Exploring the quality of social information disclosed in non-financial reports of Croatian companies

Abstract By enacting the provisions of Directive 2014/95/EU and the Croatian Accounting Act on disclosing non-financial and diversity information, companies of public interest registering 500 and more employees are required to disclose non-financial information. The purpose of this research is to assess the quality of disclosed social information in non- financial/sustainability reports of Croatian companies. The assessment of the social information was grounded on the framework defined by globally accepted sustainability reporting standards by assessing the quality of social sub-categories of human rights, labour practice, community/society and product, measured by attributes of relevance, clarity, verifiability, comparability and clarity. With the overall quality score of 13.16 (out of possible 36), the results prove that Croatian companies do disclose certain social information, but the reliability of this information for benchmarking and competitiveness assessment is questionable, as a consensus on the minimum of information to be disclosed as a fundamental requirement for benchmarking has not yet been reached.


Introduction
The provision of Directive 2014/95/EU on disclosing non-financial and diversity information was enacted in the Croatian Accounting Act (CAA) on 1 January 2017, requiring all companies registering 500 and more employees to disclose non-financial information in any of the forms of non-financial disclosures (sustainability reports, environmental reports, annual reports, social reports, etc.). In doing so, companies are to rely on any of the national (if existing), supranational (Kollman & Prakash, 2002, p. 48 (EC 2014, para. 9;GRI 2014aGRI , 2014b). Yet, the existing regulation defines neither the minimum nor kind of non-financial information to be disclosed,

Theoretical background
With the abundance and diversity of frameworks dealing with sustainability reporting, organisations sometimes find it difficult to determine which one to report by, particularly as this segment of accounting is not yet developed and precisely defined either in theory or practice. These frameworks vary in their purpose, and social information is usually emphasised as just one part of the triple bottom line approach to sustainability. Some authors have tried to classify these standards according to their features (Choi, 2003;Perrini, 2005) and assess them according to the dimension they focus on (Perrini, 2005). Regardless of their features, it is very notable that almost all of the frameworks analysed have included a social dimension of sustainability. In fact, many of the non-financial reports contain in their titles the term 'social' while reporting on environmental and economic dimensions within the same document.
The environmental dimension of sustainability has been considered as a clearer and less complicated dimension to report on than the social dimension, as environmental science is considered 'a useful guide to understanding the sustainable relationship of humans to nature' (Herriott, 2016, p. 169). However, the social dimension is founded in the concept of a modern corporation as a legal form created by society; it 'has a responsibility to conform to the laws of the society', and a key concept in defining a material aspect framework for this dimension is the stakeholder (Herriott, 2016, p. 164). A stakeholder is defined as a person, group or organisation that has an interest or concern in an organisation, and can affect and/ or be affected by the organisation's actions, objectives and policies (BusinessDictionary. com, n.d.). Most common stakeholders considered in non-financial reports are customers, employees, the local community, and nation-state (Herriott, 2016, p. 174).

Defining social impact
Measuring and consequently reporting on organisations' impact poses a practical challenge for several reasons. It should be noted that the definition of (a social) impact is unclear, and that impact upon society is difficult to measure and quantify (Maas, 2009, p. 47).
Organisations can have a positive or negative impact upon society through many components, resulting in difficulty in linking an organisation's activities to its impact (Maas, 2009, p. 47). Consideration should be given to sustainability frameworks focusing on different stakeholders, ranging from just one stakeholder (e.g., staff or customer) to a broad stakeholder focus (Perrini, 2005). In the context of sustainability reporting, disclosing non-financial information on corporate social responsibility (Herriott, 2016 , p. 170) should take into account the needs of stakeholders as a target group for specific information.
In the subsequent paragraphs attention will be given to the definition of the term 'social impact' and how it has evolved, and in doing so the research of Maas (2009, p. 2) will be followed. The idea of researching the development of the term 'social impact' stems from the chronological development of sustainability accounting that took place throughout several decades (1971-1980; 1981-1990; 1991-1995/2000; 2001 onwards) and the focuses on different stages of its development, either on the social or environmental dimension (Fifka, 2012(Fifka, , p. 66, 2013Hahn & Kuhnen, 2013, p. 5;Mathews, 1997, p. 483;Ortas & Moneva, 2011, p. 19).
