Numerous studies have identified information overload in the context of financial and accounting decisions, and the related literature is extensive (Paredes Troy, 2003). Information overload occurs when the supply of information exceeds the capacity for their interpretation (Eppler and Mengis, 2003). These hinder the decision maker’s ability to optimally define the best decision. Information overload is a situation in which the decision-maker is faced with a large amount of information (i.e., a type of information with informative characteristics such as number, complexity, as well as the level of redundancy, contradiction and inconsistency), which make up a bunch of individual informative cues of different sizes and complexity (Roetzel, 2019). Hwang and Lin (1999) found that there is a relationship between information overload and the ability of managers to process information for the decision-making since it reduces decision makers’ ability to make the best possible decision (Roetzel and Fehrenbacher, 2020), and reducing the quality of decisions (Letsholo and Pretorius, 2016). The phenomenon of information overload is one of the main challenges of the information age (Roetzel and Fehrenbacher, 2020).

Information overload is related to how individuals navigate the presence of multiple choices. While having options is often a positive experience, the presence of too many options at some point can become cognitively challenging for individuals (Tandoc and Kim, 2022). Lazer et al. (2017) also argued that information overload, coupled with limited individual attention, may prevent adequate assessment of information on social media, which may contribute to misinformation. The main causes of information overload are the result of a number of developments that are mutually interdependent and relate to various levels or scales. On a societal scale, these developments are mainly the accelerated production of information through institutions such as science, the media, or the globalized business community which operates on ever faster innovation cycles, and the more efficient distribution of that information through information and communication technology. Such societal factors are usually not discussed extensively in information overload literature dealing with organizational or business-related issues (Eppler and Mengis, 2003).

An important cause of information overload is the nature of the information itself. Schneider (1987) pointed out that the extent of information overload is not only affected by the amount of information, but also by its specific characteristics (level of information uncertainty, level of ambiguity, novelty, complexity or intensity). Simpson and Prusak (1995) advocated the views that changing the quality of information can have a great impact on the probability of information overload, as it can improve the quality of information (conciseness, consistency, comprehensiveness, etc.) and improve an individual’s information processing capabilities. In the last decade, some approaches to “soft” characteristics (i.e., subjective experiences and user trust) have been identified, which shift the focus from technical to psychological aspects (e.g., Koroleva and Bolufé Röhler, 2012; Wu and Lin, 2006).

Regulators, both national and international, have made significant contributions in response to the growing amount of information and its diverse sources, recognizing the need to establish clearer standards for managerial and financial practices. This regulatory shift was particularly pronounced following major financial scandals, such as the collapse of Enron, and the global financial crisis of 2008, which exposed systemic weaknesses in investor protection and corporate governance. In the wake of these crises, both regulators and investors began to question the transparency and integrity of financial markets, as well as the ethical standards of the entities operating within them. This erosion of trust triggered a wave of regulatory and legal reforms, including the establishment of the Public Company Accounting Oversight Board and the enactment of the Sarbanes–Oxley Act of 2002, which aimed to enhance corporate accountability and financial reporting standards.

Securities regulations have traditionally been grounded in the assumption that more information is inherently beneficial for market efficiency. Accordingly, regulatory frameworks have increasingly mandated extensive disclosures in an effort to improve transparency and bolster investor confidence. However, the proliferation of information has also introduced new challenges, particularly in terms of cognitive load and decision-making complexity at the managerial level. The growing volume and complexity of available data underscore the importance of cognitive factors—defined here as both conscious and unconscious mental processes involved in processing, storing, and transmitting information. Within this context, managerial tools such as goal-setting have come under increasing scrutiny.

Although goal-setting remains a widely used and empirically supported managerial practice (Locke and Latham, 2013), recent research has raised concerns about its unintended consequences for ethical behaviour (Niven and Healy, 2016). Specifically, while well-designed goals can enhance motivation and performance, they may also create pressure that fosters moral disengagement or justifies unethical conduct, particularly when goals are overly ambitious or narrowly focused.

Managers are typically required to disclose performance-related information to shareholders. However, this information is often highly condensed, selectively presented, and potentially subject to manipulation (Desai et al., 2006). When outcomes are not objectively measurable and must be interpreted or contextualized by management, the causal relationship between managerial actions and shareholder value becomes increasingly opaque. Consequently, complexity-based information asymmetries can mask opportunistic behaviour and elevate the risk of financial reporting fraud. Empirical research by Ndofor et al. (2015) confirms that such asymmetries are positively associated with the likelihood of fraudulent financial reporting.

