.1, there is a difference in the level of analysis between the survey data and the TB score. Table 8 presents the descriptive statistics on the similarity of formal and informal EMCS between different BUs within the organization according to the participants. None of the participants indicated similarities to a small extend, through responding one on the scale. The mean of the similarities in formal EMCS (4.77) and informal EMCS (4.79) are above average. In addition, the median of both types of EMCS (5.00) indicates that more BU managers believe there are similarities with the other BUs. Similarities in the formal EMCS are mainly because of corporate influences and similar measures used across the organization. “We use the same formats for formal management control systems, the BSC and similar matters” (respondent 4). According to the participants are similarities in informal EMCS mainly due to close contact with employees and other BU managers. However, the influence of employees makes that similarities are not visible to a high extent. “Informally we talk about sustainability with each other, but everyone is free to anticipate on it.” (respondent 6).

Mean Median Max Min St. Deviation

Similarities formal EMCS 4.77 5.00 7.00 2.00 1.55

Similarities informal EMCS 4.79 5.00 7.00 2.00 1.30

N=34

(32)

32

6. Additional Results & Discussion

The research question of this paper is: What is the relation between the usage of sustainable management control systems and the score on the Transparency Benchmark? Even though a significant relation was expected based on the theory, the outcome of the analysis show that there is no relation between these concepts and the moderating effect. However, significant correlations are found between single variables.

Looking at the single correlations analysis, significant coefficients showed that there are correlations with the quality-oriented criteria, and not the content-oriented criteria, of the TB. This could be explained by the scope of the survey questions. Malmi and Brown (2008) describe five types of control, namely; “planning, cybernetic, reward and compensation, administrative, and cultural controls” (p. 291). For example, the administrative control is divided into three groups, namely “organization design and structure, governance structure, and the procedures and policies” (Malmi and Brown, 2008, p. 293). These groups can be compared with the content-oriented criteria of the TB, examples of criteria are; “explanation on the policy carried out by the company, insight into the organization structure of the company, and explanation of the (composition of the) board” (Criteria Transparency Benchmark 2017, n.d.). Through these similarities, it could be expected to observe a relation between controls and the score on the TB. However, the items of the survey are focusing on broad formal and informal EMCS that represent single items of the types of controls by Malmi and Brown. This could explain why no relation between the concepts is observed. In addition, the survey questions are focussed on the environmental aspects of control while the criteria of the TB are also covering traditional controls. To observe the expected relation between controls and the score on the TB, more items, both environmental and traditional, should be added to the survey so the controls are represented.

(33)

33

reliability score of the TB. The two variables are work teams manage sustainability problems and the cooperation between BU on social issues. In addition, three of the formal EMCS variables related to employees show a relation with the quality-oriented criteria namely; sustainability performance indicators in compensation/rewarding systems, sustainability performance indicators in promotion/career advancements, and sustainability information on employees. The interviewee mentioned that although formal controls need to be implemented, informal management is more important, “The strength of the informal organisation is really high” (respondent 6). Besides the influence of managers, employees, and informal EMCS, no other explanations could support the correlations with the quality-oriented criteria of the TB. These correlations are perceived as coincidental.

Surprisingly, the interviews learned that BU managers (N=5) have never heard about the TB. Nevertheless, BU managers indicated that the participation in the TB is in line with the philosophy of their company as the company is or aims to be transparent. The fact that the participation on the TB is not known by the BU managers could be explained by a lack of internal communication, where numbers are only reported towards the holding. However, “the information from the holding to the BU could be better and more consistent” (respondent 1). Another interviewee indicated that “Two or three years ago a focus on sustainability was desired, but I don’t know what has been done” (respondent 2). On the other hand, one participating company indicated that they have a small environment division “this division has a lot of information, however, there is not a 100% connection with the management. The management is well informed, and the information is always available from the environment division” (respondent 5). To increase support throughout the organization, the communication from the holding towards BUs should increase.

(34)

34

example Cisco Systems International, besides their annual report, Cisco has a publicly available CSR report. Although they mention only the Netherlands as their EU head office, it does not mean that they are not transparent on their business (Corporate Social Responsibility 2016 Report, 2016). Even though, because of the criteria of the TB, Cisco obtained a zero score. Understandable, certain criteria need to be set in order to perform an analysis as the TB, but this creates a biased rating.

In addition, during the most recent kickoff meeting of the TB 20173_{approximately}

100 participants were present, where 483 companies participated on the TB of 2016. In addition, people involved in the process of the TB highlighted that high scoring companies are looking for ways to receive the maximum points to get the highest score. This suggest that the TB may be perceived as a publicity tool instead of reflecting the EMCS or performance of the company, “In an annual report you describe what you have done, after that you get into marketing atmospheres where it is about nice to have” (respondent 6). Siew (2015) explains that unsupported environmental and social disclosure could be an attribute of public relations rather than a tool to address CSR. These practices make it more difficult to assess sustainable performance based on external reporting and to distinguish leaders from laggards. In addition, as Darnall, Seol, and Sarkis (2009) indicates, a company could report on environmental performance using just CSR reports instead of creating formal and informal EMCS.

Finally, since the control variable showed a significant effect, it is important to include this outcome in the discussion. The literature review by Hahn and Kühnen (2013) showed that the company size is a determinant on sustainability reporting which shows significant consistent results. The reason company size is considered as an effect on external reporting, is not because of the company size per se, but the company’s resources, knowledge, legitimacy, and visibility which is a result of the company size (Bowen, 2002; Hörisch, Johnson, and Schaltegger, 2015). Larger companies possess more resources, either financial or human, which makes it easier to gather information for external reporting. In addition, through the influence of knowledge the effect of the company size on the use of sustainability management tools is even stronger (Hörisch et al., 2015). Finally, since large, mostly multinational, companies possess sufficient financial resources they are able to create more detailed reports regarding content and design (Daub, 2007). Although there are still large companies receiving a low score on the TB, the company size has an effect on external reporting and through that also on the TB score.

