We propose that corporate directors are in greater need of soft information about the firm when the quality of hard accounting information is low. We further propose that board meetings constitute a key opportunity for corporate directors to gather soft information about the firm, and empirically investigate the relationship between financial reporting quality and the board's soft information gathering, as revealed by board meeting frequency. Consistent with expectations, we find that boards meet more frequently when accruals quality is low. We further find that the proportion of outside directors, insider ownership, and SOX regulation moderate this relationship. The evidence is reinforced by analysis of management earnings forecasts, financial restatements, and internal control weaknesses and is robust to several alternative earnings quality specifications. Additional empirical tests suggest that our results are incremental to the alternative explanation of increased meeting frequency to address problems in the reporting process per se. We conclude that corporate directors seek more frequent board meetings as an alternative information source to low earnings quality.
Notes
We would like to thank Roni Michaeli, Juan Manuel Garcia Lara, Elizabeth Demers, Haresh Sapra, Andreas Milidonis, Marios Panayides, seminar participants at the University of Cyprus workshop, and an anonymous reviewer for their numerous useful comments and suggestions on earlier drafts of this paper. The University of Cyprus provided funding for this project.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
Financial reporting information is a core component of a firm's overall information environment and a key source of hard firm-specific information for boards of directors (e.g., Armstrong et al. 2010). In this paper, we posit that financial reporting quality affects the amount of soft information demanded by the board. We propose that board meetings constitute the best opportunity for soft information gathering by members of the board of directors and hypothesize that the demand for soft information is associated with the quality of hard accounting information.
Our hypothesis builds on prior academic research on the determinants of board structure. Several empirical studies, across a variety of research designs and samples, find evidence to suggest that firms endogenously adjust their board size and composition to account for information gathering and processing costs (e.g., Boone et al. 2007; Linck et al. 2008; Duchin et al. 2010; Cai et al. 2009). Our study extends this literature by using board meetings as a dimension of board structure that we expect to be more responsive to changes in the information environment than other relatively time-invariant board characteristics (Adams et al. 2021), and by suggesting that board meetings constitute an alternative source of information gathering by directors in carrying out their oversight responsibilities (Brickley and Zimmerman 2010; Kim et al. 2014; Alam et al. 2014; Ji et al. 2020).
Advertisement
Our hypothesis also builds on prior accounting research that investigates the effects of earnings quality measures on governance mechanisms and other formal contracts, such as compensation and financing agreements, as well as CEO termination decisions (e.g., Bushman et al. 2004; Armstrong et al. 2010; Chaigneau and Sahuguet 2023). While there is ample evidence that accounting earnings are used extensively in formal contracts, we know little about how earnings quality affects the board's internal workings. We pursue this inquiry and use board meetings as an informal contract that defines the multiperiod relation between the CEO and directors, contributing to the literature by studying the link between earnings quality and board meeting frequency.
We empirically address the relation between financial reporting quality and board meeting frequency on a sample of 2721 Russell 3000 firms between 1996 and 2017. We argue that firm-specific information on cash flows is pertinent to the board's responsibilities and identify accruals quality, which measures the mapping of accounting earnings into cash flow realizations (Dechow and Dichev 2002), as our primary financial reporting quality measure. We hypothesize that accruals quality facilitates the board's advising and monitoring responsibilities by providing a more precise measure of a firm's investment opportunity set and reducing information asymmetries between inside and outside directors. Accordingly, we predict that where accruals quality is low, boards will substitute for more costly information-gathering activities, as revealed by an increase in board meeting frequency.
While our empirical evidence confirms our expectations that boards meet more frequently when accruals quality is lower, importantly, this relationship is at least as pronounced at the high end of the earnings quality continuum. Further, we decompose accruals quality into innate and discretionary components following Francis et al. (2005) and find that the innate, rather than the discretionary, accruals component mostly drives our results. Both of these results suggest that the observed effects are not a by-product of simply calling more board meetings to address problems in accounting quality but suggest that the accounting system's inherent limitations in capturing the firm's economic performance increase the boards' demand for soft information through board meetings.
Next, we examine whether observable firm characteristics that vary cross-sectionally and proxy for access to information moderate the association between earnings quality and board meetings. We find that board meeting frequency is more sensitive to earnings quality in the presence of a greater proportion of non-executive directors. Since non-executive directors are less informed and more likely to rely on hard information for their information needs, this finding reinforces our interpretation that boards meet more in response to poor earnings quality. We also find a stronger response to earnings quality through board meetings in firms with more dispersed ownership, where board monitoring is likely to be more prominent in the overall governance structure. Finally, we find a more pronounced effect in the post-SOX period, consistent with SOX accentuating the value of hard accounting information.
Advertisement
In further tests, we show the robustness of our results to several alternative financial reporting quality measures. We also find that boards meet more frequently when there is a greater disagreement between management and analyst forecasts and presumably a more uncertain information environment. Finally, we find that boards meet more frequently in the period following a financial restatement and around material weaknesses in internal controls.
To isolate the relationship between earnings quality and board meetings, we rely on the variation that is orthogonal to the determinants of earnings quality and board meetings, as well as a large number of fixed effects; firm and year fixed effects in the main tests, and industry-year interaction and CEO-firm fixed effects in further robustness tests. Thus, our findings hold beyond unobserved time-invariant firm characteristics, macroeconomic conditions that affect all firms in a given industry and year, and unobserved time-invariant CEO-firm heterogeneity that can affect both earnings quality and board meetings. Finally, given that the thrust of our results is based on association tests, to address the challenge of correct identification and the direction of causality, we employ General Method of Moments tests and find evidence that is also consistent with our inferences. Together, these results corroborate the view that accounting information quality is associated with the intensity of board information seeking through increased meeting frequency.
