Better or worse? Revealing the impact of common institutional ownership on annual report readability

The paper investigates whether common institutional ownership (CIO)—situations where institutional investors hold stakes in multiple firms within the same industry—improves or impairs the readability of corporate annual reports (ARR) in China. In the information age, narrative disclosures in annual reports play a crucial role alongside numerical data, yet can be manipulated to obscure negative information. Given China’s evolving market integrity and regulatory frameworks, understanding how ownership structures like CIO influence textual transparency is important for investor protection and market efficiency. The study sets up competing hypotheses: CIO may enhance ARR via synergistic governance and superior monitoring, or decrease ARR through collusive manipulation and investor limited attention. Using Chinese listed firms (2007–2021), the authors examine whether CIO drives more transparent reporting or strategic obfuscation, and explore mechanisms and boundary conditions.

Two streams of CIO literature are reviewed: (1) Synergistic governance perspective—CIO can enhance firm outcomes by facilitating information sharing, monitoring, and coordinated governance across portfolio firms, reducing earnings management, managerial rent extraction, and improving efficiency and innovation (He and Huang, 2017; Ramalingegowda et al., 2021; Chen et al., 2023; Bai et al., 2023; Li and Liu, 2023). (2) Collusive manipulation perspective—CIO may induce collusive behaviors across portfolio firms, raising market power, hindering digital transformation, and lowering CSR (Azar et al., 2018; Cheng et al., 2022; Wang et al., 2023). For ARR, poor readability is linked to concealment of negative information and adverse outcomes including opportunistic insider sales, equity mispricing, reduced trade credit financing, and altered stock return synchronicity (Yin et al., 2024; Chen et al., 2023; Li et al., 2024; Gangadharan and Padmakumari, 2023). Determinants of ARR studied include board independence, organizational capital, CEO power, and CSR, but few examine shareholder network characteristics such as CIO. This study fills that gap by analyzing CIO’s effect on ARR and testing competing hypotheses (synergy vs. collusion/limited attention).

Data and sample: Chinese A-share listed firms from 2007–2021, excluding financial industry firms, ST and *ST firms, and observations with missing key data, yielding 35,916 firm-year observations. Data sources: CSMAR and WinGo financial text database. Continuous variables are winsorized at 1% (two-sided). Measures: ARR is computed via the sentence production probability method (Shin et al., 2020): Readability = (1/N) Σ log(Pi), where Pi is the sentence generation probability. A higher value indicates higher readability. CIO is measured at three levels: Coz1 (dummy for existence of CIO in a year, based on whether institutional investors holding >5% in other listed firms in the same industry also hold >5% in the focal firm), Coz2 (degree of CIO linkage: ln(1 + number of common institutional investors) averaged quarterly to annual), and Coz3 (CIO shareholding ratio: annual average of the sum of common institutional investors’ shareholding ratios). Controls include Dual, growth, size, BoardSize, Indep, Insit, cfo, ppe, lev, btm, Shrcr1, Big4, and soe. Empirical strategy: Baseline regressions use OLS with industry and year fixed effects. Robustness and endogeneity checks include: (1) Propensity score matching (PSM) with one-to-one nearest neighbor matching on covariates, followed by OLS (PSM-OLS); (2) Two-stage least squares (2SLS) using HS300 index membership as an instrument for CIO (following Gao et al., 2019); (3) First-difference model regressing changes in readability on changes in CIO measures and controls; (4) Using T+1 readability as the dependent variable; (5) Shortening the sample by excluding 2007–2008 to mitigate global financial crisis effects. Additional analyses: Moderation tests for analyst attention (Atten, based on number of analysts; high vs. low by year-industry median), industry concentration (Concern, based on HHI above median), and media coverage (News, based on media mentions above industry median). Mediation analysis: operational risk (Risk) measured via the cumulative distribution probability of the rolling 4-year standard deviation of pre-tax depreciation and amortization profit margin (Wang et al., 2017). Three-step regressions assess whether Risk mediates the CIO–ARR link.

