“Analyzing the quality disclosure of Global Reporting Initiative G4 sustainability report in Indonesian companies”

The establishment of a company cannot be separated from its environmental and social factors. Sustainability reports start from those applied to current companies because there are forms of corporate accountability to stakeholders and community consider- ations of the company to provide social responsibility. This study finds out and empirically proves that there are differences in each Global Reporting Initiative (GRI) G4 indicator in the company’s sustainability report in each industry classification. The authors investigate the dominant indicators in each industry classification based on sustainability reports. The data are obtained from 28 GRI G4-based company sustainability reports in 2016 and 2017. The analytical method in the study is the K-means clustering analysis. The results of study indicate the differences in GRI G4 in 2016 and 2017. The researchers find out that the dominant indicator expressed in the financial industry is an economic indicator. Meanwhile, in the mining, transportation and infrastructure industries, basic and chemical industries etc. the dominant indicators to be disclosed are environmental indicators. This research provides a theoretical basis for sustainability and environmental reporting, particularly in the context of developing countries. It is expected that this study should also inform business practitioners as well as policymakers vis-à-vis sustainability reporting in practice.


INTRODUCTION
Environmental damage has become a serious problem in recent years. Many companies exploit natural resources and human resources to increase their profits. Damage arising from the production of automatic goods or services will increase so that taxes and fees for cleanliness, health, and environmental budget will also continue to increase. There are demands from the public for companies to provide social responsibility. A concept introduced by Elkington (1988), i.e. people, planet, and profit, is called the triple bottom line concept that measures the success of a company. The concept is a term known as sustainability, where the company can survive as long as possible and is called the long-life company.
Many companies in the world are required to provide accountability reports. Compilation of sustainability reports is important because there are disclosure principles and standards that reflect the overall level of company's activity. Reports do not only focus on financial aspects as in financial statements. Stakeholders are also particularly interested in understanding how the firm's approach and performance In Indonesia, sustainability reporting is still voluntary in contrast to such reporting as annual reports and financial reports. According to the Global Reporting Initiatives (GRI), at the end of 2016, 120 companies in Indonesia published sustainability reports according to the Global Reporting Initiatives (GRI). In fact, in 2015, there were only 63 companies. Not all companies that report sustainability reports are in the same industry. In Indonesia, there are hundreds of companies listing on the IDX, but not all publish sustainability reports because they are still voluntary.
For this reason, we are interested in researching and proving the notion that in each of the industrial classifications, the quality of disclosure of sustainability reports is different according to the GRI G4 indicator. Central to this research, are there differences in the quality of disclosure in each of the GRI G4 indicators in the sustainability reports of public companies in each industry? Moreover, what indicators become dominant and are expressed in each industry? This research is intended to shed light on the theme of sustainability disclosure. This research is also expected to provide the information for companies in carrying out sustainable and responsible performance as well as establishing regulations in company's operations.
The structure of this paper is as follows. Section 1 discusses the background, problems, purposes, and objectives of the research. Section 2 contains the theory and literature review as the basis for this research, including previous studies and hypotheses formulation. Section 3 elaborates the research framework, population and sample, including research models, operationalization of variables, and testing methods. Section 4 analyzes the research results and implications for the research model presented.
Final section concludes the results as well as presents the limitations and suggestions.

THEORETICAL BACKGROUND AND LITERATURE REVIEW
Stakeholder theory shows that the firm is not only responsible for the welfare of the company but also must have social responsibility, taking into account the interests of all parties affected by the company's strategic actions or policies. The success of a company depends on its ability to balance the various interests of stakeholders (Lako, 2011). Stakeholders are all internal and external parties that have a direct or indirect influential relationship with the firm. Firms should pay attention to their stakeholders because they are the parties who both influence on and are being influenced by the policies and action taken by the firm. If the firm does not take care of its stakeholders, it is likely to reap protests and eliminate the stakeholders' legitimacy (Adams, 2002).