Our study makes a useful contribution to several constituents. Academically, this study adds to previous accounting research advancing earnings quality as a determinant of corporate governance characteristics, investment and dividend policies, debt and compensation contracts, and a long line of finance research attempting to understand the determinants of board structure. For practice, this study provides a new perspective to board chairpersons, CEOs, board secretaries, and corporate directors by highlighting the importance of delivering high-quality accounting information to corporate directors and the need to facilitate soft information gathering by directors. Finally, the results provide investors and creditors with a new prism through which to view information disclosure on board activity.
The remainder of this study is organized as follows: Sect. 2 discusses the background literature; Sect. 3 develops the hypothesis; Sect. 4 discusses research design methods and variable measurement; Sect. 5 presents and discusses the main empirical results; Sect. 6 presents further tests; and Sect. 7 concludes.
2 The role of information in board workings
The board of directors has a broad range of monitoring and advising responsibilities (Raheja 2005; Adams and Ferreira 2007). The monitoring function requires directors to scrutinize management to guard against harmful behavior and ensure that the firm's activities are consistent with its targets and goals. The board's advising function involves helping management make good resource allocation decisions. To effectively exercise their monitoring and advising responsibilities, directors need firm-specific information to help them understand how and why equity values change (e.g., Bushman and Smith 2001; Bushman et al. 2004; Armstrong et al. 2010; Duchin et al. 2010). Importantly, monitoring performance requires accurate information about the CEO's ability and effort and timely information about deviations from the firm's plans. Likewise, advising performance depends on the completeness and precision of information about the firm's investment opportunity set (Forbes and Miliken 1999; Armstrong et al. 2010).
Financial accounting information is a major source of hard (verifiable) firm-specific information that can help the board of directors perform its advising and monitoring responsibilities effectively (e.g., Weisbach 1988; Bushman and Smith 2001; Bushman et al. 2004). Financial statements convey credible, low-cost information about the firm's financial position, financial performance, and changes in financial position, forming the foundation of firm-specific information available to directors in performing their responsibilities (Bushman et al. 2004).
However, due to their inherent limitations, financial statements cannot provide the complete information set required by directors. Thus, directors use soft or non-verifiable information from pertinent sources to complement financial accounting information and filter out noise in hard performance data (e.g., Brickley and Zimmerman 2010; Versano 2021). Liberti and Petersen (2019) define soft information as information that cannot be codified and transferred, requires contextual knowledge to understand fully, and becomes less useful when separated from the environment in which it is collected. Notably, soft information can only be acquired from personal observation or face-to-face interactions (e.g., Stein 2002; Liberti and Petersen 2019).
Given that director time is a valuable resource and that boards meet only as necessary, we postulate that board meetings constitute the best opportunity for soft information gathering by members of the board of directors. Preparing for and attending board meetings is the quintessential activity through which board members acquire the necessary knowledge to make informed decisions about strategy, monitor the management team, and develop succession plans (Brickley and Zimmerman 2010; Adams et al. 2021). Directors can obtain soft information about the firm by reading material supplied by management as they prepare for board meetings and through interactions with other directors, the management team, and other employees during meetings.1 We, therefore, posit that as boards meet more frequently, directors have greater exposure to soft information about the firm.2
Prior research provides ample evidence in this direction, highlighting both the critical role of soft information in aiding directors with their advising and monitoring responsibilities and the value of board meetings in enhancing the amount of information to which directors are exposed. For example, Cornelli et al. (2013) show that soft information plays a more significant role than hard information in CEO replacement decisions. Kim et al. (2014) and Fernandes and Fich (2023) find that soft information accumulated through increased director tenure on the board, a measure highly correlated with the number of board meetings attended, significantly improves board advising effectiveness.
In a similar spirit, Quan and Zhang (2021) find that geographically distant directors participate in fewer board meetings and are less effective monitors. Finally, Fich and Shivdasani (2005) document the declining effectiveness of busy and overstretched directors, consistent with their constrained ability to interact with other board members and acquire the soft information needed to carry out their tasks. Firms also appear to consider the vital role of board meetings in making board appointment decisions. For example, when the quality of hard information is low, firms are more likely to appoint unaffiliated directors who live near the corporate headquarters (Alam et al. 2014), and less likely to reappoint directors who relocate further from headquarters (Quan and Zhang 2021).3
3 Hypothesis development
We hypothesize that financial reporting quality, which captures the quality of hard information available to the board, affects the amount of soft information demanded by the board, as revealed by board meeting frequency. In developing our hypothesis, we adopt the perspective that financial reporting quality is a component of a firm's overall information environment (e.g., Bushman et al. 2004).
More specifically, we focus on earnings quality as the premier source of financial reporting quality, assuming that earnings is the most important measure on which directors rely to perform their advising and monitoring responsibilities. Our assumption is supported by extensive empirical research documenting the importance of earnings in firm valuation and corporate governance contexts (e.g Biddle et al. 1995; Liu et al. 2002). We argue that firm-specific information on cash flows is pertinent to the board's advising and monitoring responsibilities and identify the accruals quality measure introduced by Dechow and Dichev (2002), as our primary earnings quality characteristic.
