PROBLEMY EKOROZWOJU – PROBLEMS OF SUSTAINABLE DEVELOPMENT

This study examines the sustainability reports (SRs)of 200 firms in both developed and emerging economies in order to identify the words most frequently used in disclosing sustainability practices within the Triple Bottom Line (TBL) approach to reporting (which emphasizes economic, environmental, and social dimensions). Its aim is to evaluate these sustainability reports under the umbrella of the GRI framework. It adopts a semi-automated TextMining (TM) technique to evaluate the corporate SRs of select firms from the top ten economies by GDP at current prices. Based on the GRI Standards guidelines, a total of 208 keywords were identified for analysis. The disclosures were then awarded points based on the appearance of these keywords so that the appearance of one resulted in the awarding of a score of one; if a keyword did not appear then the report was scored a zero for that word. Furthermore, a wordcloud was also generated in order to better understand the inclination of reporting language towards various TBL reporting categories. This analysis of the SRs of 200 firms from the top ten economies of the world sheds light on the differences in reporting practices and priorities as they relate to various aspects of the GRI Standards guidelines. The results indicate that SR practices have grown rapidly in the last half decade of the period selected for study (2013-2017) as compared to the first half (2008-2012). Canada ranked highest for its disclosure practices in this analysis followed by the UK, Germany, US, Japan, France, Italy, Brazil, India, and China. This study found that all included countries improved their sustainability performance over the period 2008-2017.


Introduction
Climate change is a major barrier to sustainable development. The footprints of this change can be traced by analyzing the frequently observed unexpected natural disasters throughout the world, such as floods, droughts, earthquakes, etc. In addition, 2016 was classified as the warmest year according to statistics, indicating that global warming will be more brutal to the present civilization in the near future, if not effectively addressed (NASA, 2016). However, greenhouse gas (GHG) emissions are identified as one of the biggest causes of global warming and climate precariousness (IPCC, 1996). Due to major anthropogenic activities such as the combustion process of fossil fuels, carbon gasses are released into the environment. Therefore, carbon emissions, the most important GHGs, are responsible for stepping up the issue of atmospheric instability (Kaygusuz, 2009). About 90% of these anthropogenic activities are related to various business firms. Therefore, it is time for society to impose checks and balances on various business activities to decelerate the speed of environmental degradation and sustainability reports provide stakeholders with an effective tool to track their activities. The first landmark report on sustainability was published in 1987, titled Our Common Future, it discussed the then current standards for sustainability reporting followed by major businesses (UNWCED, 1987). The report led to an increase in the responsibilities of corporations and led to the development of terms such as corporate social responsibility (CSR), sustainability reporting (SR), and corporate disclosing practices (CDPs). Interest in such issues grew and in just a few decades about 90-95% of large businesses began publishing their CSR reports or sustainability reports (SRs). The use of SRs has grown exponentially (Cho et al., 2012), resulting in the submission of millions of digital files and creating a virtual ocean of information on the internet. Many organizations including the Sustainability Accounting Standards Board (SASB), The Prince's Accounting for Sustainability Project (A4S), World Business Council for Sustainable development (WBCSD), Account Ability (AA), Carbon Disclosure Project(CDP), Indian Centre for Corporate Social Responsibility (ICCSR), and the Global Report- 1 The first version of the GRI framework G1 was pub- ing Initiative (GRI) are involved in attempts to improve the disclosure quality of SRs. These agencies have produced many guiding principles for SR practices, but the continuous revision of the GRI framework 1 and the simplicity of incorporating these guidelines into reporting practices has established GRI as the de facto standard globally (Spuerket et al., 2017). Most firms follow the GRI guidelines for filing their SRs in order to increase their credibility or reputation (Hedberg & von-Malmborg, 2003) and different stakeholders refer to them when investigating the content of disclosures (Searcy & Buslovich, 2014). This study investigates the SRs of publicly listed companies in the top ten global economies as classified by the International Monetary Fund on the basis of GDP at current prices (World Bank, 2018). As compared to previous research in which content analysis was used to investigate such reports and which is limited to a few variables over a short span of time or a single country, this study is based on an investigation of corporate SRs produced by top ten economies between 2008-2017. This paper utilizes Text-Mining (TM) technique to help us gain a better understanding of international SR language in the dynamic business environment. The remainder of the paper proceeds as follows. The next section 2 describes the review of literature and research objective while section 3 presents the methodological aspects of study. The results and discussion are shown in section 4 while the conclusion makes up section 5.