Legitimacy theory states that an organization can only survive if the community in which it is located feels that the firm performs based on a system of value that is commensurate with the value system that is owned by the community. Thus, organizations continually strive to act in accordance with the boundaries and norms in society, so that their activities are accepted according to the perceptions of external parties (Deegan, 2002). The rationale for this notion is that the firm will continue to exist if the community acknowledges that the firms operates within a similar value system. Legitimacy theory encourages and unarguably promotes the firms to make sure that their performance and activities are acceptable to the society.
Firms utilize their reports to describe the image of environmental responsibility so that they are accepted and supported by society. With the acceptance of the community, the value of the firm is expected to raise, so that it can increase profits. This can simulate and attract the investors in making the investment decisions. Regarding the legitimacy theory, it can be said that this study considers the image of the company from the point of view of society. Social and environmental activities carried out by the company to the community are a form of corporate responsibility (corporate social responsibility) to build a good corporate image and can also encourage or increase profits for the company.
Meanwhile, according to Elkington (1997), sustainability report means the report that contains both financial and non-financial information regarding its social and environmental activities that can help them to grow sustainably. According to GRI (2013), sustainability report is a practice in measuring and disclosing the company's activities, a commitment to external and internal stakeholders regarding the firm performance in realizing the new development agenda. Sustainability report is voluntary and complimentary but completely separated from the firm's financial statements in its submission (Iman, 2019). The World Business Council for Sustainable Development (WBCSD) explains the benefits of SR, among others: providing the information to stakeholders and improving the company's prospects, and helping to realize the transparency; help establish an image as a tool that contributes to increasing market share, brand value, as well as long-term consumer loyalty; a reflection of how the company manages the risks; a stimulus of leadership thinking and competitive performance; developing and facilitating the implementation of a management system that is better in managing the environmental, economic, and social impacts.
The principle of reporting plays an important role in achieving the transparency and must therefore be applied by all organizations when preparing the sustainability reports. One of the initial images used by companies in developing SR is to adopt an accounting method called the triple bottom line. Companies that want sustainability must pay attention to "3P." The company must be able to fulfill the welfare of the people contribute to maintaining the environmental preservation (planet), and pursue the profit (Iman, 2019).
The sustainability report disclosure is guided by the Global Reporting Initiative (GRI) report. GRI  From the statement above, to show that industry has carried out management activities from the viewpoint of economic, environmental, and social impacts, it is necessary to disclose in the sustainability report to build the firm's image and perception in each of the industry classifications. In GRIbased economic, environmental, and social disclosures, companies can present information on all indicators of what impacts are expressed through the index in the closing section of the report or present impact information directly without an index in the report commonly referred to as GRI reporting citing.
Research on sustainability reporting (SR) is still rarely done, but in recent years empirical research related to the sustainability reporting (SR) has grown rapidly from various types of sectors and variables. For example, Roca (2012) analyzes the disclosure of indicators in the company's sustainability report. This research is a case study of a company in Canada. The indicators were determined through content analysis in the year 2008 reports. The study showed that 585 different indicators were used in the report. As many as 31 out of 94 reports disclose the special GRI-based standards.
For instance, Tarigan (2014) reexamined the relationship between sustainability report disclosure and financial performance. This study uses the companies that consistently publish sustainability reports, and, second, this study uses all measures of financial performance that include asset management, profitability, leverage, liquidity, and market. The samples used were 54 observations from companies that consistently published sustainability reports. Similarly, Marwati (2015) tested and analyzed the impact of liquidity, return on assets, company size as well as earnings per share (EPS) on such disclosure. The sample used is a non-financial company registered on the Indonesian Stock Exchange (IDX), which issues the sustainability reports in accordance with the GRI standard in the 2009-2013 period, and found 12 companies. The data analysis technique is the classical assumption test.