In the Dechow and Dichev (2002) model, accruals quality is measured by the extent to which working capital accruals map into prior, current, and future period cash flow realizations. The intuition of this model is that the accruals component of earnings alleviates the mismatching of current cash flow realizations to the economic performance by shifting cash flows across time periods. However, the benefit of accruals comes at the cost of making assumptions and estimates that must be corrected in future accruals. Estimation errors add noise to earnings and reduce the precision by which earnings measure economic performance. Thus, accruals quality is defined as the extent to which accruals map into cash flow realizations, where a better match signifies high accruals quality.
We argue that better accruals quality facilitates the board's advising and monitoring activities by providing a more precise measure of investment-generated future cash flows and by reducing information asymmetry between inside and outside directors, thus enhancing the board's ability to monitor top management. In line with this contention, prior research has linked accruals quality to the effectiveness of various operating dimensions over which boards have a responsibility, including investment efficiency (e.g., Biddle et al. 2009) and dividend policy (e.g., Ramalingegowda et al. 2013). Further, prior research has linked accruals quality to lower information risk (e.g., Francis et al. 2005) and lower information asymmetry among market participants (e.g., Aboody et al. 2005; Bhattachayarya et al. 2012). Against this backdrop, we predict that when the quality of hard accounting information, as proxied by accruals quality, is low, the board of directors will substitute for soft information, obtained through increased director contact and measured by the frequency of board meetings.
This prediction parallels results supporting the notion that firms substitute towards costly monitoring by large shareholders when the information provided by the financial accounting system is low (e.g., Bushman et al. 2004). It is also consistent with evidence suggesting that traders increase costly, private information gathering and processing activities as the precision of accounting disclosures diminishes (e.g., Verrechia 1982), that firms prefer private over public debt when earnings quality is poor (Bharath et al. 2008), and that incentive plans rely relatively more on non-accounting performance measures when the quality of financial accounting information is poor (e.g., Ittner et al. 1997; Hayes and Schaefer 2000). More formally, our hypothesis is as follows:
Research hypothesis: As accruals quality decreases, boards of directors increase their demand for soft information, as revealed by an increase in board meeting frequency.
4 Research design and variable measurement
We rely on OLS methodology for our main specification and estimate the following cross-sectional regression model to examine how earnings quality affects board meetings:
Board meetings, our main variable of interest, is defined as the number of board meetings in a year, including teleconferences but excluding board actions by written consent, and is consistent with the definition used by the MSCI GMI Ratings database. We log-transform board meetings in our main tests, assuming that the efficacy of board meetings exhibits decreasing returns to scale. To corroborate that the choice of a linear model does not drive our findings, we also report the coefficient estimates of fixed-effect Poisson models using board meetings as the dependent variable.
Our measure of accruals quality (AQ) is based on the cross-sectional Dechow and Dichev (2002) model as modified by McNichols (2002):
where ΔWCAj,t is the change in working capital accruals4; CFO is cash flows from operations (OANCF) from the statement of cash flows; ΔREV is the change in revenue (SALE); and PPE is property, plant and equipment (PPEGT). All variables are scaled by average assets (AT). We estimate Eq. (2) for each two-digit SIC industry with at least ten firms in year t. AQjt is the annual decile rank of the standard deviation of the regression residuals from t−4 to t of firm j. We take the negative value of this variable so that higher values of AQ indicate higher quality accruals. Following prior research (e.g., Francis et al. 2005), we use deciles to mitigate measurement error in the estimates and reduce concerns about nonlinearity.
To isolate the effect of earnings quality on board meetings frequency, we control for several time-varying firm characteristics shown by prior research to affect either board meetings or earnings quality. Specifically, our main specifications control for measures of business uncertainty and information asymmetry that have been shown to affect board meeting frequency (e.g., Demsetz and Lehn 1985; Adams et al. 2021) and earnings quality (e.g., Lafond and Watts 2008). In addition, the board of directors composition and ownership structure can also affect the demand for board meetings (e.g., Mayers et al. 1997; Vafeas 1999) and earnings quality (e.g., Ahmed and Duellman 2007; Garcia Lara et al. 2009; Basu and Liang 2019). Therefore, we control for board size (measured by the number of directors), CEO/chair duality, the percentage of outside directors on the board, and insider ownership.
Furthermore, we control for firm performance, measured by stock returns and an indicator variable for negative earnings realizations (e.g., Vafeas 1999; Li 2008); growth opportunities, measured by research and development to total assets (e.g., Adams et al. 2021); the number of analysts following the firm (e.g., Adams et al. 2021); leverage (e.g., Adams et al. 2021); firm age and firm size, measured by total assets and the number of firm employees (e.g., Vafeas 1999; Francis et al. 2005); diversification, measured by the number of business segments (Adams et al. 2021); acquisition activity (Adams et al. 2021); and SOX regulation. Finally, all regression specifications include firm-fixed effects to control for time-invariant unobserved differences among firms and year-fixed effects to absorb common shocks to the macroeconomic environment.,5,
6
5 Empirical results
5.1 Sample selection and descriptive statistics
Our initial sample comprises all firms that are listed on the Russell 3000 list at least once between 1995 and 2017. This choice ensures that we have a long time series of data for each firm and allows robust within-firm tests. Since commercial databases on corporate boards do not provide coverage at the start of our sample period and only limited firm coverage for later periods, we manually gather all governance data on boards of directors and stock ownership from annual proxy statements. This data collection strategy ensures increased consistency in the data used in the tests. Consistent with prior work, we exclude financial and utility firms due to the potentially confounding role of their regulatory environment. Financial statement data are collected from Compustat, stock returns from CRSP, and management earnings forecasts from IBES. Firm-year observations without Compustat and CRSP data are eliminated from the sample. This screening process results in a maximum number of 2721 firms and 37,850 firm-year observations between 1996 and 2017 (as 1995 is a reference year used to construct lagged values in some variables).
