Environmental photograph use in corporate sustainability reporting: A machine‐supported visual content analysis

Fenk, Lorenz Article — Published Version Environmental photograph use in corporate sustainability reporting: A machine‐supported visual content analysis Business Strategy and the Environment Provided in Cooperation with: John Wiley & Sons Suggested Citation: Fenk, Lorenz (2024) : Environmental photograph use in corporate sustainability reporting: A machine‐supported visual content analysis, Business Strategy and the Environment, ISSN 1099-0836, Wiley Periodicals, Inc., Hoboken, NJ, Vol. 34, Iss. 1, pp. 1097-1112, https://doi.org/10.1002/bse.4035 This Version is available at: https://hdl.handle.net/10419/313796 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ RESEARCH ARTICLE Environmental photograph use in corporate sustainability reporting: A machine-supported visual content analysis Lorenz Fenk TUM School of Management, Technical University of Munich, Germany Correspondence Lorenz Fenk, Technical University of Munich, TUM School of Management, Arcisstrasse 21, 80333 Munich, Germany. Email: [email protected] Funding information Friedrich Naumann Foundation Abstract Despite the prevalence of photographs in corporate sustainability reporting, their use is not yet sufficiently understood. To the best of our knowledge, this paper is the first large-scale study in the field. Introducing a novel machine-supported approach, we assess environmental photograph utilization based on a sample of 45,228 photographs contained in 1,463 separately disclosed sustainability reports from European firms between 2011 and 2020. We find that against the overall trend of decreasing photograph utilization, the share of environmentally themed photographs has markedly increased. Furthermore, operating in an environmentally sensitive industry is strongly associated with a substantially larger share of photographs depicting environmental subject matter. Lastly, we observe that companies signal their superior environmental performance through greater utilization of environmental photographs. By introducing a novel machine-supported approach to analyzing photographs, this study makes a methodological contribution to the field of sustainability reporting. Our results also have important practical implications. KEYWORDS corporate sustainability reporting, environmental, photograph use, photograph use, computer vision 1|INTRODUCTION To counter legitimacy threats arising from environmental pollution or poor working conditions, companies may resort to shaping the perception of their impact on stakeholders and society at large (Cho & Patten, 2012). Deliberate attempts to influence the perception of an organization's legitimacy, e.g., by manipulating content and appearance of information, are helped by the information advantage managers enjoy over the report's readers (Merkl-Davies & Brennan, 2007,2011). Photographs, in particular, are well suited to communicate sustainability-related topics, as complex and less quantifiable messages in the indistinct future are often more effectively conveyed through photographs than numbers or words (Anderson, 1980; García-Sánchez & Araújo-Bernado, 2020; Rämö, 2011). Moreover, visuals enjoy a more prominent place in cognitive memory than Abbreviations: ENV PUT, environmental photograph utilization; ENV share, environmental photograph share; ESG, Environmental, Social and Governance; EU, European Union; PDF, portable document files; PNG, portable network graphics; PUT, photograph utilization; VIF, variance inflation factor. In the realm of corporate sustainability reporting, the role of photographs remains underexplored. This paper introduces a machine-supported approach based on computer vision to examine the use of photographs, and in particular environmental photographs. Analyzing 45,228 photographs contained in 1,463 sustainability reports from European companies, our findings challenge the prevailing notion of sustainability reports becoming ever more densely filled with photographs. In addition, we observe a notable increase in environmental photograph use. Particularly, companies operating in environmentally sensitive industries exhibit a significantly higher proportion of such visuals. This study not only facilitates a novel approach to studying photograph use but also holds practical significance, shedding light on the evolving field of corporate sustainability reporting. Received: 23 January 2024 Revised: 16 October 2024 Accepted: 17 October 2024 DOI: 10.1002/bse.4035 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Author(s). Business Strategy and the Environment published by ERP Environment and John Wiley & Sons Ltd. Bus Strat Env. 2025;34:1097–1112. wileyonlinelibrary.com/journal/bse 1097 written text (Tversky, 1974) and can engender framing effects or thought patterns (Tversky & Kahneman, 1986). Scientific studies have cautioned that photographs are used beyond decorating textual information; however, research has not kept pace with elaborate practices of visual communication employed by the private sector (Breitbarth et al., 2010; Davison, 2015; Davison & Warren, 2009). Previous articles find that photographs are used to symbolically pursue legitimacy by creating idealized visions, which do not typically reflect companies' business conduct (Boiral, 2013; Breitbarth et al., 2010; Hrasky, 2012). More recent studies confirm that managers use their discretion to project the image of a responsible and trustworthy corporate citizen, while negative aspects are all but absent in the visual realm (García- Sánchez & Araújo-Bernado, 2020; Nicolò et al., 2022). However, most existing studies rely on small samples, which often come from a single industry, single country, single reporting period, or case study (Invernizzi et al., 2021; Rämö, 2011). Understanding photograph use is also of great practical relevance, as even seasoned investors are swayed by aesthetics in corporate reporting (Townsend & Shu, 2010). Comparing different types of disclosure, photographs are most extensively employed in corporate sustainability reporting, where they constitute an integral element (Boiral, 2013; Breitbarth et al., 2010; Davison, 2015;Invernizzi et al., 2021;Rämö,2011). The medium of sustainability reports has also gained importance as the salience of corporate sustainability has markedly increased (Arvidsson & Dumay, 2022). Furthermore, the substantial cost of printing and distributing hard copy, photograph-rich sustainability reports have all but evaporated as corporate communication is most usually shared digitally, which may have influenced how managers design sustainability reports. This predilection for using photographs may also be adopted by small- and medium-sized enterprises (de Villiers & Alexander, 2014; Shabana et al., 2017), which are soon to be subjected to expanded reporting requirements in the European Union (European Parliament & European Council, 2022). Given the concerning use of photographs reported in related studies, we argue that it is important to understand their power to sway readers' opinions. Despite their important role and well-documented potency, the use of photographs remains a blind spot in corporate sustainability reporting. Given the inherent limitations to existing studies, we