Moreover, True (2015) examined the disclosure of sustainability report (SR) on the performance and value of the company guided by the Global Reporting Initiative (GRI). Population data were taken from the companies that publish sustainability reports and are publicly listed on the Indonesian Stock Exchange from 2006 to 2013. The results of this study indicate that disclosure of sustainability reports does not have a significant relationship with company performance and company value. Then, disclosure of economic performance, environmental performance, and social performance of sustainability reports also does not have a significant relationship with company performance and company value.
Similarly, Utama (2016) tested the effect of sustainability report (SR) disclosure as a moderating variable on the intellectual capital (IC) on firm performance based on 21 Indonesian publicly listed companies and registered at the national center for Indonesian SR chapter. This study uses the published model of the value-added coefficient intellectual (VAIC) to determine the company's IC. The result shows that VAIC has a positive effect on ROA and ROE. This means high ROA and ROE of the companies are associated with more VAIC. In addition, VAIC has no effect on GM. The results of the moderated analysis also show that disclosure has a positive effect on ROA and ROE, but has no effect on GM. Sustainability reporting disclosure became a pure moderator on ROA while being a pseudo moderator.
Jusmarni (2016) examined the relationship between the sustainability report indicator and the company's market value and asset management ratio. Independent in this research is the sustainability report disclosure, which is divided into economic, environmental, and social performance indicators and measured using the SRDI index. The sample of this study is 15 companies that publish sustainability reports three years in a row. As a result, sustainability reporting in the economic and environmental aspects has a significant positive impact on the market value ratio and asset management ratio, while the sustainability reporting in the social aspect is not significantly positive in increasing market value and asset management.
Lastly, Puspitandari (2017) obtained the evidence regarding the effect of sustainability report disclosure and each aspect of sustainability report on the performance of banks in Indonesia that issued GRI-based sustainability reports. The samples are 13 companies each year, and for 3 years there are 39 sustainability reports. The results show that sustainability report disclosure has a significant positive impact on banking performance, so it can be viewed that the increasing sustainability report disclosure will improve the banking performance, besides, disclosure of economic, environmental, and social performance aspects has a significant positive effect on banking performance, so it can be concluded that increasing disclosure of economic, environmental, and social performance aspects will also improve the banking performance.
The sustainability report for which disclosure is guided by the Global Initiative or GRI with the most recent reporting basis, GRI G4, which regulates the disclosure of indicators that are covered in 3 categories, namely economic categories, environmental categories, and social categories. Through cluster analysis, companies will be classified into relatively homogeneous groups, called clusters. In this case, relatively homogeneous companies show that they are a group that reports GRI G4-based economic, environmental, and social information.
Companies publicly listed on the Indonesian Stock Exchange (IDX) are the subjects of this study originating from various industrial sectors, such as finance, mining, transportation and infra-structure, various industries, and basic industries and chemistry, so that this difference enables the data processing systems in the research to detect and generate information about the differences in objects between the clusters produced. Therefore, based on this explanation, the hypotheses can be formed as follows: H1: There are differences in the quality of disclosure on each GRI G4 indicator in the sustainability report of companies listed on the IDX in 2016-2017 in each industry characteristic.
H2: There is a disclosure of the dominant indicators in the sustainability report on the financial industry classification.
H3: There is a disclosure of the dominant indicators in the sustainability report on the classification of the mining industry.
H4: There is a disclosure of the dominant indicators in the sustainability report on the classification of the transportation and infrastructure industries.
H5: There is a disclosure of the dominant indicators in the sustainability report on various industry classification.
H6: There is a disclosure of the dominant indicators in the sustainability report on the classification of basic and chemical industries.

RESEARCH METHODOLOGY
This research is focused on the disclosure of all indicators in the GRI G4, namely in the economic, environmental, and social categories, to compare the industries that have been grouped based on their classification. The indicators used in this study are 91. This research refers to all aspects, namely economic, environmental, and social impacts, in the GRI standard 4. Each indicator represents the aspects of the disclosure of indicators used as data processing materials for research.