Table 1 presents descriptive data on all the variables used in the tests. To mitigate the influence of outliers, we winsorized all continuous variables at the 1st and 99th percentile. The average board in our sample meets 7.55 times per year with an interquartile range of board meetings from five to nine and a standard deviation of 3.42, suggesting a notable variation in board meeting frequency across firm years. The mean (median) value of accruals quality across all firm years is − 0.054 (− 0.041), with an interquartile range from − 0.026 to − 0.065. In general, the descriptive statistics on board meetings, earnings quality, and control variables reported in Table 1 are similar to the corresponding variable statistics on similar samples reported by prior work (e.g., Francis et al. 2005; Biddle et al. 2009; Adams et al. 2021).
Table 1
Descriptive statistics
N
Mean
SD
p25
p50
p75
Board characteristics
Board Meetings
37,850
7.554
3.423
5.000
7.000
9.000
CEO Chair
37,850
0.517
0.500
0.000
1.000
1.000
Board Size
37,850
8.302
2.203
7.000
8.000
10.000
% NE Directors
37,850
0.799
0.112
0.750
0.833
0.875
% Insider Ownership
37,850
0.153
0.176
0.030
0.080
0.210
Earnings precision
AQ
33,491
− 0.054
0.044
− 0.065
− 0.041
− 0.026
ABS(AAadj)
37,201
− 0.112
0.156
− 0.132
− 0.066
− 0.029
EPSFREQ
37,850
1.326
2.344
0.000
0.000
2.000
DQ
37,494
0.787
0.075
0.751
0.791
0.836
FOG
20,686
− 19.628
1.753
− 20.393
− 19.457
− 18.631
Information asymmetry
Std. Dev. Returns
37,850
0.032
0.017
0.020
0.028
0.040
Idiosyncatic Risk
37,850
0.029
0.017
0.017
0.025
0.036
Bid-Ask Spread
37,850
0.042
0.022
0.026
0.036
0.052
Info Asymmetry
37,850
− 0.000
0.985
− 0.699
− 0.251
0.433
Firm characteristics
Stock Return
37,850
0.178
0.639
− 0.195
0.081
0.385
R&D
37,850
0.059
0.119
0.000
0.007
0.068
# of Analysts
37,850
8.785
7.956
3.000
7.000
13.000
Book Leverage
37,850
0.213
0.216
0.013
0.173
0.328
Acquisitions
37,850
0.027
0.064
0.000
0.000
0.019
Firm Age
37,850
21.875
15.621
10.000
17.000
31.000
# of Segments
37,850
1.864
1.264
1.000
1.000
2.000
Assets ($milion)
37,850
3,145.923
7,713.676
202.912
625.114
2,162.033
# Employees (thousand)
37,850
13.987
58.151
0.626
2.570
9.000
The sample comprises 2721 unique Russel 3000 firms and a maximum of 37,850 observations covering the period from 1996 to 2017. All continuous variables are winsorized at the 1st and 99th percentile. Board Meetings is the number of total board meetings held by the board of directors in a fiscal year as reported in the Definite Proxy Statement (SEC FORM 14F). CEO Chair is an indicator variable equal to 1 if the firm's CEO is also the Chair of the board. Board Size is the total number of directors on the board. % NE Directors is the fraction of total non-executive directors on the board. %Insider Ownership is the fraction of common stock owned by insiders. AQ is the standard deviation of the firm-level residuals from the Dechow and Dichev model during the years t-4 to t and multiplied by negative one. ABS(AAadj) is the absolute value of the difference between the firm's discretionary accruals and its closest firm in terms of return on assets within the same 2-digit industry multiplied by negative one. EPSFREQ is the number of annual management earnings forecasts during the fiscal years. DQ is the disaggregation of quality from Chen et al. (2015). FOG is the measure of financial statement readability computed by Li (2008) and multiplied by a negative one. Std.Dev. Returns is the standard deviation of daily returns during the fiscal year. Idiosyncratic Risk is the standard deviation of the residual returns from a market model regression of daily returns on the value-weighted market returns during the fiscal year. Bid-Ask Spread is the bid-ask spread defined as the annual average of daily spread scaled by the midpoint between bid and ask. Info Asymmetry is a continuous variable that measures market-based information asymmetry. It is computed as the average of the standardized values of std Returns, Bid-Ask Spread, Idiosyncratic Risk. Stock Return is the cumulative monthly stock return during the fiscal year. R&D is the research and development expense scaled by total assets. #Analysts is the number of analysts that issued annual earnings forecasts during the fiscal year. Book Leverage is the book value of short-term and long-term debt scaled by book value of assets. Acquisitions is acquisitions (AQC) scaled by the book value of assets. Firm Age is the number of years between the firm's current year and its first fiscal year-end date available in Compustat. #Segments is the number of business segments reported in Compustat segments database. Assets is the book value of assets. #Employees is the total number of employees
Table 2 presents initial insights into the determinants of accruals quality. Prior research suggests that accruals quality is jointly determined by innate factors, such as business models and operating environments, and by discretionary factors related to the financial reporting process per se, including managerial intervention and errors. Dechow and Dichev (2002) and Francis et al. (2005) identify five innate factors as affecting accruals quality: firm size (log assets), the standard deviation of cash flow from operations (std. dev. CFO), the standard deviation of sales (std. dev. sales); the length of the operating cycle (log operating cycle) and the incidence of negative earnings realizations (negative earnings). Descriptive statistics on the innate factors of accruals quality are presented in Panel A of Table 2. The sample mean values are 0.072 for Std CFO, 0.168 for Std Sale, 132 days for Operating Cycle, 27.4 percent for negative earnings, and 6.51 for log Assets. In general, the values of the innate factors for our sample are very similar to those reported by prior research (e.g., Dechow and Dichev 2002; Francis et al. 2005).