Chronologically, as a first definition of a social impact Maas introduced the one provided by Latan e (1981, p. 343), defining it as 'any of the great variety of changes in physiological states and subjective feelings, motives and emotions, cognitions and beliefs, values and behaviour, that occur in an individual, human or animal, as a result of the real, implied, or imagined presence or actions of other individuals', as a broad one, while in 1986 Freudenburg defined it as 'impacts, effects or consequences, that [are] likely to be experienced by an equally broad range of social groups as a result of some course of action' (as cited in Maas, 2014, p. 2), bringing into focus social groups and actions that affect those social groups.
The definition of social impact provided in 1996 by Burdge and Vanclay (1996, p. 59) is somewhat broader, with social impact viewed in the context of ' … consequences to human populations of any public or private actions that alert the ways in which people live, work, play, relate to one another, organise to meet their needs and generally act as a member of society; cultural impacts involving changes to the norms, values, and beliefs that guide and rationalise their cognition of themselves and society'. In 2000 this subject was addressed by Emerson et al. (Emerson, Wachowicz, & Chun, 2001) who defined social value as that which '? is created when resources, inputs, processes or policies are combined to generate improvements in the lives of individuals or society as a whole', whereas Gentile later defined social impact as '? wider societal concerns that reflect and respect the complex interdependency between business practice and society', (as cited in Maas, 2009, p. 2).
A comprehensive approach to social impact can be seen in the research conducted by Vanclay (2003, p. 8), when the definition became more dynamic than previous ones, by recognising the impact as ' … changes to one or more of the following': people's way of life, their culture, their community, their political systems, environment, their health and well being, their personal and property rights, their fears and aspirations (Vanclay, 2003, p. 8;Vanclay, Esteves, Aucamp, & Franks, 2015 , p. 2). This approach to understanding social impact encompasses its complexity by recognising many aspects when dealing with society, such as: how people live, work play and interact with one another on a daily basis; the cohesion, stability, services and facilities of their community; the extent of the participation in decision-making that affects their lives; levels of democratisation that are taking place; the quality of the air and water people use; the availability and quality of the food people eat; adequacy of sanitation; a state of complete physical, mental, social and spiritual well being of people, not just an absence of disease; violation of civil liberties; their perception of fears and aspirations for their future.
In subsequent years, social impact was recognised as a portion of the total outcome that happens as a result of the ventures' activity, above and beyond what would have happened anyway (Clark, Rosenzweig, Long, & Olsen, 2004, p. 7); the difference between what would happen with a given action, and what would happen without it (IAIA, 2009in Reeder & Colantonio, 2013; or the extent to which that change arises from the intervention (GECES Sub-group on Impact Measurement, 2014, p. i). It is the reflection of social outcomes as measurements, both long and short term, adjusted for the effects achieved by others, for effects that would have happened anyway (deadweight), for negative consequences (displacement), and for effects declining over time (drop-off) (GECES Sub-group on Impact Measurement, 2014, p. vii;Social Impact Investement Steering Group, 2014, p. 27).
Social impact is the reference of four key elements: the value created as a consequence of someone's activity, the value experienced by beneficiaries and all others affected, an impact that includes both positive and negative effects, and an impact that is judged against a benchmark of what the situation would have been without the proposed activity (The Organisation for Economic Co-operation and Development, 2015, p. 3) or the sum of or a net of synergies of positive or negative, intended or unintended, financial and non-financial outcomes (Migliavacca, 2016, p. 2).
In the timeframe of about 20 years from the time Latan e first defined social impact in 1981 until, and even after, Vanclay provided a deeper insight into the meaning of social impact in 2003, many authors seem to have taken into account only a fragment of the social impact. This could be because of the complexity of this term, as well as not having a consensus in understanding what a social impact is and what its boundaries are. These definitions of social impact suggest a bias in the available sustainability reporting frame-works/standards, particularly as sustainability is a matter that is to be managed at the company level, for each company's social and natural environments are unique, and so are their impacts. Considering this and the inconsistency present in non-financial reporting, the authors in their analysis of non-financial reports viewed a social impact as previously defined by Latan e and by Vanclay.
