学海拾珠系列之二百三十一：年报中的叙述性披露对公司价值的多维度影响

Quantitative Models and Construction Methods - Model Name: Dynamic Two-Step GMM Model Model Construction Idea: The model is designed to evaluate the impact of narrative-related disclosures on firm value while addressing endogeneity issues through dynamic panel data estimation [30][47] Model Construction Process: The model specification is as follows: $ Q_{ijt} = \beta_0 + \beta_1 Q_{ijt-1} + \beta_2 Tone_{ijt} + \beta_3 FLS_{ijt} + \beta_4 CS_{ijt} + \beta_5 CEO_{ijt} + \beta_6 Firm_{ijt} + \beta_7 BE_{ijt} + \epsilon_{ijt} $ - $ Q_{ijt} $ represents Tobin's Q for firm $ i $ in province $ j $ at time $ t $ - $ Tone_{ijt} $, $ FLS_{ijt} $, and $ CS_{ijt} $ are narrative-related disclosure variables (Tone, Forward-Looking Statements, Complexity) - $ CEO_{ijt} $ includes CEO characteristics such as age, gender, experience, and education - $ Firm_{ijt} $ includes firm-specific characteristics like age, size, and leverage - $ BE_{ijt} $ represents business environment quality variables The model uses dynamic two-step GMM estimation to address endogeneity issues, with Hansen-J test results confirming the validity of instruments [30][47][48] Model Evaluation: The model effectively captures the relationship between narrative disclosures and firm value, providing robust insights into the impact of tone, forward-looking statements, and complexity [48][51] Narrative Factors and Construction Methods - Factor Name: Tone Factor Construction Idea: Measures the overall sentiment conveyed in annual reports, focusing on positive, negative, and neutral tones [31][37] Factor Construction Process: - Sentences in annual reports are classified into positive, negative, or neutral categories using a dictionary-based approach and Naive Bayes algorithm - Positive sentences are assigned a value of 1, negative sentences -1, and neutral sentences 0 - The overall tone is calculated using the formula: $ TONE_{ii} = \frac{1}{K} \sum_{k=1}^{K} tone_{kii} $ - $ tone_{kii} $ represents the sentiment of sentence $ k $ in firm $ i $'s report - $ K $ is the total number of sentences in the report [37][38][39] Factor Evaluation: Tone is a significant positive driver of firm value, reflecting management optimism and investor confidence [48][51] - Factor Name: Forward-Looking Statements (FLS) Factor Construction Idea: Captures future-oriented information in annual reports, indicating strategic vision and growth potential [39][41] Factor Construction Process: - Sentences containing future-related keywords are identified using a dictionary-based filtering method - The overall FLS score is calculated using the formula: $ FLS_{it} = \frac{1}{S} \sum_{s=1}^{S} fls_{kit} $ - $ fls_{kit} $ represents the presence of forward-looking statements in sentence $ k $ - $ S $ is the total number of sentences in the report [41][42] Factor Evaluation: FLS positively impacts firm value by enhancing investor confidence in the company's strategic direction [51][54] - Factor Name: Complexity of Statements (CS) Factor Construction Idea: Evaluates the readability and complexity of annual reports, focusing on the potential impact on investor understanding [43][44] Factor Construction Process: - Complexity is measured using the Fog Index: $ Fog Index = 0.4 \times (Average \ words \ per \ sentence + Percentage \ of \ complex \ words) $ - Complex words are defined as technical terms frequently used in financial reporting [44][46] Factor Evaluation: Complexity generally has a weak negative impact on firm value, as overly complex disclosures may hinder investor understanding [52][62] Model Backtesting Results - Dynamic Two-Step GMM Model: - Tone coefficient: 0.4133 (significant positive impact) [50] - FLS coefficient: 0.0315 (significant positive impact) [50] - CS coefficient: -0.0095 (weak negative impact) [50] Factor Backtesting Results - Tone: - Tobin's Q: Positive impact (coefficient 0.4133) [50] - EPS: Positive impact (coefficient 0.8272) [60] - Cash Holding: Positive impact (coefficient 0.0258) [60] - Forward-Looking Statements (FLS): - Tobin's Q: Positive impact (coefficient 0.0315) [50] - Operating Cash Flow: Positive impact (coefficient 0.0043) [60] - Complexity of Statements (CS): - Tobin's Q: Weak negative impact (coefficient -0.0095) [50] - ROA: Weak negative impact (coefficient -0.0046) [60]