Table 2
Determinants of accruals quality
Panel A: Descriptive statistics
N
Mean
SD
p25
p50
p75
Innate factors of Accruals Quality
Std. Dev. CFO
36,162
0.072
0.099
0.026
0.044
0.076
Std. Dev. Sales
36,253
0.168
0.159
0.068
0.119
0.210
Operating Cycle
35,009
132.315
97.389
72.030
113.323
164.732
Negative Earnings
37,400
0.274
0.348
0.000
0.200
0.400
Assets (in $million)
37,850
3,145
7,714
203
625
2,162
Panel B: Accruals quality
Std CFO
Std Sale
Log (Op. Cycle)
Neg. Earnings
log (Assets)
Constant
Adj. R2
N
Predicted sign
–
–
–
–
+
Coefficients
− 7.918***
− 3.631***
− 0.442***
− 0.328**
0.283***
6.926***
53.30%
32,404
Standard errors
(0.687)
(0.201)
(0.098)
(0.132)
(0.054)
(0.536)
The sample comprises 2,721 unique Russel 3000 firms and a maximum of 37,850 observations covering the period from 1996 to 2017. Panel A presents descriptive statistics on the determinants of accruals quality. The model in Panels B is estimated using OLS methodology and include firm and year fixed effects. The dependent variable is the annual decile rankings of accruals quality, which is the standard deviation of the firm-level residuals during the years t-4 to t from the Dechow and Dichev (2002) model as modified by McNichols (2002), and multiplied by negative one. Std.Dev.CFO is the standard deviation of cash flow from operations scaled by average total assets from years t−4 to t. Std. Dev. Sales is the standard deviation of sales scaled by average total assets from years t−4 to t. Operating Cycle is the sum of receivables to sales plus inventory to cost of goods sold multiplied by 360. Negative Earnings is the proportion of losses over the past 5 years. Assets is the book value of assets. Standard errors are clustered at the firm level
*, **, ***, denote significance at p < 0.10; p < 0.05, and p < 0.01 respectively
In Panel B of Table 2, we regress the annual decile ranking of accruals quality on the five innate factors. Consistent with prior research, accruals quality is positively associated with firm size and inversely related to the standard deviation of cash flows, the standard deviation of sales, the length of the operating cycle, and the incidence of negative earnings. At 53%, the aggregate explanatory power of the five innate factors is similar to that reported by prior research (e.g., Francis et al. 2005). In sum, Table 2 reports initial evidence validating the accruals quality measure we use in later tests.
5.2 Multivariate analysis
Table 3 presents our tests on the relation between accruals quality and board meeting frequency. Models 1 and 2 in Table 3 estimate this relation using OLS and Poisson regression methodology, respectively. Consistent with our hypothesis, in both models, the accruals quality measure is negatively related to board meeting frequency (p < 0.01), signifying that boards require more information-seeking effort to carry out their work when accounting information is of lower quality.