The transmission of incomplete, ambiguous, or misleading information by management can cause recipients—investors, regulators, or other stakeholders—to make decisions they might not otherwise have made had they received accurate and comprehensible data. Thus, the format, content, and delivery of managerial communication play a critical role in shaping stakeholder perceptions and responses. This underscores the need for a deeper understanding of the strategic decision-making and communication processes within organizations.

The principal–agent relationship is particularly susceptible to three canonical issues in the presence of information asymmetry: moral hazard, adverse selection, and challenges related to unverifiable information. These phenomena impair the efficiency of the principal–agent dynamic, especially when the agent has both the motive and the means to distort information to serve personal interests.

Much of the literature on financial reporting fraud has focused on incentive structures, particularly the role of stock-based compensation in encouraging executives to overstate performance (Ndofor et al., 2015). In addition to economic motives, personality traits and leadership styles have also been implicated in unethical decision-making. For example, personalized charismatic leaders—those who seek power and influence over others—are more likely to manipulate followers to achieve self-serving goals (House and Howell, 1992). This leadership profile has conceptual overlap with the construct of harmful, self-centered leadership, which has been linked to unethical organizational behaviour (Niven and Healy, 2016).

Further evidence of strategic communication practices is provided by Persson (2018), who identified deliberate information obfuscation by management. Specifically, firms may intentionally induce information overload or employ “complexification”—i.e., adding unnecessary complexity to disclosures—to reduce stakeholder comprehension. While this behaviour can be strategic and intentional, complexity in communication may also arise unintentionally, due to excessive optimism or the unstructured presentation of large volumes of data. Ultimately, complexity-based information asymmetry remains a critical risk factor in corporate governance. It facilitates opportunistic behaviour, hinders effective oversight, and exacerbates the classic agency problems that underlie many instances of financial misconduct.

Among the most important sources of information provided by management are public announcements, annual reports, and interviews with senior management. As annual reports represent the primary communication channel between management and shareholders, they play a central role in shaping perceptions of corporate performance and transparency. Given their dual purpose—to inform stakeholders and to project a positive corporate image—they provide a particularly suitable medium for examining how complexity affects the clarity and accessibility of information. Accordingly, our empirical analysis focuses on the letters to shareholders contained within these reports, which are discussed in greater detail in Appendix A.

The interview sample was purposeful and non-random. There are only few empirically grounded studies on the process of information use by capital market actors (Hjelström et al., 2014), hence we engaged in discussions with individuals who are most knowledgeable about the issue - the top management of each company (i.e., purposive sampling).

The interviewees were selected using purposive sampling based on the following criteria: each was a top management board member (Chief Executive Officer (CEO), Chief Financial Officer (CFO), or similar) with at least 12 years of senior management experience in a company obligated to publish annual reports or to issue securities; each interviewee’s company was classified as a medium-sized or large enterprise under Slovenian legislation; and each company had been established for at least ten years.

The selected people were contacted, informed about how the data would be collected and how the interviews would be conducted, and asked to participate. 10 of the 14 invited interviewees responded positively and were interviewed. All 10 interviewees who agreed to be interviewed are top management members, comprising 4 CEOs, 4 CFOs, and 2 Chief Risk Officers (CROs). Their collective experience spans diverse sectors and industries (port and logistic system, banking group, power generation company, food manufacturing, asset management, government owned holding company, and financial institution), making them valuable assets to the company’s management boards. The companies, in which they act as representatives of the company’s management, are leaders in the industry and are among the largest and most important companies in the country. The average managerial experience of the interviewees in the top tier positions was 20 years. Seven of them have also had experience as members of supervisory boards (as presidents or members of supervisory boards of large companies). One of the 10 interviewees was a woman. We clarify that the 10 experts interviewed are not necessarily connected to the 9 analysed companies. We believe that such a purposive sample is appropriate for the aims and objectives of the research.

The interviews were conducted over a period of five months between November 2020 and March 2021. The basis for preparing the semi-structured interview were the research question presented in the introduction.

