Non-financial information farsightedness and capital market information efficiency

The study investigates whether and how the forward-looking (farsighted) content of firms’ non-financial disclosures enhances capital market information efficiency in China’s A-share markets. Motivated by persistent information asymmetries and the growing importance of qualitative disclosures, the authors leverage regulatory changes encouraging forward-looking statements and advances in text analytics to quantify disclosure foresight. Grounded in signaling theory and debates around stock price synchronicity as a proxy for information efficiency, the paper posits two competing hypotheses: H1a that more positive forward-looking non-financial information lowers synchronicity (improves information efficiency) by incorporating firm-specific information into prices; and H1b that it could instead spur noise trading and herding, increasing synchronicity. The study also hypothesizes (H2) that analyst coverage strengthens the positive impact of forward-looking non-financial disclosures on information efficiency.

The review highlights that stakeholders increasingly demand forward-looking disclosures, yet practices vary widely across firms and contexts. Prior work emphasizes forward-looking financial information (e.g., MD&A tone, frequency, and credibility), its determinants (e.g., firm size, leverage, gender diversity), and its behavior under uncertainty and across cultural/institutional settings. Research shows forward-looking narratives can influence stock returns and credibility, and investors rely more on such disclosures when audited earnings quality is high. However, non-financial forward-looking information remains underexplored regarding market information efficiency. On information efficiency, the literature identifies media coverage, regulation, analyst forecasts, reporting standards (e.g., XBRL), and financing mechanisms as drivers, affecting stock price synchronicity and firm-specific information incorporation. The paper fills a gap by linking non-financial forward-looking disclosure to capital market information efficiency and exploring moderating and mediating mechanisms.

Data: Chinese A-share listed firms (Shanghai and Shenzhen) from 2007–2022. Exclusions: financial industry; ST/*ST and delisted firms; leverage ratio > 1; firms with < 5 consecutive years of data; firm-years with missing data. Continuous variables winsorized at 1% and 99%. Financial and market data from CSMAR; non-financial disclosure texts from Shenzhen/Shanghai Stock Exchange and listed firms’ websites. Text collection via Python; text analysis via NLPIR.

Key variables: Capital market information efficiency measured by stock price synchronicity (SYNCH). For each firm-year, estimate weekly-return regression: R_it = α + β1 R_mt + β2 R_it-1 + (industry return may be included per CSRC industry classification), then transform R^2 to SYNCH = ln(R^2/(1−R^2)). Alternative measure SYNCH_2 uses total market value weights.

Non-financial farsightedness: Using Loughran–McDonald (LM) dictionary, compute forward-looking tone based on counts of positive and negative words in non-financial text disclosures: (pos − neg)/(pos + neg), ranging in [-1,1]; higher indicates stronger positive forward-looking content. Robustness measures: (i) NTU dictionary-based (Nonfinancial_farsightedness_2); (ii) LM-based scaled by total annual report words (Nonfinancial_farsightedness_3).

Analyst coverage (Follow): log(1 + number of analysts/teams covering the firm in a year). Controls: firm size, age, leverage, ROA/ROE, book-to-market, growth, turnover (VOL), ownership structure (Top1, Shrcr, InsInvestor), governance (Board size, Dual), audit quality (Big4, Audit), year and industry fixed effects (see Table 2 for definitions).

Models: Four-step multilevel regressions. (1) Baseline: SYNCH on controls with year/industry FE. (2) SYNCH on Nonfinancial_farsightedness with FE (no controls). (3) SYNCH on Nonfinancial_farsightedness plus full controls and FE. (4) Moderation: add Follow and interaction Nonfinancial_farsightedness × Follow. Robustness: alternative SYNCH measure; alternative farsightedness measures; instrument variable (industry-year median farsightedness as IV) with 2SLS; Heckman two-stage selection (binary split of farsightedness around mean to compute IMR); models with lagged explanatory variables. Diagnostics: VIF < 2 (mean 1.54); White’s test p=0.0569 (no heteroskedasticity); autocorrelation test p=0.0741 (no autocorrelation). Descriptives: 4,926 firms; 15,830 firm-years; mean SYNCH 0.435; mean Nonfinancial_farsightedness 0.002.