The research method used in this study is descriptive quantitative analysis, namely by finding the information about the existing circumstances and defining them as clearly as possible to achieve the research objectives. This research begins by providing the values for each indicator in GRI G4 based on the company's sustainability report. This value indicates a challenge or benefit for each indicator. Furthermore, industrial classification carried out into sectors in this type of industry was continued by the classification of companies into relatively homogeneous groups (clusters).
This study uses 6 (six) variables, which are the aspects of the GRI G4 disclosure indicators, specifically, in sustainability, namely economic, environmental, and social reports (including the sub-categories of employment practices and work convenience, human rights, community/society, and responsibility for product). Next we divide the sample into 5 (five) industry classifications: mining (including coal, oil and gas, metals and other minerals, and rocks), finance (banks, financial institutions, securities companies, and insurance), and transportation and infrastructure (energy, toll roads, ports, airports and the like, telecommunications, transportation, and non-building construction), as well as various industries (automotive and components, textiles and garments, footwear, cables, electronics), and basic and chemical industries (cement, ceramics, porcelain and glass, metals and the like, chemicals, plastic and packaging, animal feed, wood and its processing, and pulp and paper).
Cluster analysis begins with the formulation of the problem, or the selection of indicators that will be used for cluster formation. The set of indicators that will be selected must describe the objects. There are 6 (six) variables identified as company KPIs in disclosure of environmental impacts, with statements on a scale where 0 = there is absolutely no disclosure in the GRI G4 indicator, 1 = disclosure in the GRI G4 indicator included in each variable, the company is perceived as having a challenge and 2 = if the company feels it finds benefits in conducting or disclosing the GRI G4 indicator.
Second, to measure the distance or similarity between pairs of objects that are most commonly used is Euclidean distance or its square value. The Euclidean distance is the root of the sum of squares of differences or deviations in the values for each indicator.
Third, the selection of clustering procedures using non-hierarchy, often called K-means clustering. The method used is the optimizing partitioning method, where objects are then reassigned to the cluster to optimize a comprehensive criterion. Non-hierarchical procedures or K-means clustering were chosen to find out the preferences of researchers about the desired cluster, so that in the non-hierarchical procedure, the number of clusters and the selection of arbitrary clusters must be stated/predetermined. In addition, non-hierarchical procedures are more beneficial if the object or case or observation has a large number of samples.
Fourth, determining the number of clusters based on the selection of cluster procedures, namely non-hierarchy and considering the objectives of the research. We identified the indicators of economic, environmental, and social impacts, which become the priorities of the industry classification. We analyzed the clusters in each industry or as a whole that are obtained based on data processing in SPSS automatically.
Finally, interpreting and profiling the clusters includes the study of centroids, namely the average value of the objects contained in the cluster on each indicator. Centroid values allow the researchers to decipher each cluster by giving a name or label. The cluster profiling stage is made based on information obtained from the results of data testing for each cluster formed. Then it was developed so that information on economic, environmental, and social disclosure was dominant in each of the industrial sectors.
The research sample was taken by purposive sampling method from the data population of companies listed on the Indonesian Stock Exchange (IDX). The purposive sampling method itself is a sampling method taking into account the selection criteria: being publicly listed and having released their sustainability reports in 2016 and 2017, respectively. As many as 53 companies that publish sustainability reports and are listed on the Indonesia Stock Exchange were found, but of the 53 companies, only 30 companies have published the sustainability reports in 2016 and 2017, respectively.
The data in this study were taken from secondary data or indirect data taken through the sustain-ability report of firms listed on the Indonesian Stock Exchange between 2016 and 2017. The firm's sustainability report is obtained through the company's official websites and the related websites, and website of the Indonesia Stock Exchange. Researchers also conduct the library studies by reading books, scientific works, theses, the internet and studying the literature contained in the libraries or in other sources that aim to gather the information that is relevant to the topic or problem that is the object of research.