contribute generalizable findings on the use of photographs in European corporate sustainability reporting, which feature both longitudinal coverage and cross-sectional variation. We quantify the overall use of photographs and assess the themes depicted therein. Our machinesupported analysis of photographic content focuses in particular on photographs conveying environmental subject matter to account for the outsized role of the environmental pillar within corporate sustainability. While the significance of environmental concerns within corporate sustainability has been a constant, its relative importance has gained new momentum as evident by the ambitious regulatory agenda pursued by the European Union (EU), an increased general public awareness, e.g., global climate demonstrations, and intensified involvement of the financial community, e.g., ‘Climate Action 100+ group’. In addition, we argue that environmental themes can also be more reliably identified than more abstract concepts, such as labor relations or corporate governance. Our research is thus guided by the following research question: To what extent do companies use environmental photographs in their corporate sustainability report? Moreover, we have identified three supportive research questions that are intended to guide our research and shed light on specific aspects of our study's focus: •How has photograph use changed over time? •How is the relationship between photograph use and sustainability performance? •How does photograph use vary across different industries? We employ Google's Vision API, a commercial pre-trained computer vision service (Google LLC, 2023), to derive sophisticated insights into large image data sets. Advances in artificial intelligencepowered software and computing resources enable researchers to analyze and draw conclusions based on large-scale data sets. While machine-supported software has been put to use in many fields of academia, its application in the context of corporate sustainability reporting is to date largely confined to textual analysis, and recent examples include Niehoff (2022) and Bhandari et al. (2022). In particular, we use the label detection feature to retrieve photographic content by ascribing labels. To classify whether photographs depict environmental subject matter, we developed a codebook based on the detected labels ascribed by the vision models. Our machine-based approach allows for the analysis of much larger image data sets than previously feasible. To the best of our knowledge, this is the first large-scale study on the use of photographs in corporate sustainability reports, which applies a machine-supported visual content analysis. We assess overall photograph utilization and focus our analysis on the use of photographs conveying environmental subject matter. Based on a sample of 45,228 photographs contained in 1,463 separately disclosed reports from 2011 to 2020, we cannot confirm the general belief that sustainability reports are becoming ever more densely filled with photographs. Against this observed trend of falling photograph utilization over the last decade, the percentage of environmentally themed photographs has markedly increased. From our regression models, we find that, firstly, operating in an environmentally sensitive industry (Branco & Rodrigues, 2008; Clarkson et al., 2011) is associated with a substantially higher percentage of photographs depicting environmental subject matter. Secondly, we find that companies with higher environmental credentials signal their superior performance through a higher share of environmentally themed photographs. Both findings are highly significant and very robust with regard to changes in the methodology for identifying environmental photographs. In this study, we introduce a novel approach to assessing photograph use in corporate reporting. Furthermore, we contribute to the existing literature by providing empirical evidence of overall photograph use and, in particular, the utilization of environmental photographs by European companies. The remainder of this article is organized as follows: Section 2 consists of the literature review and the derivation of our research 1098 FENK question. In Section 3, we outline our research approach, which is followed by the empirical results in Section 4. Section 5discusses the findings as well as limitations and suggestions for future research. Lastly, Section 6provides the conclusion of this paper. 2|BACKGROUND 2.1 |Characteristics of sustainability reporting Over the last two decades, sustainability-related topics have increasingly occupied annual reports (Davison & Warren, 2009). Simultaneously, many companies have begun to voluntarily disclose a sustainability report along with their financial reporting. Since 2017, large legal entities domiciled in the European Economic Area have been legally obliged to annually disclose non-financial information to the public. The 2014/95/EU directive (European Parliament & European Council, 2014) mandates the comprehensive discussion of a wide range of topics, ranging from environmental considerations and labor relations to anti-slavery practices. While the law dictates that the report includes certain topics, companies do not need to abide by a strict reporting framework (Fiechter et al., 2022). Most of the sustainability information reported by companies can still be considered voluntarily disclosed information (Christensen et al., 2021). In the absence of a harmonized reporting standard, managers enjoy large discretion over the design of their sustainability reporting. The companies' motivations for disclosing voluntary sustainability reports vary greatly (Herzig & Schaltegger, 2011). These include showcasing the company in a positive light, gaining a comparative edge over competitors (Frias-Aceituno et al., 2014), and/or mitigating stakeholder pressures and interference by government agencies (Gray et al., 1995; Prado-Lorenzo & García-Sánchez, 2010; Reverte, 2009). Legitimacy theory is most commonly used to explain sustainability disclosure by companies. Following Deegan (2002), companies require legitimacy not only from core stakeholders but also seek to repair or maintain legitimacy in the eyes of society, on which they continuously depend for resources and approval (Beattie & Jones, 2008; Hahn & Lülfs, 2014; Milne & Patten, 2002). Legitimacy can be understood as a temporary social contract, which is granted by society to responsibly operating companies (Ashforth & Gibbs, 1990; Hrasky, 2012). Following Suchman (1995), organizational legitimacy can be attained either as moral legitimacy, backed up by the firm's business conduct, or through symbolic attempts to project the image of a caring corporate citizen. As their legitimacy rests on staying within a set of social boundaries (Deegan, 2002), companies may use sustainability disclosure to legitimatize their operations to their stakeholders (Cho & Patten, 2012; Guthrie & Parker, 1989). On the other hand, voluntary disclosure theory and, by extension, signaling theory are also extensively used to explain voluntary reporting of non-financial information to the public (Clarkson et al., 2008; Mahoney et al., 2013). Therefore, companies with superior sustainability performance choose to voluntarily disclose otherwise unobservable sustainability information to increase their financial value (de Villiers & Marques, 2016; Dhaliwal et al., 2011; Hummel & Schlick, 2016). In this context, signaling refers to the attempt to communicate a firm's superior sustainability achievements (Connelly et al., 2011; Rezaee, 2016). Even so, disclosure is costly; firms with superior sustainability credentials benefit from the comparison differentiating them from their peers through voluntary sustainability disclosure (Mahoney et al., 2013). To present these achievements through corporate sustainability reports, managers have a versatile communication toolbox at their disposal. 