Independent monitoring across board and audit committee layers: Evidence from disclosure violations in Vietnam

Technical Summary: Independent Monitoring Across Board and Audit Committee Layers

Problem and Research Questions
This study addresses a gap in corporate governance literature regarding the joint monitoring role of the board of directors and its audit committee. While prior research often models board independence and audit committee independence as separate, additive sources of oversight, this paper argues that these bodies are organizationally nested (the committee is a specialized subgroup of the board). The study investigates whether the association between independent audit committee membership and regulator-observed disclosure violations depends on the level of board independence.

The research poses two specific questions:

1. RQ1: How does board independence condition the association between independent audit committee membership and regulator-observed disclosure violations?
2. RQ2: Is this conditional association present across the entire distribution of board independence, or is it concentrated in specific ranges?

The study focuses on Vietnam, utilizing the 2020 Law on Enterprises which specifies a board-level audit committee model. The outcome variable is not general reporting quality, but specifically "regulator-observed disclosure violations" (sanctioned breaches by the State Securities Commission), acknowledging that zero sanctions do not equate to verified compliance.

Methodology

• Data and Sample: The analysis uses an unbalanced panel of 539 firm-year observations from 150 Vietnamese firms recorded between 2021 and 2024. The sample is restricted to firms with recorded audit committee information. The dataset combines annual reports, governance reports, audited financial statements, and official enforcement decisions.
• Variables:
  • Dependent Variable: Annual count of disclosure breach instances formally sanctioned ($FDV$).
  • Independent Variables: Audit Committee Independence ($ACI$), measured as the count of independent members; Board Independence ($IB$), measured as the proportion of independent directors. Both are standardized for interaction analysis.
  • Controls: Firm size, age, profitability, leverage, Big Four auditor status, modified audit opinion, and year fixed effects. Sensitivity models include governance controls (committee size, meetings, board expertise) and ownership controls.
• Econometric Specification: The primary model is a Poisson pseudo-maximum-likelihood (PPML) regression with firm-clustered standard errors and year fixed effects. This specification accommodates non-negative count data with zeros and allows for within-firm dependence.
  • The core equation: $E(FDV_{it}) = \exp[\beta_0 + \beta_1 ACI_{it} + \beta_2 IB_{it} + \beta_3 (ACI_{it} \times IB_{it}) + \gamma'X_{it} + \lambda_t]$.
  • Hypothesis $H1$ predicts a positive interaction coefficient ($\beta_3 > 0$), implying that the negative association between $ACI$ and violations weakens as $IB$ increases (a substitution effect).
• Robustness and Diagnostics: The study employs extensive diagnostics, including inverse-probability weighting to address selection bias within the committee population, lagged specifications, negative binomial and binary logit models, and leave-one-firm-out influence checks. It also examines the interaction across different ranges of board independence (low, middle, high).

Key Results

• Interaction Effect: The primary PPML estimates reveal a statistically significant positive interaction between audit committee independence and board independence ($\beta = 0.501$, $p = .018$; Incidence Rate Ratio [IRR] = 1.650). This supports the directional prediction that the monitoring effectiveness of independent committee members diminishes as board independence rises.
• Conditional Associations:
  • Low Board Independence: At one standard deviation below the mean of board independence, an increase in independent audit committee members is associated with a 57.8% lower model-implied violation count ($IRR = 0.422$, $p = .001$).
  • Mean Board Independence: At the mean level, the association is negative but imprecise ($p = .061$).
  • High Board Independence: At one standard deviation above the mean, the association becomes positive and statistically insignificant ($IRR = 1.148$, $p = .658$), though the authors note this does not establish that independence is harmful.
• Sensitivity and Precision: The positive interaction holds across various specifications (governance controls, ownership controls, lagged variables, and inverse-probability weighting). However, the precision of the estimate is highly dependent on the sample composition:
  • The interaction becomes imprecise when restricting the sample to the central range of board independence (10th–90th percentile).
  • The interaction becomes imprecise when excluding the single most influential firm.
  • Alternative models (Negative Binomial, Firm Fixed Effects) show weaker evidence or imprecision, qualifying rather than confirming the main finding.

Significance and Claims
The paper makes a modest, bounded contribution to the literature:

1. Theoretical Contribution: It challenges the additive assumption in governance models by demonstrating that the effectiveness of independent audit committee membership is conditional on the broader board structure. The findings are consistent with a "substitution" mechanism, where independent committee members provide a distinct reporting channel primarily when board-level independent representation is scarce. However, the authors explicitly state they do not identify the underlying organizational process (e.g., information exchange vs. coordination failure) as the data is archival.
2. Empirical Evidence: The study provides enforcement-based evidence from Vietnam, a setting where the 2020 law explicitly defines the board-committee relationship. It highlights that the location and classification of independent membership matter for disclosure accountability.
3. Limitations and Cautions: The authors emphasize that the estimates are provisional. They do not establish a causal mechanism, nor do they prove that changing director appointments improves compliance. The study does not claim that extensive independence is harmful, nor does it suggest a numerical ceiling for board independence. The findings are limited to firms with recorded committee data and cannot be generalized to firms without such structures or to jurisdictions with different enforcement regimes. The paper concludes that the association is estimated most precisely at lower levels of board independence, and the results should be interpreted as a conditional relationship rather than a universal governance rule.