We conducted cluster analysis, which is a technique for classifying the objects into homogeneous or relatively homogeneous groups. This group is called a cluster. The collected data are then processed by using the non-hierarchy or K-means clustering procedure and then explaining the results obtained objectively and systematically. The output produced in the form of a final cluster center shows that the cluster pairs are truly separated. Using an ANOVA/F-test, objects were systematically entered in the clusters to maximize the differences in clustering of each indicator. The greater the value of F and (sig < 0.05), the greater the difference in indicators on the cluster formed.
Furthermore, we conducted the advanced data processing such as analyzing each cluster's output to determine the number of memberships per cluster, membership cluster output to provide the information about objects or cases that have been classified into each cluster and distance or similarity between pairs of objects, final output cluster centers to determine whether or not there are differences in the dominant indicators in each cluster, and output iteration history to get the right number of iterations and the minimum distance between cluster centers from the iteration results.

RESULTS AND ANALYSIS
The population and research sample in this study were taken from publicly listed companies in Indonesia. This analysis uses purposive sampling by looking at sustainability reports of the companies listed on the Indonesian Stock Exchange (IDX). Data were selected by 53 companies that published the sustainability reports, then the companies that issued sustainability reports for two consecutive years, namely in 2016 and 2017, were selected. We got the results of 30 companies. Next we classify in 7 types of industries, and cross out 2 companies from the agricultural classification and consumer goods industry. Finally, we obtained a sample of 28 companies with 5 industrial fields. The ANOVA test results to test the differences in disclosure on each indicator are shown in the following table. In general, in all industries, the quality of disclosure of the GRI indicator in the sustainability report has differences, indicated by a significance value of less than 0.005 on the indicator.
In economic indicators, there are 4 out of 9 indicators, which are different in their disclosure, namely EC5, EC6, EC8, and EC9. In environmental indicators, there are 26 out of 34 indicators, which are different in their disclosure, namely EN1, EN2, EN4, EN7, EN9, EN10, EN11, EN12, EN13, EN15,  Meanwhile, in the sub-category of responsibility for the product, there are 6 out of 9 indicators, which are different in their disclosure, namely in PR1, PR2, PR4, PR6, PR7, and PR9. Thus, it can be said that the research hypothesis H1, which states that there are differences in the quality of disclosure in indicators of sustainability reports in 5 industry fields, is accepted.
Next, we conducted a clustering test without following a hierarchical process called K-means clustering. Each object that has been inputted will be assigned or combined to the classification of the closest cluster center automatically by the system. The central classification will be updated until stopping criteria are reached. The next stage is to do clustering in each industry and find membership clusters with the smallest distance values as findings as in the following table. In the financial sector, the most dominant cluster indicator in cluster 2 is case number 7, namely CIMB Niaga in 2016 with a distance of 4.883. This shows that the quality of disclosure of sustainability indicators at CIMB Niaga is reported more evenly on each indicator, but with the same quality of disclosure. While in the mining sector, the most dominant cluster indicator in cluster 1 is case number 2, Bukit Asam in 2017 with a distance of 4.865. This indicates that the sustainability indicators are more evenly distributed, but with the same quality of disclosure.
In the infrastructure and transportation sectors, the most dominant cluster indicator in cluster 1 is case 10, namely PT AKR. In 2017, the data distance is 4.356, indicating that sustainability indicators are more evenly distributed, but with the same quality of disclosure. As for the various industries, it can be seen that membership clusters that are the most dominant in cluster 2 are cases 3 and 4, namely United Tractors in 2016 and 2017 with a distance of 4.123 for both. This shows the sustainability indicators more evenly, but with the same quality of disclosure. Finally, in the basic and chemical industry sectors, it can be seen that the cluster indicators dominant in cluster 1 are cases 1 and 2, namely Pupuk Indonesia in 2016 and 2017 with distance of 4.330 for both. This implies that it is a more equitable indicator of sustainability, but with the same quality of disclosure.