Review of Literature
The first well-established definition of sustainable development published in the report of the Brundtland Commission in 1987 defines it as meeting the needs of the present without compromising the ability of future generations to meet their own needs (UNWCED, 1987). More often the terms CSR, CDP, and SR are used in an organizational context, yet they lack a concrete definition which is universally accepted (Linnenlueck & Griffiths, 2010). Despite this, SR practices can broadly be categorized into three types: economic, environmental, and social as mentioned in the Triple Bottom Line (TBL) approach promulgated by Elkington in 1994 (Kumar & in 2011, G4 in 2014, and the latest framework GRI Standards in 2016. Das, 2018). The TBL concept focuses on the disclosure of facts and figures related to various parameters in the socio-economic and environmental areas. There has been a common perception that SR practices are primarily a western concern (Chapple & Moon, 2005) and that the firms in Asian countries often fall behind western firms on various parameters of SR (Baughn et al., 2007). But by publishing an increasing number of SRs (KPMG, 2016) and contributing to research in this area (Fifka, 2012), Asian countries such as India and China have begun contributing to the development of sustainable practices (Kumar & Das, 2018). The frequent financial shocks experienced by firms over the last few decades and an increasing awareness of sustainable development practices by civil society has also resulted in the rapid evolution of the concept of SRs globally (Beattie, 2014 (Freundlieb & Teuteberg, 2013). A still developing branch of research in this field also adopts critical disclosure analysis (CDA) which is an interdisciplinary approach analyzing spoken and written text in order to investigate SR issues (Merkl-Davies & Koller, 2012). Some of these studies use predefined terms that limit the environment (i.e. ecological limit) by using context analysis (Bjørn et al., 2016). Content analysis is generally understood to be a quantitative form of research, yet it significantly differs from the conventional approach of quantitative research as it involves the analysis of texts created by others and data that has not been generated by a pre-coded survey questionnaire method ( Every year a large number of SRs are filed by various organizations and evaluating these together is a tedious job. Hence, there is a need for a smart and intelligent framework to evaluate these reports effortlessly and which would also have the capability of extracting hidden information present in these reports. Prior efforts have been made to apply the ML technique to the evaluation of corporate SRs but it has been unable to uncover the types of content the researchers sought. Although, Modapothala & Issac (2009) in their study used this technique successfully to find the pattern line, but the scope of results was limited to a few variables only. Some scholars like Shirata et al., (2011) have developed a TM system to examine the financial reports of firms and concluded that it is easier to identify the potential threats from non-financial data compared to financial data. TM is also successfully implemented in areas of medical science for evaluating radiology reports (Prasad et al., 2010) and medical reports (Friedlin et al., 2011). TM has the potential to uncover latent information hidden in the different texts produced by firms. It also has the potential to save firms from future losses by revealing the true picture of political and consumer preferences. However, few scholars have explored this area through the application of text-mining (Liew et al., 2014, Modapothala et al., 2010), with most studies using a very limited number of reports for their analyses. TM uses a combination of various computer programming methods and tools to produce a text-corpus (i.e. words of bag) and to uncover discursive constructions used in the communication of information related to various parameters in corporate SRs (Jaworska & Nanda, 2016). Compared to other types of professional correspondence, corporate SRs remain an under-investigated area in management science. Previous studies have either focused on mandatory or voluntary reporting practices using disclosuresanalytics or linguistic analysis, which is conceptual in nature and often unable to highlight the dynamic nature of corporate SR practices. Qualitative disclosures analysis scholars consider the quantitative analysis of disclosures as deductive in nature  while qualitative disclosures analyses are often criticized for being excessively subjective and lacking sound empirical grounding by quantitative scholars (Beattie, 2014). The primary research objective of this study is to generate a scorecard based on various parameters reported in SRs and the secondary objective is to uncover the inclination of language used in them by analyzing a wordcloud of the most frequently used words in order to increase our understanding of reporting practices broadly.