Table 3
The relationship between accruals quality and board meeting frequency
1
2
3
4
AQ
− 0.004***
− 0.004***
− 0.004***
(0.001)
(0.001)
(0.001)
InnateAQ
− 0.016***
(0.005)
DiscAQ
− 0.003**
(0.001)
Post-SOX
0.075***
(0.008)
Info Asymmetry
0.066***
0.078***
0.061***
0.049***
(0.005)
(0.005)
(0.005)
(0.004)
CEO CHAIR
− 0.038***
− 0.034***
− 0.036***
− 0.038***
(0.007)
(0.008)
(0.007)
(0.007)
Log Board Size
0.013
− 0.005
0.012
0.020
(0.020)
(0.022)
(0.021)
(0.020)
% NE directors
0.137***
0.153***
0.139***
0.166***
(0.038)
(0.041)
(0.039)
(0.038)
% Insider Ownership
− 0.102***
− 0.108**
− 0.096**
− 0.121***
(0.038)
(0.042)
(0.039)
(0.038)
Loss
0.066***
0.070***
0.067***
0.065***
(0.007)
(0.007)
(0.007)
(0.007)
Stock Return
− 0.019***
− 0.022***
− 0.020***
− 0.017***
(0.003)
(0.004)
(0.003)
(0.003)
R&D
0.096*
0.050
0.111*
0.089*
(0.050)
(0.054)
(0.057)
(0.050)
# Analysts
− 0.001
− 0.001
− 0.001
− 0.000
(0.001)
(0.001)
(0.001)
(0.001)
Book Leverage
0.035
0.040*
0.026
0.022
(0.023)
(0.023)
(0.023)
(0.023)
Acquisitions
0.419***
0.383***
0.416***
0.438***
(0.033)
(0.036)
(0.034)
(0.033)
Log Firm Age
0.132***
0.134***
0.136***
0.102***
(0.024)
(0.026)
(0.025)
(0.015)
Log # of Segments
0.005
0.009
0.005
0.007
(0.012)
(0.013)
(0.012)
(0.012)
Log Assets
0.044***
0.044***
0.050***
0.037***
(0.009)
(0.009)
(0.010)
(0.009)
Log Employees
− 0.017**
− 0.021**
− 0.016*
− 0.017**
(0.008)
(0.008)
(0.008)
(0.008)
Constant
1.135***
1.155***
1.190***
(0.085)
(0.088)
(0.067)
F-stat/Chi-square
33.23
1,205
31.55
60.36
Adj. R-square
47.00%
47.30%
46.30%
Clusters
2,667
2,667
2,615
2,667
Observations
33,479
33,479
32,404
33,479
Models 1, 3, and 4 are estimated using OLS methodology and model 2 is estimated using Poisson regression methodology. All models include firm and year fixed effects. The dependent variable is the natural logarithm of the number of board meetings in OLS models and the number of board meetings in the Poisson model. Accruals quality (AQ) is the annual decile ranking of the standard deviation of the firm-level residuals of years t−4 to t from the Dechow and Dichev (2002) model as modified by McNichols (2002), multiplied by negative one. Innate AQ is the predicted value obtained from the regression of AQ on the innate factors of accruals quality presented in Panel B of Table 2. Discretionary AQ is the residual from the above regression. Post-SOX is set to one for fiscal years 2002–2017, and zero otherwise. All other variables are defined in Appendix 1. Standard errors are clustered at the firm level in all OLS models and are heteroskedasticity robust in the Poisson model
*, **, ***, denote significance at p < 0.10; p < 0.05, and p < 0.01, respectively
A running concern is that directors meet more frequently not just to gather more information but also to actively influence the quality of accounting information. This influence is primarily exercised through the well-documented oversight of the reporting process (e.g., Xie et al. 2003). To address this concern and to distinguish between the two possible explanations for the documented association (i.e., information gathering versus monitoring), we decompose accruals quality into innate and discretionary components.
Innate accruals quality (innate AQ) represents the component of accruals quality that is inherent to the financial reporting process of the firm. It includes factors that are relatively stable and less influenced by managerial control, and we view them as predetermined in any given financial reporting date. In contrast, discretionary accruals quality (discretionary AQ) is more malleable and can be shaped by managerial decisions and actions. It encompasses various factors, including unintentional errors, performance measurement, and opportunistic behavior (e.g., Guay et al. 1996; Francis et al. 2005). The rationale behind this test is to understand whether boards are meeting more frequently to gather more information due to lower accruals quality or if they are actively working to influence and improve the quality of financial data. The hypothesis is that if the frequency of board meetings is primarily driven by the need for more information due to lower accruals quality, then the effect of the innate accruals component should be more significant. Conversely, if boards are meeting more often to influence and shape accounting quality (i.e., discretionary AQ), then the effect of the discretionary component on board meetings should be more substantial.
Following Francis et al. (2005) we estimate innate accruals quality (innate AQ) using the predicted values from the regression of accruals quality on the factors presented in Panel A of Table 2, and the discretionary component of a firm's accruals quality in year t (discretionary AQ) using the residuals from the above regression.
Model 3 of Table 3 presents the results of this analysis. Both coefficients are negative and statistically significant. The negative coefficient on the discretionary component of accruals quality suggests that, on average, discretionary accruals provide useful information to the board of directors. This result is consistent with evidence from prior research that the net effect of discretion over accruals quality in a broad sample of firms improves earnings quality (e.g., Subramanyam 1996; Francis et al. 2005). Importantly and consistent with our expectations, we find that the effect of the discretionary component of accruals quality, which includes the opportunism and error subcomponents, is both smaller in magnitude and weaker in statistical significance than the effect of the innate component of accruals quality. The estimated coefficient on the innate component of accruals quality is over five times as large as the estimated coefficient on the discretionary component (coefficient sizes of 0.016 and 0.003, respectively). The relative import of innate compared to discretionary accruals in explaining board activity further supports our hypothesis.
Model 4 in Table 3 excludes year-fixed effects and controls for the effect of SOX on board activity, given that SOX was found to increase the responsibility and work burden on boards of directors (e.g., Adams et al. 2021). Boards meet more frequently after SOX and, as predicted by our hypothesis, the effect of accounting information quality on board meeting frequency persists after controlling for the effects of SOX.
Evidence on the other determinants of board meetings from Table 3 corroborates the notion that boards compensate for the lack of relevant information by increasing director contact time through more frequent board meetings. Specifically, boards meet more frequently when firms operate in an environment of greater information asymmetry; when a director other than the CEO chairs the board, who presumably requires more effort to collect and process the information necessary to carry out the tasks; when there is a greater fraction of outsiders on the board, who have an information disadvantage compared to inside directors; and when ownership is more dispersed, given that insider ownership is a substitute incentive alignment mechanism. Each of these variables is significant at p < 0.01 in all four model specifications, with models 1–3 including firm and year fixed effects. Importantly, pre-meeting performance is a key control variable in our models, given that board meetings are responsive to poor performance (Vafeas 1999; Brick and Chidambaran 2010; and Adams et al. 2021) and that corporate performance has been linked to earnings quality (e.g., Li 2008). In line with prior research, our empirical results confirm that board meeting frequency is preceded by abnormally poor firm performance. Other control variables are also found to be related to board meeting frequency, consistent with earlier work.