2.2 |Visual communication Photographs have become ubiquitous in our visualized and digitized world of business. In the context of corporate reporting, the growth of visual material over the last decades has been recognized by several studies (Beattie & Jones, 2008; Davison & Skerratt, 2006; Lee, 1994; Usmani et al., 2020). Comparing different venues of corporate reporting, studies have shown that photographs are especially prevalent in corporate sustainability reporting and constitute an integral feature of such reports (Boiral, 2013; Invernizzi et al., 2021; Rämö, 2011). Many researchers have expressed concern that the use of richer media gives companies a very potent tool, which can be leveraged for rhetorical or persuasive purposes (Cho et al., 2009). Photographs may construct what corporate sustainability means to whom (Breitbarth et al., 2010), divert attention from adverse information (Arora & Lodhia, 2017), or alter the perception of non-financial performance (Cho et al., 2010; Hrasky, 2012; Mahoney et al., 2013; Peeples, 2011). Hopwood (2009) argues that sustainability disclosure filled with visual imagery may obfuscate the true content and thus potentially decrease corporate transparency. High-quality photographs, professionally shot and at times covering an entire report page, are intentionally and deliberately used (Breitbarth et al., 2010; Davison, 2015; García-Sánchez & Araújo-Bernado, 2020). Graphic elements such as photographs naturally stand out from the surrounding text and attract the reader's attention through a so-called ‘pop-up effect’(Treisman, 1985). Richer media seems to be especially well suited when there is ambiguity about the interpretation of the underlying issue (Cho et al., 2009; Daft & Lengel, 1986). Photographs serve several purposes in long text documents. They orientate readers, thus enhancing information processing fluency (Invernizzi et al., 2021). In addition, readers are more likely to select texts with photographs, which enhances the overall attractiveness of reading a text document (Knobloch et al., 2003). On the other hand, visuals may be used to guide readers to more favorable sections of the report (Usmani et al., 2020). The capacity of photographs to communicate complex messages to a diverse audience makes them a very powerful tool for disseminating information (Anderson, 1980; Coleman, 2010). This is especially true compared to textual information, where complex messages may be lost to complicated language (Rämö, 2011). Studies suggest that visuals are not only easier to recall in many situations (Paivio, 1969) but are also better retained over the long-term (Paivio et al., 1968). Individuals also tend to spend more time looking at visual material than texts (Tversky, 1974) and become FENK 1099 biased toward the image if the two media give opposite cues (Zillmann et al., 1999). Furthermore, the capacity of photographs to ‘create reality’ (Preston et al., 1996) makes them especially persuasive, as they are perceived to capture the real world (Pesci et al., 2015). Photographs can be leveraged to misrepresent reality (Caron & Turcotte, 2009), especially in fairly contested domains such as corporate sustainability (Hallin et al., 2021; Meuer et al., 2020). Following Bansal and Kistruck (2006), photographs can therefore be used to create a convincing account of the subject matter irrespective of actual implementation. Photographs also have the capacity to engender thought patterns and engender framing effects to guide the readers' thinking (Tversky & Kahneman, 1986). The strategic and deliberate attempt of firms to positively influence public opinion about their image is called impression management (Hooghiemstra, 2000; Merkl-Davies & Brennan, 2007). Following Davison (2015), corporate sustainability reporting has inherent characteristics that make it ideal for leveraging visual communication. Compared to financial disclosure, it speaks to a much wider audience, including non-professional and less experienced stakeholders (Rämö, 2011). Furthermore, it tends to be forward-looking and more qualitative in nature, stemming from the long time horizon of some of the topics covered. The dissemination of mostly qualitative information also results in less comparable measures of performance (Anderson, 1980;Cho&Patten,2007;Rämö,2011). In light of the substantial discretion reporting managers enjoy over their sustainability disclosure, photographs are a very potent tool used to showcase the company in the most favorable light. The importance of understanding companies' visual reporting behavior is furthermore underscored by the findings of Townsend and Shu (2010), who showed that even seasoned investors are swayed by the aesthetics of corporate reporting. 2.3 |Related work and research gap We argue that it is very important to shed more light on the subject as research has lagged behind the sophisticated practices of the private sector (Davison, 2015). Despite a number of existing studies, the understanding of photograph use in corporate sustainability reporting remains insufficient. As these studies rely on the manual assessment of photographic content, the sample size of the underlying reporting data is inherently limited. While the majority of studies provide a lot of analytical depth, their findings can seldom be generalized beyond the observed phenomenon. Most are based on cross-sectional data from a single industry or country (Boiral, 2013; Cabrera-Narváez & Quinche- Martín, 2021;Invernizzietal.,2021;Rämö,2011;Zengetal.,2022), multiple cases studies (Ali et al., 2021;Breitbarthetal.,2010;Corazza et al., 2020;Pérez-Cañizares,2022), or jointly analyze graphs and photographs as visual disclosure (Hrasky, 2012; Nicolò et al., 2022). The remaining contributions, which feature both longitudinal coverage and cross-sectional variation, include only a few dozen reports from a single country (Chong et al., 2019; García-Sánchez & Araújo-Bernado, 2020). Most studies underscore the centrality of photographs in corporate sustainability reports, which are an integral part of the reporting toolbox (Boiral, 2013; Breitbarth et al., 2010; Zeng et al., 2022). While the extent of photograph use seems to vary markedly between companies (Invernizzi et al., 2021), many authors observe that photographs have become more heavily used in recent years (Ali et al., 2021; Anantharaman et al., 2020; Beattie et al., 2008; Chong et al., 2019). However, we believe that these differences may not only be due to sample selections but also to different measures of photograph use, e.g., counting the number of photographs may produce very different results than assessing photographs against the number of