The understanding of social impact as a term has clearly been evolving through the decades. Nevertheless, the definition of social impact by Latan e, in recognising physiological states, subjective feelings, motives, emotions, cognitions, beliefs, values and behaviours that can occur in humans or animals, seems to have laid a foundation for the forthcoming interpretation of this term. However, interpretations of social impact in the decades that that followed gradually recognised the role of organisations in the occurring changes and acknowledged the complexity of social impact as a sum or net of synergies of positive or negative, intended or unintended, financial and non-financial outcomes (Migliavacca, 2016, p. 2). When reporting on non-financial information, most organisations follow some form of sustainability framework (guidelines, principles or standards) (Freundlieb, Gr€ auler, & Teuteberg, 2014, p. 23), and in doing so, they tend to disclose mostly qualitative information (Jones, Comfort, & Hillier, 2006, p. 13;Jones, Hillier, & Comfort, 2014, p. 339), which is disclosed as textual description, or expressed as absolute numbers or percentages measured against the very same measure (Per si c & Halmi, 2016 , p. 198). Drawing on the concept of social impact defined by Vanclay, for the purpose of this research any recognition of interaction of the organisation with society was deemed as social-impact information that was further considered in the content analysis.

Measuring social impact
It is possible to measure direct and indirect impact (World Business Council for Sustainable Development, Corporation, & International Financial Corporation, 2008, p. 41). Direct impact is largely within an organisation's control, such as inputs and outputs resulting from the day-to-day activities of an organisation, whereas indirect impacts are not within the organisation's control, but are within the organisation's influence and may include creation of jobs within the supply chain, change in the quality of life for the consumers who buy a product or service, or it can also be viewed as the additional value derived by other firms (small and large) that deal with the company. Diverse methods have been used over a period of time for measuring social impact. Since 2010 Tools and Resources for Assessing Social Impact (T.R.A.S.I.) has been providing support to organisations for measuring social impact (Tools and Resources for Assessing Social Impact, n.d.-a). Today, in the T.R.A.S.I. database there are available over 150 frameworks developed by many institutions and organisations with the purpose of measuring social impact, classified as (1) best practice (assessing and building the organisation's capacity, charting impact, cluster evaluation, due diligence framework for scaling initiatives, evaluating development cooperation … ); (2) methods (appreciative inquiry approach, balanced scorecard, base of pyramid impact assessment framework; BluePrint 1.0, building a performance measurement system, KaBOOM method, GRI Reporting Framework.); and (3) tools (assessment and improvement indicators, B.A.C.O. ratio, beneficiary perception report, benefit-cost ratio, efforts to outcomes software, operational benchmark report, political return on investment … ) (Tools and Resources for Assessing Social Impact, n.d.-b).
In general, methods for measuring social impact can be recognised as qualitative methods (e.g., storytelling, content analysis, interviews), and quantitative methods (e.g., process, impact and monetarisation method), used for purposes of screening, monitoring, reporting and evaluating, with different orientation (input and output) and time perspective (prospective, ongoing and retrospective) as a base for short-and long-term decision-making on the micro (individual), meso (corporation) and macro (society) level (Maas, 2014, pp. 51, 10).
Qualitative methods deal with finding answers to the 'why' and 'in what way' questions, gathering data and finding answers while being exposed to the observed phenomenon through observation, interviews (in-depth, structured, unstructured, face-toface, telephone), content analysis, and other qualitative research methods (Sekeran & Bougie, 2013; Tkalac Ver ci c, Sin ci c Cori c, & Polo ski Voki c, 2010, p. 22). The assumption of qualitative methods is a socially constructed reality; variables are complex, interwoven and difficult to measure; the purpose is interpretation and description of experiences, gaining knowledge and understanding perspectives of others (Burns, 2000, p. 391). Qualitative methods are focused on a small sample, for such an approach allows understanding and deeper analysis of the phenomenon that is the focus of the research (Tkalac Ver ci c et al., 2010, p. 22), with data analysis done by themes determined by the researcher, and with descriptive reporting on the phenomenon (Burns, 2000, p. 392).
These and other attributes of qualitative methods are well depicted in reporting on social impacts, which are related mostly to non-metric measures, such as textual interpretation of the impact an organisation has had on the society, very often including information about charity work, volunteering, or donating for a certain cause in society (providing financial means for building playgrounds for children or donating for the education of underprivileged children and adults). In sustainability reports, these methods are very common, and along with other information concerning social (and environmental) aspects, even interviews with employees can be found. Excessive use of qualitative methods in non-financial reports can leave the impression of a story-telling publication, rather than of a report that is to be used in decision-making. Qualitative methods do not require actual, numerical data on the impact, and can be used for indicating positive or negative social impact (Maas, 2009, p. 150).