To probe further into the relation between accruals quality and board meetings, we re-estimate model 1 of Table 3 using indicator variables for each earnings quality decile. This analysis addresses an important nonlinearity concern: If the relationship between board meetings and earnings quality stems from boards responding to accounting problems by meeting more frequently and not because they seek more information to carry out their tasks, we would expect this relationship to be more pronounced in the lowest accounting quality deciles; i.e., when accounting problems are the greatest. As depicted by the decile coefficient sizes in Fig. 1, we find no such evidence. Specifically, even though the coefficients on accruals quality are negative in every decile, their magnitude and significance are relatively greater in deciles six to ten, when accounting information quality is generally higher. The coefficient estimate of − 0.04 on the highest AQ decile suggests a modest difference of 4% or 0.3 meetings per year between the worst and best AQ deciles. The more pronounced negative relation between accruals quality and board meetings in the higher accruals quality deciles contrasts the alternative explanation that our results are solely driven by boards meeting more frequently to address problems in accounting quality.
Fig. 1
Coefficient plot of accruals quality deciles on board meetings. The figure presents the estimated coefficients of accruals quality (AQ) deciles on board meeting frequency. The coefficents are estimated using OLS methodology with firm and year fixed effects. The dependent variable is the natural logarithm of the number of board meetings and the explanatory variables of interest are indicator variables that take the value of one for each accruals quality decile and zero otherwise. The regression includes all controls from Table 3. Standard errors are clustered at the firm level. The coefficients are depicted with 95% confidence intervals
×
5.3 Moderators of the relationship between earnings quality and board meetings
The evidence thus far suggests a negative effect, on average, of accruals quality on board activity. This section explores the possibility that these effects are not uniform across firms. We hypothesize and test second-order effects drawing on the general notion that firm characteristics that vary cross-sectionally and proxy for access to information help explain the usefulness of accounting information to corporate boards.
First, we examine how the relationship between earnings quality and board meetings varies with the proportion of outside directors. Prior research suggests that outside directors have an information disadvantage compared to inside directors and that boards meet more frequently in firms with a higher proportion of outside directors (Vafeas 1999). Because outside directors are more likely to rely on accounting information for their information needs, we expect the negative effect of accruals quality on board meetings to be more pronounced when relatively more outside directors serve on the board.
Second, we examine whether insider ownership mitigates the effect of earnings quality on board meetings. Inside ownership is an alternative mechanism through which firms alleviate information gathering and processing costs and managerial misalignment costs (e.g., La Porta et al. 1998). Further, prior research suggests that high levels of inside ownership substitute for costly board monitoring in disciplining management (e.g., Mayers et al. 1997). Thus we expect the negative effect of accruals quality on board meetings to be less pronounced in firms where insiders own a greater percentage of common stock.
Third, we examine the moderating role of the firm's general information environment. Information asymmetry and business uncertainty accentuate information gathering and processing costs, exacerbate managerial misalignment costs, and impede the board's advising and monitoring effectiveness (e.g., Coles et al. 2008). Consistent with the above arguments, prior research shows that boards meet more frequently in firms operating in a more volatile environment (e.g., Adams et al. 2021). We expect earnings quality to help alleviate the adverse effects of information asymmetry.
Finally, we expect that SOX has accentuated the value of accounting information. SOX strove to improve the quality and reliability of financial reporting, an aim that some prior studies have substantiated (e.g., Bartov and Cohen 2009; Defond and Lenox 2011; Kalelkar and Nwaeze 2011). Thus, even though greater board activity is expected after SOX (e.g., Adams et al. 2021), the value placed on financial reporting properties (i.e., accruals quality) is also expected to have increased after SOX.
Table 4 presents the moderating effects of outside directors, insider ownership, and information asymmetry on accruals quality. These effects are estimated by building on the baseline model (model 1 of Table 3), adding interactive terms between accruals quality and moderating factors separately and together (models 1–4). Last, model 5 tests whether SOX enhances these relationships. To ease the interpretation of the results, we center all interacted continuous variables around their mean.
