Empirical Studies in Financial Accounting

Abstract Abstract

Purpose: This research investigates the impact of financial disclosure complexity on the reaction of the Iranian capital market, emphasizing the moderating role of social media attention. Grounded in behavioral finance, specifically limited attention theory and information processing costs, the study addresses two questions: (1) whether more complex financial disclosures lead to weaker market reaction in the Tehran Stock Exchange (TSE), and (2) whether attention generated on Persian Twitter moderates this negative relationship. The study provides empirical evidence from an emerging market context, contributing to the literature on cognitive biases in financial decision-making and the informational role of social media.



Theoretical Framework and Hypothesis Development: The theoretical framework integrates Limited Attention Theory (Kahneman, 1973; Hirshleifer & Teoh, 2003) and Information Processing Cost Theory (Bloomfield, 2008). Investors face cognitive constraints, operating as selective information processors. Complex disclosures—characterized by length, numerical density, and ambiguous language—elevate processing costs, leading to incomplete or delayed price adjustments. This phenomenon is particularly pronounced in emerging markets like Iran, where retail investors with lower financial literacy constitute a significant portion of market participants (Moradi et al., 2021). Concurrently, the digital attention economy paradigm suggests social media platforms have become primary arenas for attracting investor attention. Platforms like Twitter facilitate collective discourse and cognitive facilitation, potentially mitigating the adverse effects of information complexity. Based on this synthesis, two hypotheses were developed: H1 posits a negative relationship between financial disclosure complexity and immediate market reaction (measured by cumulative abnormal returns and abnormal trading volume). H2 proposes that pre-disclosure social media attention positively moderates this relationship; the negative effect of complexity on market reaction is weaker for firms receiving higher levels of digital attention.



Methodology: The study employs a quantitative, event-study-based design with panel data spanning 2016–2023. The final sample comprises 1,024 firm-year observations from 128 non-financial companies listed on the TSE, selected through systematic screening. Disclosure complexity was measured using a multidimensional composite index, standardizing and averaging three proxies: (1) report length (logarithm of word count in explanatory notes), (2) numerical density (ratio of numerical digits to total words), and (3) degree of ambiguity (frequency of uncertain words such as "may," "probably," "risk"). Textual data were extracted from financial reports on the Codal database using Python libraries (PyPDF2, pdfplumber, BeautifulSoup) and processed with Hazm, a specialized Persian NLP library. Investor attention was quantified using data from Persian Twitter. Twitter was selected due to its textual nature, relatively open API, and prevalence as a forum for stock exchange discussions in Iran. Over 250,000 Persian tweets containing company-specific keywords (stock symbols, full names, abbreviations) were collected using snscrape and tweepy libraries. The attention metric (ATTN) was defined as the natural logarithm of the average daily count of unique, firm-specific posts during the five-day window preceding annual report releases. Control variables included firm size, book-to-market ratio, profitability (ROA), leverage, institutional ownership, Big Four auditor quality, and market return index. The analytical approach employed fixed-effects panel regression models with firm-clustered robust standard errors. The main model tested the interaction between COMP and ATTN, with robustness ensured through alternative variable constructions, different event windows, GMM estimation, and subsample analyses.



Key Findings: The empirical analysis provides strong support for both hypotheses. First, a statistically significant (p < 0.01) negative coefficient was found for the complexity variable (COMP). Economically, a one-standard-deviation increase in complexity leads to an approximately 1.4% reduction in short-window cumulative abnormal returns (CAR[0,+1]). This confirms that greater disclosure complexity dampens the market's immediate price adjustment, supporting limited attention and information processing cost theories. Second, the interaction term between COMP and ATTN was positive and statistically significant (p < 0.05). Marginal effect analysis revealed that the negative slope of complexity on returns flattens as social media attention increases. For firms in the highest attention quartile, the adverse effect of complexity becomes statistically indistinguishable from zero, demonstrating that high levels of digital attention can effectively neutralize the information processing barrier created by complex disclosures. Notably, the direct effect of social media attention alone on returns was insignificant, highlighting that its primary role is moderating rather than directly price-informative. Robustness checks using alternative windows, volume measures (CAV), different complexity specifications, and subsample analyses consistently reaffirmed the core results. Subsample analysis showed that the moderating effect of attention was stronger for firms with lower institutional ownership, where retail investors predominate.



Discussion and Implications: The findings offer substantial contributions to theory and practice. Theoretically, the study extends limited attention theory to the digital age within an emerging market context, demonstrating that social media attention acts as a critical moderating variable capable of mitigating market inefficiencies stemming from information complexity. It bridges the literature on formal financial reporting with informal information diffusion through digital networks. It introduces a theoretical distinction between "mere attention" and "quality information processing," evidenced by the insignificant direct effect of attention versus its significant interactive effect with complexity. Methodologically, the research pioneers the use of Persian social media data and NLP techniques in accounting and finance research. Practically, the findings carry important implications. For regulators like the Securities and Exchange Organization of Iran, the evidence underscores the need for policies promoting simplification and readability in mandatory disclosures, such as setting maximum limits for report length or minimum readability scores. Regulators could establish "attention monitor" systems to identify firms experiencing attention deficits during critical disclosure periods. For corporate managers, the results highlight the strategic importance of active investor relations in the digital sphere; proactive engagement on social media—publishing simplified management summaries or hosting online Q&A sessions—can ensure complex financial information is effectively communicated. For retail investors, the study serves as a caution about cognitive biases and suggests consulting credible informal channels when confronting complex reports. For auditors, the findings emphasize considering "understandability" as a practical objective during financial reporting consultations.