On the other hand, quantitative methods are fact-driven with objective reality, variables can be measured and identified, and the purpose is to predict, generalise and provide causal explanation (Burns, 2000, p. 391), and not to provide deeper insight into the results obtained (Tkalac Ver ci c et al., 2010, p. 22-23). Quantitative methods include testing and measuring, providing statistical reporting (Burns, 2000, p. 391-392) requiring large numbers of observations in order to provide statistical power and unbiased research (Tkalac Ver ci c et al., 2010, p. 22). These attributes of quantitative methods are evident in reporting on both environmental and social information, with social information being rather descriptive in nature, unlike environmental information (Chaplin-Kramer & Green, 2016 ;Herriott, 2016, p. 52). Nonetheless, some management tools (Rigby, 2015) have been recognised as quantitative methods for measuring social impact as related to metric measures, and have been recognised and classified according to their purpose, time perspective, orientation, time frame, beneficiaries, and approach.
Quantitative methods require a lot of data, demanding lots of time and resources. In her research on defining quantitative methods for measuring social impact, Maas defined some 30 tools, but also came to the conclusion that although those tools are applicable, they are not used for reporting purposes (Maas, 2009(Maas, , pp. 58-59, 60, 2014, which could be attributed to the nature of the quantitative methods and data, requiring lots of time and resources (Dagilien & Gokiene, 2011, p. 26;Todorov & Marinova, 2009, p. 1218. This variety of quantitative methods or tools provides an opportunity to approach social issues in the way most suitable to the company's needs for monitoring and managing, as well as to the outside pressures for disclosing social issues. While several authors have tried to classify the guidelines, principles and standards for reporting on sustainability issues (Choi, 2003;Perrini, 2005) according to their features, and have classified them as aspirational principles and codes of practice, management system and certification schemes, rating indices, accountability and reporting framework (Perrini, 2005), Maas (2009, p. 51) did not include any of the guidelines, principles and standards in her analysis, but did come to the conclusion that qualitative and quantitative approaches combined ought to form a basis for developing social impact measurement, as the characteristics and purpose of the results dictate whether 'a full quantification of social impact' (Kucukusta, Mak, & Chan, 2013) is needed and consequently worthwhile. In general, social impact has been recognised as the synergy of key areas of sustainable development, for example nature (environmental benefits and impacts), society (community impacts and involvement), economy (financial health and economic influence) and well being (effect on individual quality of life), as a starting point in preparing and presenting sustainable reports based on the GRI framework.
A conclusion can be drawn that these methods and tools should be intertwined with the globally accepted guidelines, principles and standards (e.g., GRI, AA1000, SA8000, and ISO 26,000). Based on the research of several authors and with the statistical approach to viewing the data, measures have been described as (1) categorical measures, which place an entity into one of several defined categories which are not related to each other along any continuum (gender, race, occupation … ), (2) ordinal-scaled measures, sequencing categories in value, with no meaningful concept of distance between categories (corporate rankings, rating firms … ), (3) interval-scaled measures, having no meaningful zero point, yet differences between scores are meaningful, meaning that intervals of measurement have a meaning (an increase in temperature .), and (4) ratio-scaled measures, having a meaningful 'zero' point, making it possible to calculate meaningful ratios and proportions, and making it possible to track improvement in meaningful statements, or ratios (percentage of improvement per year … ) (Hair, Black, Babin, & Anderson, 2009, p. 3;Herriott, 2016, pp. 77-78;Horvat & Mijo c, 2012, p. 22, 26;So si c, 2006., p. 7). These viewpoints were taken into consideration in the process of analysing non-financial reports disclosed by Croatian companies, and in assessing the quality of the non-financial information disclosed therein.
In dealing with social issues, the GRI G4 Guidelines (2014b) divides social category into sub-categories: (1) labour practice and decent work, (2) human rights, (3) society and (4) product responsibility, listing indicators for each of the sub-categories. Another globally recognised sustainability standards is the Sustaianbility Accounting Standards Board (S.A.S.B.), in which the S.A.S.B. recognises sustainability issues (environment, social capital, human capital, business model and innovation, leadership and governance) which can be traced and measured through specific indicators (human rights and community relations, access and affordability, customer welfare, data security and customer privacy, fair disclosure and labelling, fair marketing and advertising) (Sustaianbility Accounting Standards Board, n.d.). Just a glimpse of the approach to social impact provided by these two globally accepted standards highlights the question of different views and approaches to this matter.