report pages. Despite employing different measures, both Anantharaman et al. (2020) and Nicolò et al. (2022) find that companies operating in environmentally sensitive industries make more use of visual disclosure. Moreover, some evidence exists that photograph use negatively correlates with both the number of topics disclosed in the report as well as CSR performance (Anantharaman et al., 2020). On a similar note, Hrasky (2012) and García-Sánchez and Araújo-Bernado (2020) both observe that companies that are less motivated by sustainability use a more symbolic approach to photograph use, e.g., through the use of non-specific photographs. Furthermore, the photographs generally convey idealized visions of sustainability disconnected from the operational reality, while negative cues are almost completely absent from the visual level (Boiral, 2013; Breitbarth et al., 2010; Chong et al., 2019). Regarding the subject matter depicted in the photographs, existing studies find very different patterns that may plausibly be explained using industry-specific samples. Rämö (2011) and Boiral (2013) find large percentages of environmentally themed photographs, while Invernizzi et al. (2021) and Hrasky (2012) observe socially themed as well as product- and operations-related photographs to be most extensively used. Longitudinal studies by Pérez- Cañizares (2022) and Chong et al. (2019) find an increased focus on depicting people in the photographs. However, as previously pointed out, all studies are characterized by a very limited sample size, often based on a single industry, country, and/or reporting period. Thus, their findings should be interpreted with caution. Despite the richness of their analysis, their results may not be generalizable beyond the particular phenomenon. Hence, we still observe substantial knowledge gaps in the understanding of photograph use in corporate sustainability reports. Firstly, there is a lack of coherent measures to assess photographs in reporting documents and, as a consequence, a basic understanding of the extent of overall photograph use. Secondly, the relative importance of the specific subject matter depicted in photographs is only documented in industry-specific studies. Thirdly, there is scant evidence of how overall photograph use as well as the use of certain themes varies across company characteristics. Some studies observe that industry affiliation and sustainability performance are positively associated with photograph use. We are interested in whether this relationship also holds for subject matter-specific photographs. To address these gaps, we conduct a longitudinal study to provide comprehensive empirical evidence of the role of photographs in corporate sustainability reporting. Accordingly, we seek to contribute more generalizable findings on visual reporting behavior by European 1100 FENK companies. To this end, a new methodological approach is introduced to quantify overall photograph use and facilitate a machine-supported content analysis of large photographic data sets. We thus shed light on the subject matter conveyed through photographs. Given the wide range of visual themes communicated in the reports, we focus our analysis on photographs depicting environmental themes. While each stakeholder has a different set of priorities with regard to corporate sustainability, we argue that the environmental pillar of corporate sustainability is arguably the most pressing. In Europe, this is manifested by an ambitious regulatory agenda centered around the European Green Deal (European Commission, 2019). It covers a range of initiatives including steering financial capital towards green activities, i.e., EU Taxonomy (European Parliament & European Council, 2020), more stringent sustainability reporting requirements, i.e., Corporate Sustainability Reporting Directive (European Parliament & European Council, 2022), and a framework for achieving climate neutrality, i.e., European Climate Law (European Parliament & European Council, 2021). Furthermore, these priorities can also be observed in the heightened interest from the financial community, e.g., the ‘Climate Action 100+group’, green bond issuance, as well as general public awareness through the prominence of the reports by the Intergovernmental Panel on Climate Change or societal and political movements. On a more practical note, we also argue that objects indicating environmental subject matter can be more reliably identified compared to more abstract concepts, such as labor relations or corporate governance. To live up to the relative importance of the environmental pillar within corporate sustainability, we focus our analysis on the visual communication of environmental subject matter. 3|METHODOLOGY In the following section, we introduce our novel machine-supported content analysis approach. Firstly, we outline the sample selection process and describe how we access photograph data from corporate sustainability reports. Thereafter, the identification of environmental subject matter within the photographs is presented. Lastly, we put forward a measure of photograph utilization and specify the regression models employed. 3.1 |Sample selection and data extraction For this study, we collected data from 10 consecutive reporting periods between 2011 and 2020. We consider all companies from countries in the European Economic Area, which have continuously been part of the STOXX Europe 600 index between January 2011 and December 2020. In this way, we can ensure that all firms are subject to a very similar regulatory environment and financially stable. To accurately assess photograph utilization in European corporate sustainability reporting, we limit our sample to stand-alone sustainability reports. This allows us to unequivocally attribute every photograph to a company's sustainability disclosure. For this study, we require that a separately published sustainability report is not additionally contained as a chapter within the annual report. Furthermore, the reporting period has to match the corresponding calendrical year. The corporate sustainability reports were downloaded as portable document files (PDFs) from corporate websites and the GRI Sustainability Disclosure Database (Global Reporting Initiative, 2020) in September 2020 and November 2021. Our final sample contains 1,463 stand-alone sustainability reports from 221 companies consisting of 128,875 pages of sustainability-related disclosure. We summarize the sample selection process in Table 1. We extracted the photographs contained in the reports with a Python script built upon the ‘fitz’functionality module (Kastman, 2017). The program retrieves the image files as portable network graphics (PNGs). In four instances in which the script was unable to extract the image files from the report, we used a webbased extraction tool (Spikerog SAS, 2020). As not all image files contained in the reports are actual photographs, we had to clean the data set from non-photographic PNG files (Rämö, 2011), e.g., charts, TABLE 1 Sample selection process. Step Explanation Number of companies We chose the STOXX Europe 600 index which covers the 600 largest publicly listed companies in Europe. 