Limitations and Future Research Directions: The study acknowledges several limitations. First, data collection was limited to Persian Twitter; future studies could incorporate other platforms like Instagram and Telegram. Second, attention was measured quantitatively (post counts); future research could employ advanced NLP techniques to analyze content quality, sentiment, and user influence, distinguishing between "productive" and "superficial" attention. Third, the focus was solely on annual reports; future studies could examine quarterly reports, earnings announcements, or event-based disclosures. Fourth, despite advanced econometric methods, the observational data precludes definitive causal claims; mixed-method approaches could provide deeper insights. Finally, cross-country comparative studies could explore how institutional differences shape the relationship between disclosure complexity, attention, and market reactions.



Conclusion: This research establishes that the efficiency of the Iranian capital market in incorporating complex financial information is not determined by disclosure complexity alone but is fundamentally contingent upon the level of investor attention it garners in the digital ecosystem. It provides compelling evidence that social media activity can function as a societal corrective to cognitive limitations, enhancing price discovery. The study advocates for an integrated perspective in market regulation and corporate strategy that simultaneously addresses the quality of formal disclosures and the dynamics of informal digital communication channels.

Abstract Abstract Individuals are thought to make biased judgments under uncertainty, because limited time and cognitive resources lead them to apply heuristics like representativeness. Representativeness is the tendency of individuals to classify things into discrete groups based on similar characteristics. In order to measure the representativeness bias, we examine the relation between past trends and sequences in financial performance and future returns in Tehran stock exchange (TSE) between1380-1390. We also investigate the impact of consistent sequence of financial performance in future return. Finally, the study examines the effect of subsequent performance that confirms or contradicts past pattern of growth on the predictability of future returns. The study uses annual data consisted of 800 firms-year and 3200 firms-quarter. The main research methodology is portfolio study.  This study calculates financial growth rates over two periods: one year (four rolling quarters) and five years (using annual data). Three accounting measures of performance are calculated: sales, net income, and operating income. The results indicate that the abnormal returns in one-year trend are significantly positive. But abnormal returns in the year after five years of high or low growth are statistically and economically insignificant. In addition, we find little evidence about the consistency or pattern of firm performance effects on expectations of future returns. Finally, the past trend and pattern of growth do not lead to predictable returns following subsequent performance that confirms or contradicts this past trend.  

Abstract  

Herding behavior is among the most noticed biases in behavioral finance. This bias implies that investors unknowingly neglect personal information and analyses; instead they tend to follow other investors or the whole market. Using Tehran Exchange stocks transactions data, this study empirically examines herding behavior in this market. Two models based on cross-sectional deviation of stock returns and another model based on beta in state-space structure are utilized in our research. The sample covers 21,112 weekly returns and transaction volumes observations from April 2005 to April 2011. Our findings indicate that participants often lack independent investment decisions; i.e. they prefer following other investors’ decisions to taking an independent approach. This confirms that herding behavior exists in Tehran Stock Exchange. Moreover, evaluating comparative power of our three models suggests that the model based on beta is more powerful in explaining herding behavior than the other model.

Abstract As the amount of dividend is one of the effective and final factors in investor decision making, dividend policy of companies that decides the amount of dividend, could play a very important role in making decisions. Dividend policy is reciprocally influenced by individual behavioral specifications. These specifications have an extensive coverage from which a few has been selected for this study. By measuring limits of patience and loss aversion of shareholders this study is trying to analyze their relations with companies’ dividend policy.
The necessary data for trial assumptions, belonging to 77 companies for a period of 5 years between 1386-90, has been collected from Tehran stock exchange data base .the relation between variables have been examined by testing correlation coefficients and multiple regressions. Study shows there is no meaningful relation among percentage of dividend and stockholders patience, based on effecting date of dividend payment measurement and the amount of adjusted earnings per share measurement of companies accepted in Tehran stock exchange. On the other hand relation between dividend ratio percentage and measurement of companies profit growth is meaningful but to the contrary. So based on index of companies profit growth, a meaningful relation exists between payable dividend policy and loss aversion of shareholders. According to the market risk measurement, has no meaningful relation between loss aversion and dividend policy.
 Dividend Policy, Behavioral Finance, Patient Shareholders, Loss Averse Shareholders.
 Assistant

Abstract Two techniques which are used in stock markets and most investigators and market analysts used are contrarian and momentum investing strategies. These strategies help investors to predict future performance based on past   performance. Momentum investing strategy move the same way as the market stock move. In contrast, contrarian investing strategy acts vice versa. In this paper, for the period of 2005- 2007 different formation periods (1 to 6 month) and holding periods (1 to 36 month) are examined. The results show each strategy is profitable in a specific formation periods and holding periods. For formation periods from 1 to 4 month, momentum strategy is profitable and for formation periods from 5 and 6 month, contrarian strategy creates profitable portfolios. The best momentum effect is seen in formation periods of 4 month and holding periods of 36 month. Also, the best contrarian effect is seen in formation periods of 5 month and holding periods of 35 month.