At the beginning of each interview, we posed a general question regarding the communication of management or supervisory board members regarding annual reports. Subsequently, we progressed to more specific questions pertaining to the study of the reasons, conditions, and consequences of the information overload. Throughout, we maintained a consistently meaningful order of questions, facilitating data analysis. The number of all questions varied slightly between individual interviewees based our judgment that the respondent had already answered a given question directly or indirectly. The content of the recorded interviews (literal transcription) was typed into an individual document. The recorded interview therefore enabled us to make an accurate record and subsequently transcribe the content of the interview. After the completion of the translation into text, we rechecked the content of the transcription for the adequacy of each interview, whether the recording corresponds to the reality of the audio recording.

We followed the standard protocol when conducting the individual semi-structured interview. The conversation took place via Zoom application (audio only, no video), with an average interview length of 38 minutes. By agreeing to the interview, the interviewees gave their consent for the recording of the conversations.

The basis for preparing the questions for the semi-structured interview were research questions from the research, which we expanded or deepened after the analysis of the literature and placed them into three sets of questions covering: (1) managers’ role in communication with the company’s key stakeholders, (2) the position of management in the field of communication, especially in relation to investors, and (3) information overload and the complexity effect. The interviews (see Appendix B (Appendix Table 8) for the full question list) were designed to elicit the personal opinions and experiences of the company leaders regarding their communication in form of annual reports.

For the analysis of the interviews, we used the content analysis method, which structures qualitative data in a formally valid way. The method was implemented in the following stages: categories were determined by categorization, and codes were determined by coding. The categorization process was carried out according to the following steps: distribution of interviews by person; classification of answers by content groups and content analysis; defining and numbering categories; and content re-analysis.

For the analysis, we developed a coding scheme based on eight behavioural economics concepts: moral hazard, excessive optimism (overconfidence and arrogance), attention manipulation (adverse selection), heuristic-driven bias, herd behaviour, bounded rationality, information overload, and the complexity effect. Comments in the interviews that correspond to the answers to the research question and sub-questions have been marked separately. Codes were assigned to the coding units (sentence, paragraph, chapter), which were used to mark the essential components of the data. A code is a word, characteristic, or thing that indicates the meaning of words. It also refers to quotations, which are a verbatim extract of a specific part of the content. We grouped the codes into categories, which are repeating patterns, extreme phenomena, processes, etc. Appendix C (Appendix Table 9) provides an example of the coding structure. Our aim was to understand how managers communicate while balancing impartiality and transparency against vested interests, and to illuminate the information relationship between agent and principal.

To study letters to shareholders in the annual reports of all 9 Slovenian companies which are traded on the Prime Market of Ljubljana Stock Exchange (see Table 1), we used the Flesch Reading Ease formula. The conditions for inclusion of a public company in the primary listing of the Ljubljana Stock Exchange are represented by the criteria listed in Table 2.

We estimate that one of the most appropriate forms of showing increased complexity in a corporation is their letter to shareholders within annual report, in which management reports to stakeholders about all aspects of the corporation’s operations.

Abrahamson and Amir (1996) argue that a letter to shareholders may be the most interesting part of an annual report to read, as it is the only section where the author has considerable freedom in choosing the content and faces few restrictions on how information is presented. Hyland (1998) likewise suggests that letter to shareholders is the one part of an annual report that readers scrutinize thoroughly. Furthermore, Amernic et al. (2010) contend that the CEO’s shareholder letter is a key medium through which corporate leaders communicate their attitudes and values, providing insight into the “tone at the top” of a company.

In theory, the effectiveness of financial reporting rests upon annual reports that facilitate reading and interpreting financial information. Readability can be defined as the perception of a text, which is usually measured by the speed of reading, comprehension, and memorization. In the world, mostly in English-speaking countries, several different readability methods and measures have been developed, the aim of which is to determine whether a text is too difficult for a target age group. The Flesch Reading Ease Scale uses two variables to determine the readability score (Srinivasan et al., 2015): average sentence length in words (W/S) and syllables per 100 words (S/W). The Flesch Reading Ease is a calculated score tied to standards defined in written texts and classified according to reading difficulty.

Studies have confirmed that among the many readability formulas, the Flesch formula is the most widely used, and it is also considered one of the most tested and reliable (DuBay, 2006). The readability of an important element of the annual reports—the letter to shareholders—was calculated using the Flesch Reading Ease formula, which measures text complexity using two variables:

$\bullet$ average sentence length, measured by the number of words, and

$\bullet$ average syllable length per word, which indirectly measures word complexity.

