The narrative content of annual reports has become increasingly important as a source of information for stakeholders beyond traditional financial data. This is particularly true in the wake of major accounting scandals, highlighting the need for a more nuanced understanding of corporate communication. Managers use narrative disclosures to convey information about company performance, risks, and sustainability, potentially influencing investor perceptions. Prior research demonstrates a link between narrative tone and profitability as well as bankruptcy risk, primarily focusing on Western developed markets. This study contributes new evidence by examining this relationship in the context of Vietnam, an emerging East Asian market with distinct cultural and linguistic characteristics. Specifically, it addresses the following research questions: 1. How does the narrative tone of annual reports respond to profitability, bankruptcy risk, and the COVID-19 pandemic? 2. Are there differences in narrative tone between profitable and unprofitable companies? 3. Are there differences in narrative tone between financially distressed and non-distressed companies? 4. How does the narrative tone of annual reports change before, during, and after the pandemic? The study distinguishes itself from previous research in three key ways: (1) It focuses on an emerging East Asian market (Vietnam), acknowledging significant cultural and linguistic differences compared to Western contexts. (2) It develops a lexicon of Vietnamese words and phrases expressing emotional nuances in finance and accounting, a unique contribution to the field. (3) It compares narrative tone across diverse company groups (profitable vs. unprofitable, bankruptcy risk vs. no risk, and different pandemic phases). The data comprises 4,212 annual reports from 468 Vietnamese non-financial companies spanning nine years (2015-2023), encompassing the pre-, during, and post-pandemic periods.

The "narrative turn" in social science research, including accounting, emphasizes the importance of analyzing the narrative content of corporate disclosures. Studies have shown that narrative tone can predict profitability and bankruptcy risk. Several methods exist for measuring narrative tone, including dictionary-based approaches and machine learning techniques, mostly focusing on English-language texts. This study uses a novel approach by developing a Vietnamese-language lexicon of positive and negative words to analyze the tone. Prior research shows conflicting results on the relationship between narrative tone and various firm characteristics such as profitability, financial distress, stock returns, and size. Some suggest a positive correlation between profitability and positive tone while others point to the use of impression management where firms may manipulate the tone of disclosures to portray a more favorable image. Studies examining the impact of the pandemic on corporate disclosures are also emerging, pointing to varied emotional tones in communication with stakeholders. Existing research regarding the relationship between profitability and narrative tone largely focus on Western markets, leaving a gap for research in an emerging Eastern market like Vietnam.

This study collected data from 468 non-financial companies listed on the Hanoi Stock Exchange (HNX) and Ho Chi Minh City Stock Exchange (HOSE) from 2015 to 2023. The data included 4,212 annual reports, totaling 250,013 pages. Annual reports were converted into plain text using OCR. The researchers created a Vietnamese lexicon of positive and negative words, incorporating translations of existing English lexicons and adding terms identified through analysis of 200 annual reports. Python was used to calculate the frequency of positive and negative words, generating the variables POSTONE, NEGTONE, and NETTONE. The primary regression model included independent variables representing return on assets (ROA), profitability (PLOS – a dummy variable for negative ROA), bankruptcy risk (DIST – a categorical variable based on Z-score), and pandemic phase (COVID – a categorical variable representing pre-, during, and post-pandemic periods). Control variables included firm size, age, audit quality, stock return, stock return volatility, book-to-market ratio, liquidity, leverage, firm growth rate, accruals, and report length. The study employed OLS, REM, and FEM regression methods, followed by FGLS to address heteroscedasticity and autocorrelation. Bootstrapping and SGMM were used to test robustness and account for endogeneity and dynamic relationships. A Hausman test helped select the most appropriate regression model for each dependent variable (POSTONE, NEGTONE, NETTONE). Separate models were also created to test the moderating effects of bankruptcy risk and the pandemic on the relationship between profitability and narrative tone.

The analysis revealed significant relationships between profitability and narrative tone. High profitability was associated with a positive tone and a lower negative tone, supporting signaling and agency theories. Unprofitable firms used significantly more negative tone and less net tone compared to profitable firms, while no significant difference was found in positive tone. Financially distressed firms also showed a significantly more negative tone and less net tone than non-distressed firms. Regarding the pandemic, pre-COVID-19 firms had less positive and negative tone but higher net tone. During the pandemic, positive and negative tone increased, while after stabilization positive tone rose, and negative tone fell. The moderating effects of bankruptcy risk and the COVID-19 pandemic were assessed using interaction terms in the regression models. The results showed that bankruptcy risk significantly moderated the relationship between profitability and negative tone: in high-bankruptcy-risk firms, decreasing profitability led to a sharper increase in negative tone. However, the pandemic did not significantly moderate the relationship between profitability and narrative tone. Bootstrapping and SGMM analyses confirmed the robustness of the main findings, showing a significant influence of the preceding year's tone on current year's tone. Additional analyses showed that firm size, book-to-market ratio, liquidity, leverage, and firm growth rate also had significant relationships with narrative tone, along with audit quality and report length.

The findings support several theoretical frameworks. The positive association between profitability and positive tone aligns with signaling theory, where firms with strong performance use positive language to signal their quality. The increased negative tone among financially distressed firms aligns with agency theory, reflecting managers' response to poor performance and financial risk. The impact of the pandemic highlights the importance of considering macroeconomic factors in understanding corporate disclosure choices. The absence of a consistent effort to manage impressions during periods of financial distress and during the pandemic is noteworthy, suggesting that managers might prioritize accurate information conveyance over carefully crafted messages to boost investor confidence when faced with significant adversity. The study’s use of a Vietnamese language lexicon and its focus on an emerging market add to the growing literature on cross-cultural differences in financial reporting practices and their impact on narrative tone.

This study provides empirical evidence on the relationship between profitability, bankruptcy risk, pandemic, and narrative tone in annual reports of Vietnamese firms. The results offer practical insights for investors and analysts interpreting management disclosures, particularly in an emerging market setting. Further research could expand the scope to include other emerging markets, analyze more nuanced aspects of narrative content such as readability and storytelling, and utilize advanced text analysis techniques like deep learning. Investigating the predictive power of narrative tone for future performance, incorporating additional contextual factors, and analyzing the impact of different regulatory environments are potential future research directions.

The study's limitations include the sample being confined to Vietnamese non-financial companies, potentially limiting the generalizability of the findings. Furthermore, the focus on narrative tone alone restricts the investigation into other aspects of narrative content that could provide additional insights. The methodology relies on a specific lexicon developed for Vietnamese annual reports, which may not directly translate to other linguistic contexts.