Whereas in the mining sector, environmental variables dominate more, especially on indicators: • EN2: material used for recycled material input; • EN5: energy intensity; • EN7: energy reduction in products and services: • EN13: protected habitat; • EN15: direct greenhouse gas emissions; • EN16: intensity of greenhouse gas (GHG) emissions; • EN20: emission of Ozone-Depleting Substances (ODS); • EN22: significant air emissions; • EN23: total weight of waste based on type and method of construction; • EN26: habits associated significantly affected by wastewater; • EN29: report of fines and non-monetary sanctions due to non-compliance with environmental laws and regulations; • EN32: percentage of screening of new suppliers using the environmental criteria.
As expected, in the transportation and infrastructure sectors, the dominant variable is the environment, i.e.: • EN2: percentage of recycled material input material; • EN6: comparison of senior management employed from the community; • EN7: energy reduction in products and services; • EN23: total weight of waste based on type and method of construction; • EN29: report of fines and non-monetary sanctions due to non-compliance with environmental laws and regulations; • EN34: number of complaints by environmental impacts.
Meanwhile, in various industrial sectors, the dominant variable is also the environment, which is on indicators: • EN22: significant air emissions; • EN23: total weight of waste based on type and method of construction; • EN25: the total weight of waste transported, imported, exported; • EN29: report of fines and non-monetary sanctions due to non-compliance with environmental laws and regulations.
Finally, in the basic industrial and chemical sectors, environmental variables are also dominant, namely on indicators: • EN5: energy intensity; • EN6: comparison of senior management employed from the community; • EN7: energy reduction in products and services; • EN8: total water withdrawal based on source; • EN10: total volume of water recycled; • EN11: operational locations owned, adjacent to protected areas; • EN12: significant impact of activities, products and services on biodiversity; • EN13: protected habitat; • EN15: direct greenhouse gas emissions; • EN19: emission reduction (GHG); • EN20: Ozone-depleting substances (ODS) emissions; • EN23: total weight of waste based on type and method of construction; • EN25: total weight of waste transported, imported, exported; • EN28: product percentage those sold and their packaging reclaimed according to category; • EN29: report of fines and non-monetary sanctions due to non-compliance with environmental laws and regulations; • EN31: total expenditure on environmental protection; • EN32: percentage of screening of new suppliers using environmental criteria; • EN34: number of complaints by environmental impacts.

CONCLUSION
This study aims to find out and prove empirically that there are differences in the quality of disclosures of each indicator in sustainability reports of companies listed on the IDX whether it is dominant in each industry characteristic. The data used are derived from sustainability reports, with 28 companies studied during the 2016 and 2017 reporting periods.
The data were obtained and processed to carry out the tests on the problem using the cluster analysis methods, namely non-hierarchical procedures or K-means clustering. Based on the conducted study, it can be concluded that there are differences in the quality of disclosure on each GRI G4 indicator in the sustainability reports of companies listed on the IDX in 2016-2017 in each industry characteristic; the difference in indicators is 66 out of 91 indicators.
The dominant indicator expressed in the financial industry is an economic indicator. In the mining, transportation and infrastructure industries, various industries, and basic and chemical industries, the dominant indicators to be disclosed are environmental indicators. This is understandable, considering that the financial industry is being very focused on economic interests (e.g., Iman, 2018), while other industries are more closely related to environmental factors.
Of course, there are some limitations in this study. First and foremost, there are still very few companies in Indonesia that publish their sustainability reports. This is feared to interfere with a more comprehensive quantitative analysis. In addition, the companies` samples obtained are limited to five industrial sectors. Therefore, in further research it`s recommended to expand the reporting period and, if possible, make comparisons with other countries. Nevertheless, it is expected that the findings of this study can be a stepping stone to open horizons related to sustainability in the context of developing countries.