Framework, methodology and research sample
Much of the initial research (Chauvey, Giordano-Spring, Cho, & Patten, 2015;Freundlieb et al., 2014;Hąbek & Wolniak, 2015) on non-financial information defined and measured primarily qualitative characteristics from the point of the F.A.S.B. and I.A.S.B., internationally recognised accounting bodies for financial reporting (Freundlieb et al., 2014); inferential statistics was used to shed further light on the research problem.

Research framework
In defining the research framework and methodology it is of particular importance to consider the current global standards relevant in sustainability reporting, with an emphasis on the current GRI G4 Guidelines and announced GRI Standards. Namely, the Global Sustainability Standards Board was established in 2014 with an assignment to create GRI Standards, which are still being developed and will be effective on or after 1 July 2018 (GRI, n.d.). The G4 Guidelines are to be used until then, for which reason the concept of the research was grounded in the GRI G4 Guidelines (GRI, 2014b, pp. 84, 143, 173, 198, 221). Michelon et al. (2014, p. 65, 69) recognise the dependency of quantity of information disclosed with the quality of non-financial disclosures, providing among other disclosure variables mathematical formulae for both. Hąbek and Wolniak (2015, p. 415) have discovered a positive effect of legal obligation on the quality of nonfinancial reporting. As there is no legally prescribed form for disclosing sustainability information or defining the type and minimum of information to be disclosed, after determining the quality of social information, a correlation between the quantity of disclosed social information and the number of disclosed non-financial disclosures will be tested with the purpose of detecting an interrelationship between these two. The aim is to determine whether the number of (publicly) disclosed non-financial disclosures affects their quality, assuming that organisations bound by regulation only to disclose eventually do improve the quality of their non-financial disclosures and practices.

Research methodology
GRI G4 Guidelines (2014a, p. 43) divide the social category into the following four subcategories: labour practices and decent work, human rights, society, and product responsibility.
Although they established the reporting attributes for non-financial information, in this research the GRI G4 Guidelines were not considered for determining the attributes of the disclosed non-financial information because of their complexity, that is, each principle being defined by a set of tests. For this reason similar research and examples of best practice (Chauvey et al., 2015;Hąbek & Wolniak, 2015) will be used in defining the attributes; hence these results can be observed in relation to the results of research conducted in France on the sample of companies listed in the French S.B.F. (Chauvey et al., 2015, pp. 794-795).
The indicators searched for were those provided by the GRI G4 Guidelines. The research method used was a content analysis, with the purpose of recognising the presence of social sub-categories in the non-financial reports as defined by the GRI G4 Guidelines.

Research sample
The approach to the research conducted in France can be applied as a guideline in designing a model for assessing the quality of non-financial reports disclosed in Croatia by companies registering 400 and more employees, because under Directive 2014/95/EU Croatian companies have to publicly disclose non-financial information. The Directive is primarily focused on large undertakings, groups and big companies registering 500 and more employees (EC, 2013(EC, art. 2, 2014. With these starting points as guidance, this research was focused on companies in Croatia registering 400 and more employees in 2014. The motive for including companies with 400-499 employees in the sample was a possibility of their growth in forthcoming years. The purpose of analysing the 2014 sustainability reports was to assess the starting point of Croatian companies in the year of the enactment of Directive 2014/95/EU, so that further research can track the impact of the Directive on the non-financial reporting from the time of its issuing and becoming mandatory in all Member States by bringing into force laws and regulations regarding non-financial reporting. The list of companies was retrieved from Croatian Company Directory (C.C.D.), a website provided and maintained by Croatian Chamber of Economy (Croatian Chamber of Economy, n.d.). After obtaining the list, company websites were searched for non-financial reports, and when not found a search via Google search was performed using predefined keyword phrases.
The unit of analysis was any form of non-financial disclosure (annual reports, social reports, environmental reports, etc.), and the method employed was content analysis. The focus of the research was non-financial reports published for year 2014, mainly due to these being publicly available in the second quarter of the financial and calendar year in Croatia, making non-financial reports for 2015 unavailable at the time of collecting the reports, and also due to companies deciding their own pace of publishing these reports, which could be every year, every other year, or could even change in the meantime. As many of the retrieved companies were a member of a group of undertaking, and by the Directive 2014/95/EU non-financial reports can be prepared by a parent undertaking and still be valid for a member of a group (EC, 2014, paras. 14, 15), it was necessary to perform coding of the companies, which resulted in a total of 221 companies retrieved and only 38 reports available for analysis.