Data Source and Research Methodology
The study adopts semi-automated-TM to evaluate the corporate SRs of select firms from the top ten economies. In this method, every text file itself is considered a concept vector and each concept vector contains a large number of terms used with a different frequency commonly referred to as the vector's dimensions. In other words, every unique term used in the text file denotes a different dimension of a vector. In order to reduce the dimensionality of the vector and remove unwanted information, pre-processed text files have been used in this study (Crain et al., 2012). It utilizes the multiple keyword extracting technique to find information about the various GRI parameters. These keywords are unique terms that occur in the concept vector and altogether de-  The firms for which sustainability reports, integrated reports, or CSR reports were found missing for any year (i.e. 2008-2017) were dropped from the sample and another firm was selected randomly from the affected index. 2008 was chosen as the starting point for this study because of the drastic increase in the filing of SRs by various organizations as per GRI guidelines. Before this year, very few SRs were filed (GRI, 2016). Furthermore, only SRs which were published in English were found appropriate for analysis Ultimately, 2000 SRs published during 2008-2017 were included in the analysis. Various keywords from all 33 aspects of economic, environmental, and social reporting were extracted from the latest GRI guideline commonly known as GRI Standards. The list of keywords used is provided in annexure-1. A total of 208 keywords were identified for analysis. The appearance of each keyword leads to the awarding of a score of one for a report for that word and if the word is not disclosed then the score is zero (Kumar & Das, 2018). For example, if all 208 keywords were present in a report then a firm can earn a maximum 208 score for that report. Similarly, the maximum score which can be earned by the firms is 2080 for the ten years of reporting. All semi-automated computer program codes were scripted in R using various packages such as tm, SnowballCC, RColor-Brewer, ggplot2, wordcloud, biclust, Rcampdf, etc.

Results and Discussions
During analysis, it was observed that the scores of firms for their SR practices in developed economies improved over the time period under study (i.e. 2008-2017). The results for SR practices in top ten economies have been classified into two categories: developed economies and emerging economies. The developed economies consist of US, Japan, Germany, UK, France, Italy, and Canada while emerging economies consist of China, India, and Brazil. The empirical results are shown in the following subsections. The annexure 1 represents the mean scores obtained by the respective countries, listed from the highest ranked to the lowest ranked country. In addition, at end of this section, figure 2  The categorical investigation shows Canadian firms are more focused on environmental disclosures than social and economic disclosures. A possible reason for this may be the dominance of the Canadian mining industry throughout the world (MAC, 2017) and therefore in Canada, the orientation of its legal compliance is focused on preservation of the environment. GRI_302, GRI_305, and GRI_306 were among the top three reported GRI parameters while GRI_407, GRI_206, GRI_401 were least reported. The linguistic analysis also confirms the pattern found during the categorical investigation as four out of ten words-emission, water, reuse, and pollutionwere related to environmental aspects in the list of top ten most used words. The most frequent words used in reports filed by Canadian firms, as represented by the wordcloud in Figure 2a, are mine, emission, water, reuse, pollution, employees, program, support, service, and networks. United Kingdom (UK): The number of UK firms filing SRs to the GRI has also been continuously increasing. According to the GRI database, only 68 firms had filed SRs in 2008 but it increased at CAGR of 13.53% and in 2017, a total of 242 firms had filed SRs. After Canada, firms in the UK also performed well in sustainability disclosures and scored the second highest mean average score. The overall GRI scores of the 20 sampled UK firms jumped from 64.67 in 2008 to 134.61 in 2017 as shown in annexure 1. The categorical investigation shows firms in the UK are more focused on social and environmental disclosures as compared to economic disclosures. GRI_301, GRI_304, and GRI_306 were among the top three reported GRI parameters while GRI_409, GRI_205, GRI_402 were least reported. The linguistic analysis of the SRs of UK firms is inclined towards social aspects of disclosure as seven out of ten words are related to society. The most frequent words used in reports filed by UK firms is represented in Figure 2b's wordcloud. It shows that the top ten words are people, community, ethics, positive, procurement, engagement, planet, business, performance, and scorecard. Germany: The number of German firms filing SRs to the GRI has been increasing as well. According to the GRI database, only 48 firms had filed SRs in 2008 but it increased at CAGR of 11.68% and in 2017, a total of 145 firms had filed SRs. German firms are also doing well in sustainability disclosures and scored the third highest mean average score. The overall GRI scores of 20 sampled firms jumped from 62.12 in 2008 to 130.92 in 2017 as shown in annexure 1. The categorical investigation shows