600 1 We included only companies in our sample which have continuously been part of the STOXX 600 Europe index between January 1st, 2011, and December 31st, 2019, the start and end date of the ten-year study period. Companies which ceased to exist or entered the index in more recent years were not considered. (256) 2 We excluded companies which are not incorporated in the European Economic Area, i.e. Swiss companies, as they are not subject to the European Union's sustainability reporting regulation. (31) 3 We considered only stand-alone sustainability reports as opposed to integrated report chapters included in annual reports. Thus, we can unequivocally attribute every photograph employed to the company's sustainability reporting. Moreover, we required that the reporting period has to match the corresponding calendrical year. (92) Our final sample contains 1,463 stand-alone sustainability reports from 221 companies consisting of 128,875 pages of sustainabilityrelated disclosure. The 1,463 sustainability reports contain 45,228 photographs 221 Note: The sample contains 1,463 stand-alone sustainability reports disclosed by 221 STOXX Europe 600 companies sample which have continuously been part of the index between January 1st, 2011, and December 31st, 2019. FENK 1101 data tables, icons, corporate logos, drawings, or cartoons. We also excluded collages because they represent several photographs in one image file. We removed duplicate photographs in each report, in order to ensure that recurring photographs such as a chapter separator, are only included once. The data cleaning process had two steps: first, we applied a 50-kilobyte memory size threshold on all files to exclude the substantial number of corporate logos and icons from the data set. Second, the remaining procedure was based on a semi-automatic process, whereby image files are flagged based on the subject matter identified by Google's computer vision models. Our final data set of photographs consists of 45,228 photographs contained in 1,463 sustainability reports. 3.2 |Classification of Environmental Photographs Advances in artificial intelligence-powered software and computing resources enable researchers to analyze and draw conclusions from large-scale data sets in an automated way. Computer vision facilitates sophisticated information retrieval from graphic material without having prior knowledge of the presented inputs. It allows for the analysis of much larger data sets than previously feasible and ensures high consistency in the assessment of photographic content. Apart from open-source frameworks, there are several commercial providers of pre-trained computer vision solutions, which do not require additional training of the software. In our study, we use Google's Vision API, which can be accessed through its cloud computing infrastructure (Google LLC, 2023) and has been employed in many scientific projects (Chen & Lin, 2014; Nanne et al., 2020). Our approach, which we outline below, is in part similar to the protocol for analyzing graphic material on web pages outlined by Araujo et al. (2020). To analyze the photographic data set, we deploy Google's label detection feature to retrieve the photograph content by assigning labels. The photographs are thus tagged with up to 10 descriptive labels, as well as a corresponding confidence score for each label ranging between 0 and 1. We performed the data labeling in April 2022. In total, the data set of 45,228 photographs were tagged with 452,004 labels, of which 3,059 are unique values. To classify whether photographs depict environmental subject matter, we developed a codebook based on the list of unique labels assigned by the machine. Two researchers jointly coded the labels, as to whether they unequivocally indicate that the photographs display an environmental theme. We defined labels indicating nature scenes, animals, or renewable energy projects as environmental themes. Furthermore, we only included labels that unambiguously indicate that the relevant object is depicted in the photograph. Hence, we do not consider labels that indicate objects that are very likely present alongside the relevant object, e.g., ‘Shepherd’, who is likely present along a sheep flock. Also, we excluded labels that have a double meaning, e.g., a ‘Python’may both be a programming language or a snake. Lastly, we excluded the labels ‘Sky’and ‘Cloud’because they are assigned to too many photographs rendering them too unspecific. The labels ‘Sky’and ‘Cloud’may still be assigned to a photograph, but we do not include them in the list of classifying labels. From the initial list of 3,059 unique labels, we classified 472 labels as environmental labels. The environmental labels and their corresponding confidence scores are then used to determine whether a photograph is classified as an environmental photograph. In order to qualify as an environmental image, we stipulate that a photograph has to be equipped with at least one environmental label with a confidence score of over 0.9 or two environmental labels with a confidence score of over 0.7. This results in 12,462 photographs labeled as environmental photographs representing 27.55% of all photographs in the data set. 3.3 |Quantifying photograph use To date, there is no broadly accepted measure for assessing photograph utilization in corporate sustainability reporting. Most studies use the sum of photographs to determine the level of photograph utilization, the share of a certain image category (Ali et al., 2021; Chong et al., 2019; Rämö, 2011) or in a further iteration include the size of the photographs into their assessment (García-Sánchez & Araújo-Bernado, 2020; Hrasky, 2012; Invernizzi et al., 2021; Nicolò et al., 2022). However, there are only two groups of authors who calculate photograph utilization against the extent of the report in which they are contained (Anantharaman et al., 2020; Breitbarth et al., 2010). We argue that 20 photographs contained in a 40-page report amount to a substantially different use of photographs than 20 photographs as part of a 100-page report. Thus, we assess the use of photographs against the report corpus measured by the number of pages of the report document. The number of report pages is often referred to as reporting quantity (Arvidsson & Dumay, 2022; Fifka & Drabble, 2012). Accordingly, the use of all photographs as well as specifically environmental photographs is assessed relative to the reporting quantity of the particular report. We call these measures photograph utilization (PUT) and environmental photograph utilization (ENV PUT), respectively. In addition to these absolute measures of photograph use, we calculate environmental photograph use against the number of all photographs contained in the particular report, which we call environmental photograph share (ENV share). In this study, the different measures of photograph use serve as our dependent variables, which are derived below. Overall photograph utilization (PUT) for a particular report is measured against reporting quantity: PUT ¼photograph count page count ð1Þ Environmental photograph utilization (ENV PUT) is equally assessed relative to reporting quantity: ENV PUT ¼environmental photograph count page count ð2Þ 1102 FENK The environmental photograph share (ENV share) for a particular report is measured as a share of all photographs contained in the report: ENV share ¼environmental photograph count photograph