Research results and discussions
The aim of the research was to identify and to record social topic subcategories, that is, human rights, labour practice, community (society), and product. The assessment was performed by awarding one point for the identified presence of each of the attributes defined for the subcategories. The results of each report were then summed up and presented in a matrix table. The sub-category scores could range from 0 to 9, reflecting the nine assessed attributes, and are presented in the final row, whereas the attributes' scores could range from 0 to 4, reflecting the four assessed sub-categories, and are presented in the final column.
Information pertaining to social topic sub-categories has been assessed with considerably lower score, ranging from 2.95 the lowest to 3.74 the highest (out of maximum 9). In a way, this proves that the business world does not yet have a sufficiently complete science of social sustainability, and its methods for evaluating social impact are still underdeveloped (Herriott, 2016, p. 189), thus affecting the quality of social information disclosed in sustainability reports. The sub-category under social topics with the highest score is labour practices and decent work, with 3.74 (out of 9), whereas the sub-category with the lowest score is product responsibility with 2.95 (out of 9). Ranking the scores in a descending order indicates the importance of social information in companies registering 400 and more employees and disclosing non-financial information in Croatia, which are: labour practices and decent work (3.74); community / society (3.34); human rights (3.13); product responsibility (2.95).
The most common information disclosed regarding labour practices and decent work is related to categorical data, such as gender representation, number of junior and senior positions, and so on. When referring to community/society, the most common information was categorical, providing descriptions of actions taken and charities. Human rights (3.13) and product responsibility (2.95) are the social sub-categories with the lowest scores. The reason for this could lie with underdeveloped system of measuring, recording, monitoring and reporting of data related to human rights and product responsibility. It is interesting to note that content analysis highlighted that much of the presented information is qualitative, either disclosed as absolute numbers or percentages, or simply described in the text.
Regarding the assessed non-financial information attributes, the scores are not high either, with the highest score referring to relevance and identification of a stakeholder, scoring 2.39 out of maximum 4.00, and the lowest referring to clarity, with 0.68 out of maximum 4.00. This could be explained through the expectations set by the GRI guidelines and standards in relation to identification and clarity of methods Customer health and safety, product and service labelling, marketing communications, customer privacy, compliance (PR), Source: Authors based on (GRI, 2014b, pp. 143,173,198, 221). The framework of the sustainability report Comparison with previous/next-period sustainability report Several reporting periods in a sample sustainability report (or possibility of comparing the presented information with the information presented in the sustainability report from the previous/upcoming reporting period) Clear definition of the report perimeter and boundaries -Verifiability Verifiability of information presented Different knowledgeable and independent observers could reach consensus that a particular depiction is a faithful representation.

Clarity Clear definition of presented data and indicators
Exact indicators that are being monitored, measured and reported on Explanation about the methods of elaboration, calculation, and/or reporting mechanisms.
-Source: Authors based on Chauvey et al., 2015, p. 793 when disclosing non-financial information. The stakeholder inclusiveness tests include description of the stakeholders to whom the company considers itself accountable, drawing of the report content upon the outcomes of the stakeholder engagement process specifically for the report and for the binding legal and institutional framework, and finally informing that topics included in the disclosure are material (GRI, 2016, p. 8).
As the tests identify any relationship related to the stakeholders, this relatively high score of relevance and identification of stakeholders is self-explanatory. On the other hand, for the purposes of this research, clarity was defined as providing explanation about the methods of elaboration, calculation, and/or reporting mechanisms (Chauvey et al., 2015, p. 793), none of which was identified, probably due to still being underdeveloped. The assessment of the attributes of social information point to a lower score of those attributes requiring a metric system, such as parameters needed for comparability within an organisation, industry and perhaps other industries (ranging from 1.37 to 1.24), verifiability of the information presented (1.50), clarification of the methods employed in calculating and disclosing the information (0.68), and analysis of risks factors specific to the organisation (0.95). The overall score of the quality of the non-financial reports is considerably low as well, 13.16 out of maximum 36. With many of the attributes revolving around the social impact metric system, most of those attributes were assessed rather low, thus affecting the overall quality score.