German firms are more focused on environmental disclosures than social and economic ones. The possible reason for this may be Germany's stringent compliance measures focused on the preservation of the environment. GRI_303, GRI_305, and GRI_302 were among the top three reported GRI parameters while GRI_401, GRI_203, GRI_405 were least reported. The most frequent words used in reports filed by German firms represented in the Figure 2c wordcloud include employees, health, services, program, management, support, issues, company, reduce, and initiatives. The various policies encouraged German firms to implement green practices in their business operations as evidenced in this study by reporting on parameters such asGRI_303, GRI_305, and GRI_302. The significance and current status of SR practices within Germany prove that the work of German firms toward achieving sustainable development (SD) has improved notably. As with Canadian firms, the categorical investigation shows that US firms have an approach more focused on environmental disclosures than social and economic ones. GRI_302, GRI_305, and GRI_306 were among the top three reported GRI parameters while GRI_407, GRI_206, GRI_401 were least reported. Linguistic analysis does not support the categorical analysis as the inclination of US firms leans more towards social issues. The most frequently used words in reports filed by US firms are represented in Figure 2d'swordcloud. The top ten words used by US firms are social, responsibility, principles, rights, planet, people, resources, employees, business, and services. Japan: The number of Japanese firms filing SRs to GRI has been on the rise. According to the GRI database, only 183 firms had filed SRs in 2008 but it increased at CAGR of 1.15% and in 2017, a total of 205 firms had filed SRs. Japanese firms performed well in sustainability disclosures and scored the highest mean average score. The overall GRI scores of 20 sampled firms jumped from 57.86 in 2008 to 123.54 in 2017 as shown in annexure 1. The categorical investigation shows Japanese firms have a more focused approach towards social disclosures as compared to environmental and economic disclosures. GRI_403, GRI_401, and GRI_415 were among the top three reported GRI parameters while GRI_308, GRI_307, GRI_304 were least reported. And the linguistic analysis of SRs shows that Japanese firms consider all three aspects of TBL reporting. The most frequently used words in reports filed by Japanese firms represented in the Figure 2e wordcloud include business, energy, management, corporate, production, plan, csr, income, committee, and hazardous. France: The number of French firms filing SRs to GRI was also found to increase during the time period under study. According to the GRI database, only 32 firms had filed SRs in 2008 but it increased at CAGR of 19.17% and in 2017, a total of 185 firms had filed SRs. French firms did not perform well in sustainability disclosures and obtained the second lowest mean average score in the group of developed countries. In annexure 1 overall GRI scores of 20 sampled firms jumped from 57.01 in 2008 to 119.86 in 2017. The categorical investigation shows French firms are more focused on environmental disclosures as compared to social and economic disclosures. While GRI_302, GRI_305, and GRI_306 were among the top three reported GRI parameters, GRI_407, GRI_206, GRI_401 were least reported. The linguistic analysis suggests that French firms have adopted a balanced approach to sustainability as the most frequently used words represent all three aspects of the TBL approach. Those words are found in the Figure a Figure 2. Worldscloud Depicting Sustainable Development in top ten economies, own's work 2f wordcloud and include million, water, emissions, management, performance, sustainability, co2, data, group, and solutions. Italy: The number of Italian firms filing SRs to the GRI has been continuously increasing. According to the GRI database, only 48 firms had filed SRs in 2008 but it increased at CAGR of 8.44% and in 2017, a total of 108 firms had filed SRs. Italian firms also performed below average in sustainability disclosures and scored the lowest mean average score of the developed countries examined in this study. The total overall GRI scores of 20 sampled firms jumped from 55.31 in 2008 to 108.79 in 2017 as shown in annexure 1. The categorical investigation shows Italian firms are more focused on social disclosures than environmental or economic ones. GRI_303, GRI_405, and GRI_406 were among the top three reported GRI parameters while GRI_206, GRI_403, GRI_412 were least reported. A possible reason for this may be isomorphic pressure from civil society throughout the world. The linguistic analysis shows that Italian companies give equal weight to all aspects of the TBL approach to reporting. The most frequently used words in the reports of Italian firms are shown in the Figure 2g wordcloud. These include performance, management, employees, production, environmental, indicators, emissions, company, local, and sustainability. Brazil: The number of Brazilian firms filing SRs to GRI has been increasing. According to the GRI database, only 87 firms had filed SRs in 2008 but it increased at CAGR of 10.77% and in 2017, a total of 242 firms had filed SRs. When compared to the firms of developed economies, Brazilian firms did not perform well; however, among emerging economies Brazilian firms performed the best, obtaining the highest mean average score. The total overall GRI