Descriptive statistics show negative correlations between CIO measures and ARR at the 1% level. Baseline OLS (industry and year FE): • Coz1: -0.230 (t=-5.532) • Coz2: -0.322 (t=-5.603) • Coz3: -0.709 (t=-5.354), all significant at 1%, indicating CIO reduces ARR. Robustness: • PSM average treatment effect on Readability: -0.214 (1%). PSM-OLS shows Coz2 -0.266*** and Coz3 -0.529***. • 2SLS with HS300 as IV: first stage shows HS300 predicts CIO measures at 1%; second stage shows Coz1, Coz2, Coz3 remain significantly negative for ARR. • Difference model: DCoz2 -0.231*** and DCoz3 -0.610** for DReadability. • Using T+1 Readability: Coz1 -0.222***, Coz2 -0.313***, Coz3 -0.650***. • Shortened sample excluding 2007–2008: Coz1 -0.232***, Coz2 -0.329***, Coz3 -0.772***. Moderation: • Analyst attention strengthens the negative effect: interaction Coz3×Analyst is -0.038*** in the full sample; effects are more negative in the high-analyst-following group. • Industry concentration strengthens the negative effect: Coz3×Concern is significantly negative at 5%, indicating stronger reductions in ARR when HHI is higher. • Media coverage strengthens the negative effect: Coz3×News is negative and significant (at 10% in reported tests), with stronger negative effects under higher media coverage. Mediation: • Coz3 → Readability: -0.673*** (direct effect). • Coz3 → Risk: 0.042**. • Risk → Readability: -0.525***; with Risk included, Coz3 → Readability becomes -0.651***, indicating partial mediation. Overall, evidence supports the collusive manipulation/limited attention view (H1b): CIO is associated with lower annual report readability.

The findings indicate that common institutional ownership is associated with reduced readability of annual reports, supporting the collusive manipulation and limited attention hypotheses rather than a synergistic governance effect. CIO appears to facilitate coordinated behaviors that prioritize portfolio-level gains and potentially short-term performance, leading to strategic obfuscation in narrative disclosures. Limited attention among diversified institutional investors further weakens oversight, enabling managerial self-interest and information masking. Boundary conditions show that the negative CIO–ARR relationship intensifies under greater external scrutiny or competitive pressure—namely high analyst attention, higher industry concentration (greater potential for coordinated market power), and higher media coverage. The mediation analysis suggests a mechanism where CIO increases operational risk, which in turn decreases readability, consistent with heightened agency conflicts and managerial opportunism. These results underscore CIO’s potential adverse implications for transparency and investor decision-making in emerging markets like China, suggesting that ownership networks can influence nonfinancial disclosure quality. They also inform regulatory focus on narrative disclosure practices in settings where CIO is prevalent.

This study analyzes Chinese listed firms (2007–2021) and documents that common institutional ownership reduces annual report readability, with results robust to extensive checks (PSM, 2SLS, difference models, alternative timing, and sample periods). The negative effect is stronger when analyst attention, industry concentration, and media coverage are high, and operating risk partially mediates the relationship. Contributions: (1) Extends the literature on CIO’s economic consequences to nonfinancial textual disclosures, offering evidence consistent with collusive manipulation in China’s capital market; (2) Provides a network/ownership perspective on antecedents of annual report readability; (3) Uncovers mechanisms (operational risk mediation) and boundary conditions (analyst attention, industry concentration, media coverage). Practical implications include encouraging firms to simplify and clarify narrative disclosures, regulators to monitor readability especially in high-CIO contexts, and investors to be cautious of low readability as a potential red flag. Future research could refine readability measurement in Chinese-language contexts, and examine the broader economic consequences of diminished readability and the heterogeneity across industries, ownership types, and regulatory regimes.

The study acknowledges challenges in measuring readability in Chinese due to limited language-tailored metrics, suggesting the readability index may lack a comprehensive linguistic foundation. Additionally, while the paper verifies that CIO is linked to lower readability via collusive manipulation, the broader economic ramifications of diminished readability in China remain underexplored. Future research should develop improved Chinese-language readability measures and investigate the consequences of low readability for capital allocation, market efficiency, and firm outcomes.