Regardless of the approach to the assessment of non-financial information which can be found in many studies (Bachoo, Tan, & Wilson, 2013;Chauvey et al., 2015;Hąbek & Wolniak, 2015;Michelon et al., 2014), these results point to a low quality of disclosed non-financial information and reports. The average score of quality of information in the research is also rather low, scoring 13.16 out of 36. Such low scores suggest that 'having and reporting data on sustainability is just the first step for a company that needs or wants to manage itself effectively on the long run' (Herriott, 2016, p. 213), and that existing measurement tools for social topics are not adjusted to, but can be useful for, non-financial reporting (Maas, 2014, p. 60), and might even improve it to a certain extent.
Data presented in Table 4 indicate that the strongest correlation between attributes and sub-categories is between clarity, as an attribute, and community, as a sub-  .000 .000 .000 .000 .000 .000 .000 Source: Authors' processing.
ECONOMIC RESEARCH-EKONOMSKA ISTRA ZIVANJA category (r ¼ 0.896, n ¼ 38, p ¼ 0.000), suggesting that most information disclosed was related to the community and that it was clearly defined in non-financial disclosures according to the companies' needs. The weakest correlation was between attribute of relevance and sub-category of product (r ¼ 0.724, n ¼ 38, p ¼ 0.000), suggesting that there is the least amount of disclosed information related to the product, and its relevance is questionable. Further analysis of correlation within sub-categories suggests the information disclosed in the sub-category human rights seems to be relevant (r ¼ 0.848, n ¼ 38, p ¼ 0.000), yet its verifiability (r ¼ 0.749, n ¼ 38, p ¼ 0.000) is somewhat questionable; information disclosed in sub-category labour practice seem to be clear for the reader to understand (r ¼ 0.880, n ¼ 38, p ¼ 0.000), but its relevance also seems questionable; information disclosed in the sub-category community/ society seems to be clear (r ¼ 0.896, n ¼ 38, p ¼ 0.000) to the understanding of the stakeholder, but its verifiability (r ¼ 0.802, n ¼ 38, p ¼ 0.000) is vague; information disclosed in subcategory product is relatively comparable (r ¼ 0.806, n ¼ 38, p ¼ 0.000), but its relevance (r ¼ 0.724, n ¼ 38, p ¼ 0.000) is arguable. These results of low correlations with verifiability and relevance suggest that information disclosed in nonfinancial disclosures is susceptible to the needs of the company's management in disclosing favourable information. This further points to the need for a much stronger initiative in both regulation and best practice in defining the form and a minimum of information to be disclosed in non-financial disclosures. This low score of quality of non-financial information led to a question of what can positively affect the quality of non-financial disclosures. Since this subject is in its infancy in the areas of research, regulation and practice, we further tried to discover whether the companies themselves could improve the quality of non-financial reporting. Research (Bonilla Priego & Aviles Palacios, 2008, p. 390) undertaken on hotels in Spain has shown a correlation between the number of years of owning an environmental certification and the quantity of disclosed environmental information. This research highlighted the question of whether the amount of disclosed non-financial information is correlated with the quantity of social information.
A correlation analysis revealed that there is no significant correlation between the amount of disclosed non-financial disclosures and the quantity of disclosed social information therein, meaning that the experience gained through the amount of disclosed non-financial disclosure does not improve the quality of disclosed social information. When interpreting these data and comparing them with research done by Bonilla Priego and Aviles Palacios (Bonilla Priego & Aviles Palacios, 2008), one must bear in mind that environmental information has been quantified and, due to the development of natural and environmental sciences, can be captured on all four measurement scales previously described. This will further enable environmental systems such as E.M.A.S., on which Bonilla Priego and Aviles Palacios have based their research, to define a minimum and type of information to be disclosed in environmental disclosures that are to be certified, guaranteeing a high quality of disclosed environmental information. This leads to an issue of recording and measuring social information.
In the process of developing measurement methods suitable for reporting social nonfinancial information, it is recommended to work on creating indicators, as suggested by Herriott, for the classification of measurement methods and information, classifying categorical-scaled, ordinal-scaled, interval-scaled, and ratio-scaled measures by which social information can be reported. The authors expected to find more indicators that could be easily identified and comparedmeaning they would be numerical in their nature, regardless of them belonging to interval-or ratio-scaled measures but indicators from categorical-or ordinal-scale measures were prevalent. This affected the approach to analysing non-financial reports, so that only a binary coding could be applied to this research, assigning 1 for present and 0 for non-present indicators. This has also altered the scores of the respective attributes analysed, thus affecting the overall score of quality of the disclosed social information.