scores of 20 sampled firms jumped from 36.59 in 2008 to 79.29 in 2017 as shown in annexure 1. The categorical investigation shows Brazilian firms have an approach focused on social disclosures as compared to environmental and economic disclosures. GRI_402, GRI_405, and GRI_406 were among the top three reported GRI parameters, while GRI_407, GRI_206, GRI_401 were least reported. Linguistic analysis also supports the categorical analysis as five out of ten words are related to the environment. Figure 2h shows the most frequently used words in reports filed by Brazilian firms which include year, ensure, use, area, sea, air, social, environment, risk, and emission. India: The number of Indian firms filing SRs to GRI has been growing. According to the GRI database, only 24 firms had filed SRs in 2008 but it increased at CAGR of 29.28% and in 2017, a total of 313 firms had filed SRs. Indian firms have shown the average result in the scoring of SR practices. The total overall GRI scores of 20 sampled firms jumped from 34.89 in 2008 to 73.76 in 2017 as shown in annexure 1. The categorical investigation shows that GRI_302, GRI_305, and GRI_306 were among the top three reported GRI parameters while GRI_407, GRI_206, GRI_401 were least reported. The above results are also supported by the linguistic analysis of the SRs of Indian firms. As shown in the Figure 2i wordcloud, Indian firms most often used words related to the environmental category. The most frequently used words include water, business, waste, energy, development, emissions, performance, indian, sustainable, and products. The inclination of Indian firms towards the environmental aspects of reporting is most likely due to mandated reporting overseen by regulatory authorities. Traditionally the Indian approach to sustainability was cultivated through benevolence towards society but these days a western approach (i.e. sustainability leads to improved profitability in the long run) is increasingly adopted as Indian businesses more fully participate in the global economy (Balasubramanian et al., 2005). China: The number of Chinese firms filing SRs to the GRI has grown during the period under study. According to the GRI database, only 28 firms had filed SRs in 2008 but it increased at CAGR of 40.05% and in 2017, a total of 813 firms had filed SRs. Though there was an increase in reporting, Chinese firms did not perform well in sustainability disclosures and scored the lowest mean average score. The total overall GRI scores of 20 sampled firms jumped from 30.63 in 2008 to 70.07 in 2017 as shown in annexure 1. The categorical investigation shows Chinese firms have a focus on the social aspects of disclosures rather than environmental and economic aspects. GRI_301, GRI_206, GRI_305 were among the top three reported GRI parameters while GRI_302, GRI_306, and GRI_412 were least reported. The linguistic analysis of the SRs of Chinese firms mirrors the findings of the categorical analysis, showing an inclination towards language related to the social aspect of SRs. The most frequently used words in Chinese SRs are represented in the Figure 2j wordcloud and include csr, service, company, system, urban, area, ensure, safe, access, and key. Many aspects of the government's reporting policies in China are voluntary and indicative (Marquis and Qian, 2014). These SR policies place more of an emphasis on environmental issues (i.e. GRI_301, GRI_305) than social issues, especially human-rights (i.e. GRI_412), as indicated in the categorical analysis.

Conclusion
This study has shown how firms in different nations utilize different SR practices as well as identifies differences in the disclosure of various aspects of the GRI Standards guidelines. By utilizing TM, this study was able to identify the words most often used in the writing of SRs, providing insight into what issues both companies and countries find most important. While this study identified differences across countries, it also showed that all countries have improved their sustainability performance over time (i.e. 2018-2017). This study also ranked the SR practices of sampled economies after classifying them into developed economies and emerging economies. Canada ranked highest among the developed economies, followed by the UK, Germany, US, Japan, France, and Italy. In the emerging economies category, Brazil led the way followed by India and China. China is a particularly interesting case as Chinese firms received the lowest ranking in SR scoring, but China simultaneously had the highest CAGR, 40.05%. This would seem to suggest that Chinese firms are concerned with sustainability. A general observation across all countries is that reporting related to economic issues lags behind reporting on environmental or social issues. One possible reason for this may be the publication of separate financial statements in the form of annual reports as well as an understanding among firm management that sustainability reports are mostly for socio-environmental purposes. There are some limitations to this study. It is focused on the SR practices of the largest countries and ignores the contributions of smaller countries on this issue. Considering them together might lead to a fuller understanding of global sustainability reporting practices. Another limitation is this study's focus on only the 20 largest firms in the countries examined. This opens up the possibility that the practices of medium-and small-sized enterprises might change the country rankings obtained in this study.