Formula for measuring the level of ease of reading (DuBay, 2006):

Reading Ease Level = 206.835 – 1.015 (W/S) – 84.6 (S/W)

The index takes values between 0 and 100, whereby a higher index value indicates better readability of an individual text (or easier to read text), and an index value of 50 indicates the average difficulty of the text to read. Interpretations are summarized in Table 3.

We present the findings of 10 in-depth interviews with personnel in managerial positions of companies in Slovenia. We interpreted management thinking according to eight categories of behavioural economics. We marked the categories with labels from r to y and list of ten persons from A to J. The characteristics of the category had to be sufficiently broad, and we checked whether they were supported by citations. For easier interpretation of the explanation, we classified the answers of the management and members of the supervisory board according to the categories of behavioural economics attributed to each answer, as we wanted to understand the parallels between them. Selected data highlights with comprehensive responses grouped into individual behavioural economics categories are available in Appendix C (Appendix Table 9). Table 4 provides an overview of the behavioural economics categories and their corresponding subjects of influence, distinguishing between management, individual, and combined effects. As shown in Table 4, certain categories are specifically linked to managerial decision-making, while others reflect broader individual or shared behavioural patterns.

From the LJSE website (https://seonet.ljse.si) we obtained the annual reports of all 9 listed companies from Prime Market of the Ljubljana Stock Exchange and exported the letters to shareholders from them into the Microsoft Word program, using only the text (without the title) and without separate sub-headings or management signatures. The studied annual reports of Slovenian companies, published in English for the business year 2019, represent 100% of the composition of the prime market (9 companies in year 2019). Table 5 presents the scope of the analysed annual reports by indicating the number of pages in both Slovenian and English versions for each company. As shown in Table 5, the length of the reports varies across companies, while remaining broadly comparable between language versions.

Table 6 summarises the descriptive statistics of the Flesch Reading Ease scores for the analysed reports. As shown in Table 6, the results indicate generally low readability levels, with most values concentrated in the lower range of the scale.

Table 7 presents a year-over-year comparison of Flesch Reading Ease scores between 2013 and 2019 at the company level. As shown in Table 7, readability trends vary across companies, with some showing improvement while others exhibit a decline over time.

The results of the readability analysis for the management reports of 2019 indicate a range between 15.1 and 46, with a median of 24.7 and an arithmetic mean of 27.5. The median results demonstrate that the readability of the reports of Slovenian companies is difficult, ranging between 0 and 30. The variation gap is also high, indicating significant differences in readability between the most and least comprehensible texts. The annual report of Luka Koper exhibited the highest level of readability due to the publication of an interview with the president of the management board instead of a traditional management report. The lowest Flesch index value of the worst-performing company (Telekom Slovenije) is likely due to the length of the sentences.

The Flesch index shows an arithmetic mean of 27.5 and a median of 24.7. The results of our analysis of annual reports show that letters to shareholders in Slovenia are difficult or very difficult to read, which undoubtedly makes it more difficult for investors to glean useful information. We found that most annual reports include statements that are long-winded and could be simplified accordingly without diminishing their meaning. The fact is that the readability of annual reports contributes to improving the quality of information and helps to ensure that the needs of internal and external users of financial statements are met.

We believe that the theoretical foundations and empirical findings of this study can offer valuable insights beyond the realm of corporate governance, particularly in contexts where agency relationships are present. The complexity effect may also influence communication in public administration, nonprofit organizations, healthcare, or international institutions—sectors where information asymmetry and strategic communication play a critical role in stakeholder relations.

One promising avenue for addressing the complexity effect lies in the application of artificial intelligence, particularly large language models (LLMs). These models, based on artificial neural networks and designed for natural language processing tasks, have the potential to translate complex and technical content into clear and accessible language. By identifying intricate language patterns, extracting key concepts, and generating concise summaries or explanations, LLMs can help bridge the comprehension gap between information senders (agents) and receivers (principals). This technological approach could enhance information transparency and democratize access to knowledge across a broad range of audiences.

Future research could explore the effectiveness of such tools in corporate communication settings—for instance, by measuring whether AI-assisted simplification improves investor understanding and decision-making. Comparative studies between AI-mediated and traditional forms of disclosure could yield insights into how best to balance legal, strategic, and ethical considerations in corporate reporting. Additionally, interdisciplinary research combining corporate governance, behavioural finance, and computational linguistics could open new frontiers in understanding and mitigating the impact of complexity in organizational communication.