Nonetheless, a prerequisite for any of this is providing a common definition of social impact as a term, so that globally accepted guidelines and standards dealing with the social dimension of the triple bottom line can set the framework for reporting on social information.
A theoretical discussion on social impact suggests a wide range of issues that an organisation might tackle while defining the impact it has in the social category. Although indicators for the social category are defined by the GRI G4 Guidelines, sustainability reporting is still grounded in the management perception of relevance of indicators being reported for the reporting period; an organisation might find certain social issues relevant and report on them in one reporting period, while these same issues may be assessed or perceived as irrelevant, or to have been solved, and therefore are not reported on in the subsequent reporting period.
Reporting on social issues is particularly sensitive because of its complexity, involving changes to any or many aspects combined pertaining to the quality of human life, such as people's way of life, their culture, political systems, health and well being, personal and property rights, fears and aspirations, as well as the social aspect of environment impact (water, air, noise … ), physical safety, etc., all of which are intertwined. Some of these aspects have a direct impact and as such are within an organisation's control, whereas some have an indirect impact and are outside an organisation's control; being intertwined, this makes it difficult to determine whether an impact is within or outside an organisation's control. In this regard, quantified indicators might be more suitable than qualitative ones for detecting these borderlines and managing an organisation's impact, consequently highlighting a near urgent need for developing a metric system for social impact(s), starting with the types/ scales of measurement by which information can be reported. In the development process, it might be wise to consider the industry in focus, as social impact is a synergy of key areas of sustainable development that are individual to the industry.

Conclusion
This research has revealed a low level of quality of social information disclosed in nonfinancial disclosures in Croatia. This quality cannot be correlated with the experience in disclosing non-financial information (number of disclosed non-financial disclosures), along with other research (Fifka, 2013) suggesting that the regulation might be a solution to improvement of the quality of social information and thus quality of non-financial information in general.
Another reason for the low level of quality of social information might be underdeveloped metrics and models for recording, tracking, monitoring and, perhaps particularly, reporting. One reason for this may be the vague definition of the term 'social impact'. Another reason might be the complexity of this term, implicating a net of synergies of various kinds of outcomes of an organisation, making it difficult to distinguish between outcomes as a result of organisation's activity reflected in society, and other circumstances causing and/ or influencing the outcome. A suggestion for further research would be to clearly determine the definition of social impact by analysing definitions provided by various authors and organisations, and by analysing organisations' standpoint on defining this term when reporting on social issues. In order to proceed with developing a metric and reporting system for social topics, it is of paramount importance to determine a (unified) definition of the social impact of an organisation on society. Such a definition will provide guidelines to enable organisations in determining the worth of social impact information and the importance of its monitoring and disclosure.
One possible approach for improving the quality of non-financial information is developing and applying the measurement methods dealing with the impact an organisation has on society. In this regard, particular emphasis should be placed on developing measures related to the social topics and impacts of an organisation on society. Another suggestion for further research would be to investigate whether, which, and to what extent the above-mentioned social impact measurement methods are adopted in organisations that are subject to the EU Directive/2014/05/EU, as well as in other organisations that have chosen to disclose non-financial information for reasons other than legislation. Regarding sustainability and accountancy, economic and environmental dimensions of the triple bottom line are indisputably easier to measure and manage than the social dimension, for which the metric system is not only underdeveloped, but also not unified in determining basic terms as a prerequisite for developing a metric system. Essentially, a metric system for social topics should be developed and improved to the level of environmental information, and both of these topics should reach the level of transparency, comparability, clarity, relevance and verifiability of financial information, by defining and measuring reporting principles for defining non-financial report quality as fundamental and enhancing qualitative characteristics (FASB, 2010, pp. 16-21;IASB, 2015 , pp. 27-32), as well as reporting principles for defining non-financial report content in the same way as has been defined for disclosing financial information. The challenge of reporting on the social dimension of sustainability should be viewed in the context of sustainability in general, which requires a high level of harmonisation in accountancy with the aim of bridging the gaps created by a solely financial approach and assessment of management decisions.

Disclosure statement
No potential conflict of interest was reported by the authors.