count ð3Þ 3.4 |Regression model specification The above-derived measures of photograph use serve as the dependent variables. Thus, we construct three regression models to estimate the photograph used in corporate sustainability reports. The independent and control variables which are used in all three regression models are specified below: I:PUT ¼ß0þß1ENVsensitiveindustryi,tþß2ENVperformancei,t þß3SOCperformancei,tþß4GOVperformancei,tþß5SHcountryi,t þß6RQi,tþß7CO2equrevenuei,tþß8ESGreportscopei,t þß9ESGcontroversyi,tþß10AvgBoardTenurei,tþß11SIZEi,t þß12OPMi,tþεi,t ð4Þ II :ENV PUT ¼ß0þß1ENVsensitiveindustryi,tþß2ENVperformancei,t þß3SOCperformancei,tþß4GOVperformancei,tþß5SHcountryi,t þß6RQi,tþß7CO2equrevenuei,tþß8ESGreportscopei,t þß9ESGcontroversyi,tþß10AvgBoardTenurei,tþß11SIZEi,t þß12OPMi,tþεi,t ð5Þ III :ENV Share ¼ß0þß1ENVsensitiveindustryi,tþß2ENVperformancei,t þß3SOCperformancei,tþß4GOVperformancei,tþß5SHcountryi,t þß6RQi,tþß7CO2equrevenuei,tþß8ESGreportscopei,t þß9ESGcontroversyi,tþß10AvgBoardTenurei,tþß11SIZEi,t þß12OPMi,tþεi,t ð6Þ The financial and non-financial data were obtained from the Refinitiv database (Kind et al., 2023; Landau et al., 2020; Refinitiv Inc., 2023). As we argued in Section 2.3, we are interested in identifying the relationships between sustainability performance and photograph use as well as how photograph use varies across industries. These variables serve as independent variables in our regression. The variables ENVperformance,SOCperformance, and GOVperformance denote the three pillar scores which comprise the commonly used Environmental, Social and Governance (ESG) scores by Refinitiv. The more granular pillar scores measure firm performance in the particular sustainability domain. ENVsensitiveindustry denotes whether a company operates in an environmentally sensitive industry. Following Clarkson et al. (2011) and Branco and Rodrigues (2008), we define the Pulp and Paper, Oil, Gas and Coal, Chemicals, Metals and Mining, Construction as well as Building Materials as environmentally sensitive industries given their high pollution propensity. The industry affiliation is in line with the Industry Classification Benchmark (FTSE Russell, 2023). In addition, we include several control variables in our regression models. SHcountry indicates whether a company is domiciled in a common law or civil law country. Studies by Fifka and Drabble (2012) and Kolk and Perego (2008) have shown that cultural differences play an important role in explaining differences in how companies approach sustainability reporting. RQ denotes the natural logarithm of the sustainability report's page count as a proxy for the amount of information companies choose to disclose in their reporting. To account for the difference between the environmental sensitivity of the industry in general and the individual greenhouse gas emissions of firms, we include CO2equrevenue measuring the natural logarithm of CO2-equivalent emissions divided by net annual revenue. Furthermore, ESGreportscope incorporates the share of a company's commercial activities covered by its non-financial disclosure. ESGcontroversy is included in our models to control the level of controversy surrounding a company. We perform a natural logarithm transformation for all ESG scores to ensure an approximate normal distribution of the variables. AvgBoardTenure denotes the average tenure of company board members. Lastly, we control for company size, SIZE, measured by the natural logarithm of net annual revenue, as well as operating profit margin, OPM, winsorized at the 1% level. Non-Euro figures are Eurodenominated using the European Central Bank's average exchange rate for the particular reporting period. ‘I' and ‘t’denote firm i and year t. εi,trepresents the error term. The Hausman test for all models returns that random effects estimators and fixed effects estimators are both consistent. Hence, we proceed with more efficient random effects estimators. Furthermore, we cluster the robust standard errors by economic agent (i) as well as time period (t). The results of our random effects panel regression are shown below. For Model III, stated by equation (6), we impose an additional restriction to consider only reports with at least five photographs (n =1,185). We argue that, for reports with few photographs, the distribution of the dependent variable would be heavily distorted by a high percentage of extreme values. 4|RESULTS 4.1 |Descriptive statistics and correlations Our sample includes 1,463 separate reports from 221 companies that have continuously been part of the STOXX Europe 600 index of large European companies. Table 2lists the number of observations by reporting period. The considered firms are based in 15 different European countries, of which the United Kingdom represents the largest segment with 65 sample companies, followed by 31 from France and 30 from Germany (see Table 3). With regard to the industry composition, Table 4reports the industry segmentation in line with the two-digit codes of the Industry Classification Benchmark (FTSE Russel, 2023). Table 5shows the descriptive statistics summary, including minimum, mean, median, maximum, and standard deviation values of all continuous dependent, independent, and control variables. Of all considered reports in the data set, 498 (34.04%) were published by FENK 1103 companies from shareholder-oriented countries domiciled in Ireland or the United Kingdom, and 965 from companies located in continental Europe. Considering the industry sensitivity classification, 366 reports (25.02%) were disclosed by companies operating in environmentally sensitive industries. We calculate Pearson's correlation coefficients for all independent and control variables. The strongest correlation between the variables CO2equrevenue and ENVsensitiveindustry stands at 0.607. The additional test for variance inflation factor (VIF) yields no factor exceeding 2.0. The variable SIZE exhibits the highest VIF value of 1.68, 1.76, and 1.90 in Models I, II, and III, respectively. Hence, we see no concern with regard to multicollinearity. Overall, we find that photograph utilization measured by the sum of all photographs relative to reporting quantity has fallen between 2012 and 2017 by around 13% (see Figure 1). Since 2017, the overall level of photograph utilization has been relatively stable at around one photograph for every three report pages. Against this overall trend of falling photograph utilization, the share of photographs conveying environmental themes has steadily increased from 23.12% in 2013 to 31.13% in 2020 (see Figure 2). 