(35)

35

7. Conclusion & Future Research

To conclude, this research shows that there is no relationship with formal and informal EMCS and the TB score. A reason for this could be that the EMCS in this research do not completely cover the topics of the criteria on the TB because the focus of these criteria is not only on environmental aspects. Therefore, future research on the TB score should focus on the five control types of Malmi and Brown (2008). Through the analysis of the relationship between the control types and the score on the TB it would be easier for companies to observe whether there is an influence. This would help to anticipate on their score and, if preferred, to adjust their controls to increase their score on the TB. In addition, if the company would strive to be more sustainable and get a higher score on the TB, information should be made available more consistently. Through this, the commitment and support in the organization would be higher. Future research should investigate the processing of information and lack of communication from higher management to BU managers. As participants indicate, there is awareness on the application of sustainability, but no additional information could be provided. What are the reasons this is not done by the higher management, and if this is done, would this increase the score on the TB.

In addition, the Ministry of Economic affairs should reconsider the future of the TB. Although a shift towards a monitoring instrument is advised after the changes in the EU legislation, the TB should still focus on stimulating low scoring companies. Through the stimulation, the TB would not become a publicity tool of high scoring companies. Also, the criteria on the zero score should be investigated to reduce the number of zero scoring companies even further. Through this a more general image could be created of the transparency of Dutch companies.

(36)

36

8. References

8.1 Articles

Becker, T. E. (2005). Potential problems in the statistical control of variables in organizational research: A qualitative analysis with recommendations. Organizational Research Methods, 8 (3), 274-289.

Bowen, F. E. (2002). Does size matter? Organizational slack and visibility as alternative explanations for environmental responsiveness. Business & Society, 41 (1), 118-124.

Chenhall, R. H. (2003). Management control systems design within its organizational context: findings from contingency-based research and directions for the future. Accounting, organizations and society, 28 (2), 127-168.

Chenhall, R. H., & Morris, D. (1986). The impact of structure, environment, and interdependence on the perceived usefulness of management accounting systems. Accounting Review, 61 (1), 16-35.

Chewning, E. G., & Harrell, A. M. (1990). The effect of information load on decision makers' cue utilization levels and decision quality in a financial distress decision task. Accounting, Organizations and Society, 15(6), 527-542.

Chin, W.W. (1998). The partial least squares approach to structural equation modeling. In Marcouldes, G.A. (Ed.), Modern methods for business research (pp. 259-336). New York, NY: Psychology Press.

Chin, W. W. (2010). How to write up and report PLS analyses. In Vinzi, V.E., Chin, W.W., Henseler J., & Wang, H. (Ed.), Handbook of partial least squares (pp. 655-690). Berlin: Springer.

Chin, W.W., & Newsted, P.R. (1999). Structural Equation Modeling Analysis with Small Samples Using Partial Least Squares. In Hoyle R.H. (Ed.), Statistical Strategies for Small Sample Research (pp. 307-341). London: SAGE publications Ltd.

Cohen, J. (1992). A power primer. Psychological bulletin, 112 (1), 155-159.

Cravens, D. W., Lassk, F. G., Low, G. S., Marshall, G. W., & Moncrief, W. C. (2004). Formal and informal management control combinations in sales organizations: The impact on salesperson consequences. Journal of Business Research, 57 (3), 241-248.

(37)

37

Daub, C. H. (2007). Assessing the quality of sustainability reporting: an alternative methodological approach. Journal of Cleaner Production, 15 (1), 75-85.

De Villiers, C., Rouse, P., & Kerr, J. (2016). A new conceptual model of influences driving sustainability based on case evidence of the integration of corporate sustainability management control and reporting. Journal of Cleaner Production, 136, 78-85.

DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: institutional isomorphism and collective rationality in organizational fields, American Sociological Review, 48 (2), 147-160.

Durden, C. (2008). Towards a socially responsible management control system. Accounting, Auditing & Accountability Journal, 21 (5), 671-694.

Eppler, M. J., & Mengis, J. (2004). The concept of information overload: A review of literature from organization science, accounting, marketing, MIS, and related disciplines. The information society, 20 (5), 325-344.

Fiol, C. M. (1995). Corporate communications: Comparing executives' private and public statements. Academy of Management Journal, 38 (2), 522-536.

Glaser, B.G. (2016). Grounded Description: No No. Grounded Theory Review An

international journal, 15(2). Retrieved from

http://groundedtheoryreview.com/2016/12/19/grounded-description-no-no/

Gond, J. P., Grubnic, S., Herzig, C., & Moon, J. (2012). Configuring management control systems: Theorizing the integration of strategy and sustainability. Management Accounting Research, 23 (3), 205-223.

Hahn, R., & Kühnen, M. (2013). Determinants of sustainability reporting: a review of results, trends, theory, and opportunities in an expanding field of research. Journal of Cleaner Production, 59, 5-21.

Hair, J. F., Hult, G. T. M., Ringle, C., & Sarstedt, M. (2017). A primer on partial least squares structural equation modeling (PLS-SEM). Los Angeles: Sage Publications Ltd.

Hair, J. F., Ringle, C. M., & Sarstedt, M. (2013). Editorial - Partial least squares structural equation modeling: Rigorous applications, better results and higher acceptance, Long Range Planning, 46 (1-2), 1-12.

The usage of environmental management control systems and the score on the Transparency Benchmark.