• H1 (main effect): Higher forward-looking non-financial tone is associated with significantly lower stock price synchronicity, implying higher information efficiency. Reported coefficients for Nonfinancial_farsightedness on SYNCH are negative and statistically significant (e.g., about −0.0498 to −0.0463 at the 1% level in baseline models; Table 5). Robustness using alternative SYNCH_2 yields −0.0524 to −0.0634 (p<0.01); alternative text measures also negative and significant (e.g., −0.0319 to −0.0304, p<0.01).
• H2 (moderation by analyst coverage): The interaction Nonfinancial_farsightedness × Follow is negative and significant (e.g., −0.0290 to −0.0433; p<0.05 to p<0.01), indicating that analyst attention strengthens the reduction in synchronicity from forward-looking non-financial disclosures (Table 6). In some specifications, Follow itself is associated with lower SYNCH (e.g., −0.0894, p<0.01).
• Mechanisms (mediation): • Analysts’ earnings forecast errors/bias: Nonfinancial_farsightedness is associated with changes in analyst forecast characteristics (Table 7 shows coefficient around 0.0355, p<0.05), evidencing an effect on forecast bias/accuracy that mediates the link to information efficiency. • Audit opinions: Nonfinancial_farsightedness reduces the likelihood of non-standard audit opinions (Audit_OP coefficient ≈ −0.1721, p<0.01), which is associated with higher information efficiency. • Corporate financialization: Nonfinancial_farsightedness is negatively related to firms’ financial asset allocation (Fin coefficient ≈ −0.0149, p<0.01), which in turn supports higher information efficiency.
• Heterogeneity: • Media attention: The negative effect of Nonfinancial_farsightedness on SYNCH is significant in high media attention firms (≈ −0.0399, p<0.01) but insignificant in low media attention firms. • Internal control quality: Significant negative relation for firms with high internal control quality (≈ −0.0433, p<0.01), insignificant for low internal control quality firms.
• Endogeneity checks: IV (industry-year median farsightedness) 2SLS confirms main results; Heckman two-stage (IMR included) and lagged explanatory variables maintain sign and significance, supporting robustness.

Findings support the signaling view that forward-looking non-financial narratives embed firm-specific information that gets impounded into prices, lowering synchronicity and improving information efficiency. Analysts amplify this channel: where coverage is higher, analysts parse forward-looking content more effectively and disseminate insights, further reducing synchronicity. The identified mechanisms suggest that forward-looking non-financial disclosure relates to audit outcomes (fewer non-standard opinions), corporate investment orientation (less financialization), and analysts’ forecasting behavior, each influencing how firm-specific signals translate into prices. Heterogeneity indicates institutional information environments matter: stronger media attention and higher-quality internal controls improve the credibility, reach, and market processing of forward-looking non-financial disclosures. Together, the results align with a more informative market price process when firms provide credible, forward-looking non-financial information within supportive external (media, analysts) and internal (governance, controls) environments.

The study documents that forward-looking non-financial disclosures by Chinese A-share firms from 2007–2022 reduce stock price synchronicity and enhance capital market information efficiency. Analyst coverage strengthens this beneficial effect, while mechanisms operate via analysts’ forecast behavior, audit opinions, and firms’ financialization choices. Effects are more pronounced among firms with higher media attention and better internal controls. Contributions include (i) extending the literature on non-financial disclosure by focusing on its forward-looking dimension and market information efficiency, (ii) developing machine learning-based measures of disclosure farsightedness, and (iii) identifying practical levers—enhanced disclosure quality, analyst engagement, and governance—to improve information efficiency. Future research should refine forward-looking measurement tailored to local contexts, explore investor interpretation of narrative foresight, examine analysts’ governance role in the information environment, investigate how corporate financialization interacts with disclosure effects, and assess how improvements in internal and external governance mitigate information asymmetry.

• External validity: Non-financial text data are primarily from listed firms; private firms’ voluntary disclosure practices differ, limiting generalizability.
• Measurement/model risk: Machine learning choices (dictionaries, models, parameters) may affect stability and accuracy of the farsightedness measure.
• Analyst attention metrics: Proxies for coverage can be influenced by subjective factors and may not fully capture analysts’ interpretation quality or market impact.