4.2 |Regression Results Table 6presents the results of random effects regressions as specified in Section 3.4. The F-test for all models is highly significant. Model I estimates the overall use of photographs against the length of the report. We find that less extensive sustainability reports feature substantially more photographs relative to report length, i.e., an increase in reporting quantity by 10% is associated with a decrease of 0.021 fewer photographs per report page (p < .0001). Controlling for the difference in report length, we also find that reports of continental European companies feature 0.225 fewer photographs per report page than in reports of their UK or Irish peers (p < .0001). Models II and III consider only the use of environmental photographs which are measured against the length of the report (ENV PUT) or as a share of all photographs contained in a particular report (ENV share). The highly significant negative associations between reporting quantity and PUT, as well as the country of domicile and PUT observed from Model I, also hold for environmental photograph utilization. However, these two relationships are not significant for the ENV share measure. With regard to the regressions on the use of environmental photographs, we find that the two main relationships shown below hold equally in both Models II and III. We report that companies operating in environmentally sensitive industries use substantially more environmental photographs on an absolute level, as well as a share of photographs depicting environmental subject matter. Both correlations are highly significant at the p < .0001 level. Holding all else equal, the share of environmental TABLE 3 Number of companies by country of domicile. Country Number of observations Austria 3 Belgium 7 Denmark 7 Finland 10 France 31 Germany 30 Ireland 3 Italy 13 Luxembourg 2 Netherlands 11 Norway 5 Portugal 3 Spain 12 Sweden 19 United Kingdom 65 Note: The sample contains 221 sample companies. TABLE 4 Number of companies by industry sector. ICB Industry Code Industry name Number of observations 10 Technology 9 15 Telecommunications 14 20 Health Care 13 30 Financials 41 35 Real Estate 7 40 Consumer Discretionary 36 45 Consumer Staples 22 50 Industrials 38 55 Basic Materials 15 60 Energy 11 65 Utilities 15 Note: The sample contains 221 companies in total. ICB Industries Codes denote the two-digit industry codes for equities by FTSE Russell (2023). TABLE 2 Number of reports by reporting period. Year Number of observations 2011 158 2012 162 2013 156 2014 162 2015 152 2016 147 2017 140 2018 130 2019 133 2020 123 Note: The sample contains 1,463 report observations. 1104 FENK gambling companies. Accounting, Auditing & Accountability Journal, 35(4), 1009–1034. https://doi.org/10.1108/AAAJ-11-2019-4239 European Commission. (2019). Communication 2019/640 (The European Green Deal). Official Journal of the European Union. European Parliament & European Council. (2014). Directive 2014/95/EU (Non-Financial Disclosure). Official Journal of the European Union. European Parliament & European Council. (2020). Regulation 2020/852 (EU Taxonomy). Official Journal of the European Union. European Parliament & European Council. (2021). Regulation 2021/1119 (European Climate Law). Official Journal of the European Union. European Parliament & European Council. (2022). Directive 2022/2464 (Corporate Sustainability Reporting). Official Journal of the European Union. Fiechter, P., Hitz, J.-M., & Lehmann, N. (2022). Real Effects of a Widespread CSR Reporting Mandate: Evidence from the European Union's CSR Directive. Journal of Accounting Research,60(4), 1499–1549. https://doi.org/10.1111/1475-679X.12424 Fifka, M., & Drabble, M. (2012). Focus and Standardization of Sustainability Reporting - A Comparative Study of the United Kingdom and Finland. Business Strategy and the Environment,21(7), 455–474. https://doi.org/10.1002/bse.1730 Frias-Aceituno, J., Rodríguez-Ariza, L., & García-Sánchez, I.-M. (2014). Explanatory Factors of Integrated Sustainability and Financial Reporting. Business Strategy and the Environment,23(1), 56–72. https://doi. org/10.1002/bse.1765 FTSE Russell. (2023). Industry Classification Benchmark (Equities). Retrieved from https://research.ftserussell.com/products/downloads/ICB_Rules_ new.pdf García-Sánchez, I.-M., & Araújo-Bernado, C. (2020). What colour is the corporate social responsibility report? Structural visual rhetoric, impression management strategies, and stakeholder engagement. Corporate Social Responsibility and Environmental Management,27(2), 1117–1142. https://doi.org/10.1002/csr.1869 Global Reporting Initiative. (2020). Sustainability Dislosure Database [Computer software]. Retrieved from https://database.globalreporting.org/ search/ Google LLC. (2023). Python google-cloud-vision [Computer software]. Retrieved from https://pypi.org/project/google-cloud-vision/ Gray, R., Kouhy, R., & Lavers, S. (1995). Corporate social and environmental reporting. Accounting, Auditing & Accountability Journal,8(2), 47–77. https://doi.org/10.1108/09513579510146996 Gugerty, M. (2009). Signaling virtue: voluntary accountability programs among nonprofit organizations. Policy Sciences,42(3), 243–273. https://doi.org/10.1007/s11077-009-9085-3 Guthrie, J., & Parker, L. (1989). Corporate Social Reporting: A Rebuttal of Legitimacy Theory. Accounting and Business Research,19(76), 343– 352. https://doi.org/10.1080/00014788.1989.9728863 Hahn, R., & Lülfs, R. (2014). Legitimizing Negative Aspects in GRI-Oriented Sustainability Reporting: A Qualitative Analysis of Corporate Disclosure Strategies. Journal of Business Ethics,123(3), 401–420. https:// doi.org/10.1007/s10551-013-1801-4 Hallin, A., Karrbom-Gustavsson, T., & Dobers, P. (2021). Transition towards and of sustainability—Understanding sustainability as performative. Business Strategy and the Environment,30(4), 1948–1957. https://doi.org/10.1002/bse.2726 Herzig, C., & Schaltegger, S. (2011). Corporate Sustainability Reporting. In J. Godemann & M. Michelsen (Eds.), Sustainability Communication (pp. 151–169). Springer, Cham. https://doi.org/10.1007/978-94-007- 1697-1_14 Higgins, C., Tang, S., & Stubbs, W. (2020). On managing hypocrisy: The transparency of sustainability reports. Journal of Business Research, 114, 395–407. https://doi.org/10.1016/j.jbusres.2019.08.041 Hooghiemstra, R. (2000). Corporate Communication and Impression Management –New Perspectives Why Companies Engage in Corporate Social Reporting. Journal of Business Ethics,27,55–68. https://doi.org/10.1023/A:1006400707757 Hopwood, A. (2009). Accounting and the environment. Accounting, Organizations and Society,34(3–4), 433–439. https://doi.org/10.1016/j.aos. 2009.03.002 Hrasky, S. (2012). Visual disclosure strategies adopted by more and less sustainability-driven companies. Accounting Forum,36(3), 154–165. https://doi.org/10.1016/j.accfor.2012.02.001 Hummel, K., & Schlick, C. (2016). The relationship between sustainability performance and sustainability disclosure –Reconciling voluntary disclosure theory and legitimacy theory. Journal of Accounting and Public Policy,35(5), 455–476. https://doi.org/10.1016/j.jaccpubpol.2016. 06.001 Invernizzi, A., Bellucci, M., Acuti, D., & Manetti, G. (2021). Form and substance: Visual content in CSR reports and investors' perceptions. Psychology & Marketing,39(5), 974–989. https://doi.org/10.1002/mar. 21635 Kastman, E. (2017). Python fitz (Version 0.0.1.dev2) [Computer software]. Kind, F., Zeppenfeld, J., & Lueg, R. (2023). The impact of chief executive officer narcissism on environmental, social, and governance reporting. Business Strategy and the Environment,32(7), 4448–4466. https://doi. org/10.1002/bse.3375 Knobloch, S., Hastall, M., Zillmann, D., & Callison, C. (2003). Imagery Effects on the Selective Reading of Internet Newsmagazines. Communication Research,30(1), 3–29. https://doi.org/10.1177/00936502 02239023 Kolk, A., & Perego, P. (2008). Determinants of the adoption of sustainability assurance statements: an international investigation. Business Strategy and the Environment,19(3), 182–198. https://doi.org/10.1002/ bse.643 Krippendorff, K. (2018). Content analysis: An introduction to its methodology (4th ed.). SAGE. Landau, A., Rochell, J., Klein, C., & Zwergel, B. (2020). Integrated reporting of environmental, social, and governance and financial data: Does the market value integrated reports? Business Strategy and the Environment,29(4), 1750–1763. https://doi.org/10.1002/bse.2467 Lee, T. (1994). The changing form of the corporate annual report. The Accounting Historians Journal,21(1), 215–232. https://doi.org/10. 2308/0148-4184.21.1.215 Lyon, T., & Montgomery, W. (2015). The Means and End of Greenwash. Organization & Environment,28(2), 223–249. https://doi.org/10.1177/ 1086026615575332 Mahoney, L., Thorne, L., Cecil, L., & LaGore, W. (2013). A research note on standalone corporate social responsibility reports: Signaling or greenwashing? Critical Perspectives on Accounting,24(4–5), 350–359. https://doi.org/10.1016/j.cpa.2012.09.008 Meng, X., Zeng, S., & Tam, C.-M. (2013). From Voluntarism to Regulation: A Study on Ownership, Economic Performance and Corporate Environmental Information Disclosure in China. Journal of Business Ethics, 116(1), 217–232. https://doi.org/10.1007/s10551-012-1462-8 Merkl-Davies, D., & Brennan, N. (2007). Discretionary Disclosure Strategies in Corporate Narratives: Incremental Information or Impression Management? Journal of Accounting Literature,26, 116–196. Merkl-Davies, D., & Brennan, N. (2011). A Conceptual Framework of Impression Management. Accounting and Business Research,41(5), 415–437. https://doi.org/10.1080/00014788.2011.574222 Meuer, J., Koelbel, J., & Hoffmann, V. (2020). On the Nature of Corporate Sustainability. Organization & Environment,33(3), 319–341. https:// doi.org/10.1177/1086026619850180 Milne, M., & Patten, D. (2002). Securing organizational legitimacy. Accounting, Auditing & Accountability Journal,15(3), 372–405. https://doi.org/ 10.1108/09513570210435889 Nanne, A., Antheunis, M., van der Lee, C., Postma, E., Wubben, S., & van Noort, G. (2020). The Use of Computer Vision to Analyze Brand- FENK 1111 Related User Generated Image Content. Journal of Interactive Marketing,50, 156–167. https://doi.org/10.1016/j.intmar.2019.09.003 Nicolò, G., Ricciardelli, A., Raimo, N., & Vitolla, F. (2022). Visual disclosure through integrated reporting. Management Decision,60(4), 976–994. https://doi.org/10.1108/MD-01-2021-0034 Niehoff, S. (2022). Aligning digitalisation and sustainable development? Evidence from the analysis of worldviews in sustainability reports. Business Strategy and the Environment,31(5), 2546–2567. https://doi. org/10.1002/bse.3043 Paivio, A. (1969). Mental imagery in associative learning and memory. Psychological Review,76(3), 241–263. https://doi.org/10.1037/h0027272 Paivio, A., Rogers, T., & Smythe, P. (1968). Why are pictures easier to recall than words? Psychonomic Science,11(4), 137–138. https://doi.org/10. 3758/BF03331011 Patten, D. (2002). The relation between environmental performance and environmental disclosure: a research note. Accounting, Organizations and Society,27(8), 763–773. https://doi.org/10.1016/S0361-3682 (02)00028-4 Peeples, J. (2011). Toxic Sublime: Imaging Contaminated Landscapes. Environmental Communication,5(4), 373–392. https://doi.org/10.1080/ 17524032.2011.616516 Pérez-Cañizares, P. (2022). Visualising Corporate Social Responsibility. Ibérica,43, 155–178. https://doi.org/10.17398/2340-2784.43.155 Pesci, C., Costa, E., & Soobaroyen, T. (2015). The forms of repetition in social and environmental reports: insights from Hume's notion of ‘impressions’.Accounting and Business Research,45(6–7), 765–800. https://doi.org/10.1080/00014788.2015.1084224 Prado-Lorenzo, J.-M., & García-Sánchez, I.-M. (2010). The Role of the Board of Directors in Disseminating Relevant Information on Greenhouse Gases. Journal of Business Ethics,97(3), 391–424. https://doi. org/10.1007/s10551-010-0515-0 Preston, A. M., Wright, C., & Young, J. J. (1996). IMag[in]ing annual reports. Accounting, Organizations and Society,21(1), 113–137. https://doi.org/10.1016/0361-3682(95)00032-5 Rämö, H. (2011). Visualizing the Phronetic Organization: The Case of Photographs in CSR Reports. Journal of Business Ethics,104(3), 371–387. https://doi.org/10.1007/s10551-011-0916-8 London Stock Exchange Group PLC. (2023). Eikon Database [Computer software]. Reverte, C. (2009). Determinants of Corporate Social Responsibility Disclosure Ratings by Spanish Listed Firms. Journal of Business Ethics, 88(2), 351–366. https://doi.org/10.1007/s10551-008-9968-9 Rezaee, Z. (2016). Business sustainability research: A theoretical and integrated perspective. Journal of Accounting Literature,36(1), 48–64. https://doi.org/10.1016/j.acclit.2016.05.003 Shabana, K., Buchholtz, A., & Carroll, A. (2017). The Institutionalization of Corporate Social Responsibility Reporting. Business & Society,56(8), 1107–1135. https://doi.org/10.1177/0007650316628177 Spikerog SAS. (2020). PDF Extractor [Computer software]. Retrieved from https://www.extractpdf.com/ Suchman, M. (1995). Managing Legitimacy: Strategic and Institutional Approaches. The Academy of Management Review,20(3), 571. https:// doi.org/10.2307/258788 Townsend, C., & Shu, S. (2010). When and how aesthetics influences financial decisions. Journal of Consumer Psychology,20(4), 452–458. https://doi.org/10.1016/j.jcps.2010.06.013 Treisman, A. (1985). Preattentive processing in vision. Computer Vision, Graphics, and Image Processing,31(2), 156–177. https://doi.org/10. 1016/S0734-189X(85)80004-9 Tversky, A., & Kahneman, D. (1986). Rational choice and the framing of decisions. Journal of Business,59(4), 251–278. https://doi.org/10. 1086/296365 Tversky, B. (1974). Eye fixations in prediction of recognition and recall. Memory & Cognition,2(2), 275–278. https://doi.org/10.3758/ BF03208995 Unerman, J. (2000). Methodological issues - Reflections on quantification in corporate social reporting content analysis. Accounting, Auditing & Accountability Journal,13(5), 667–681. https://doi.org/10.1108/ 09513570010353756 Usmani, M., Davison, J., & Napier, C. (2020). The production of standalone sustainability reports: visual impression management, legitimacy and “functional stupidity”.Accounting Forum,44(4), 315–343. https:// doi.org/10.1080/01559982.2020.1782566 Wedari, L., Jubb, C., & Moradi-Motlagh, A. (2021). Corporate climaterelated voluntary disclosures: Does potential greenwash exist among Australian high emitters reports? Business Strategy and the Environment,30(8), 3721–3739. https://doi.org/10.1002/bse.2836 Zeng, X., Momin, M., & Nurunnabi, M. (2022). Photo disclosure in human rights issues by fortune companies: an impression management perspective. Sustainability Accounting, Management and Policy Journal, 13(3), 568–599. https://doi.org/10.1108/SAMPJ-06-2019-0243 Zillmann, D., Gibson, R., & Sargent, S. (1999). Effects of Photographs in News-Magazine Reports on issue Perception. Media Psychology,1(3), 207–228. https://doi.org/10.1207/s1532785xmep0103_2 How to cite this article: Fenk, L. (2025). Environmental photograph use in corporate sustainability reporting: A machine-supported visual content analysis. Business Strategy and the Environment,34(1), 1097–1112. https://doi.org/10. 1002/bse.4035 1112 FENK