Table 4
Moderators of the relationship between accruals quality with board meeting frequency
(1)
(2)
(3)
(4)
(5)
AQ
− 0.004***
− 0.004***
− 0.004***
− 0.004***
0.003
(0.001)
(0.001)
(0.001)
(0.001)
(0.002)
AQ × % NE directors
− 0.022**
− 0.019**
(0.009)
(0.009)
AQ × % Insider Ownership
0.015**
0.013*
(0.007)
(0.007)
AQ × Info Asymmetry
− 0.001
− 0.002
(0.001)
(0.001)
AQ × SOX
− 0.009***
(0.002)
SOX
0.125***
(0.016)
Info Asymmetry
0.067***
0.067***
0.072***
0.074***
0.050***
(0.005)
(0.005)
(0.007)
(0.007)
(0.004)
% NE directors
0.255***
0.136***
0.137***
0.239***
0.166***
(0.065)
(0.038)
(0.038)
(0.067)
(0.038)
% Insider Ownership
− 0.101***
− 0.177***
− 0.102***
− 0.166***
− 0.119***
(0.038)
(0.054)
(0.038)
(0.055)
(0.038)
Other controls
Included
Included
Included
Included
Included
F-stat
32.48
32.51
32.55
31.23
57.51
Adj. R-square
47.00%
47.00%
47.00%
47.00%
46.30%
Clusters
2,667
2,667
2,667
2,667
2,667
Observations
33,479
33,479
33,479
33,479
33,479
Models 1–4 are estimated using OLS methodology and include firm and year fixed effects. Model 5 includes firm fixed effects but no year fixed effects. The dependent variable is the natural logarithm of the number of board meetings in each firm-year. Accruals quality (AQ) is the annual decile ranking of the standard deviation of the firm-level residuals for years t−4 to t from the Dechow and Dichev (2002) model as modified by McNichols (2002), and multiplied by negative one. The percentage of NE directors is the proportion of non-executive (outside) directors on the board. The percentage of insider ownership is the percentage of common stock beneficially owned by officers and directors as a group. Information asymmetry is a coninuous variable that measures market-based information asymmetry, and computed as the average of the standardized values of the standard deviation of daily stock returns in the year, the bid-ask spread, and idiosyncratic risk. SOX is set to one for fiscal years 2002–2017 and zero otherwise. For ease of interpretation, all continuous interacted variables are centered around their respective means. The models include all control variables reported in Table 3 and defined in Appendix 1, for which results are not tabulated for brevity. Standard errors are clustered at the firm level in all models
*, **, ***, denote significance at p < 0.10, p < 0.05, and p < 0.01, respectively
The results from Table 4 corroborate previous evidence that outside directors require more information in general (Duchin et al. 2010) by illuminating the usefulness of accruals quality in particular. As expected, the results on the moderating role of insider ownership suggest that the effect of accruals quality is more pronounced in firms with a more dispersed ownership structure. These results hold separately in models 1–2 and together in model 4. Furthermore, in models 3 and 4, the coefficient on the interaction between earnings quality and information asymmetry is not statistically significant, suggesting that the effect of earnings quality on board meetings is not affected by the firms' general information environment. Finally, in model 5, the results confirm the increased significance of accruals quality for directors after SOX.7 In sum, the evidence from Table 4 suggests that proxies for information availability moderate the board's demand for soft information.
6 Further tests
In this section, we discuss the results from several additional tests that probe further into the relationship between financial reporting properties and board meeting frequency.
6.1 Alternative measures of financial reporting quality and board activity
First, we examine the robustness of our results to alternative measures of financial reporting quality. We recognize that accruals quality has some limitations as a proxy for financial reporting quality, including being limited to current accruals, imposing a survivorship bias on the sample firms, and not accounting for other financial accounting information that can complement the board's information set (e.g., Francis et al. 2005; Dechow et al. 2010). Thus, we replicate our primary analysis using four additional measures of financial reporting quality: (1) the absolute value of performance-matched abnormal accruals multiplied by negative one, where abnormal accruals are estimated using the Jones (1991) model (ABS(AAadj); Kothari et al. 2005); (2) the number of management earnings forecasts in a year (EPSFREQ; e.g., Hui et al. 2009); (3) a financial reporting quality measure based on the level of disaggregation of financial data items in annual reports (DQ; Chen et al. 2015); and, (4) the FOG index of financial statement readability multiplied by negative one, which captures the overall readability of 10-K filings (FOG; Li 2008).8
The results of this analysis are reported in Table 5. Notwithstanding some variation in economic import, the alternative accounting quality measures uniformly suggest that boards meet more frequently when accounting quality is lower. This evidence is in line with the notion that directors rely on accounting information for decision-making and that lower accounting quality forces boards of directors to exert more effort to carry out the required tasks.
Table 5
Alternative measures of financial reporting quality and board meeting frequency
1
2
3
4
ABS(AAadj)
EPSFREQ
DQ
FOG
FRQ
− 0.059***
− 0.006***
− 0.272***
− 0.003**
(0.013)
(0.001)
(0.057)
(0.002)
Info Asymmetry
0.064***
0.067***
0.066***
0.063***
(0.005)
(0.005)
(0.005)
(0.006)
CEO CHAIR
− 0.036***
− 0.038***
− 0.038***
− 0.041***
(0.007)
(0.007)
(0.007)
(0.009)
Log Board Size
− 0.002
− 0.001
0.002
− 0.008
(0.020)
(0.020)
(0.020)
(0.024)
% NE directors
0.143***
0.142***
0.136***
0.162***
(0.036)
(0.036)
(0.036)
(0.041)
% Insider Ownership
− 0.095***
− 0.098***
− 0.095***
− 0.075*
(0.032)
(0.032)
(0.032)
(0.041)
Other controls
Included
Included
Included
Included
F-stat
38.42
38.71
38.47
33.42
Adj. R-square
46.00%
45.70%
45.80%
49.60%
Clusters
2,695
2,721
2,719
2,132
Observations
37,190
37,850
37,493
20,650
All models are estimated using OLS methodology and include firm and year fixed effects. The dependent variable is the natural logarithm of the number of board meetings in each firm-year. FRQ captures the alternative earnings quality measures as follows: ABS(AAadj) is the absolute value of the difference between the firm's discretionary accruals and its closest firm in terms of return on assets within the same two-digit industry multiplied by negative one. Discretionary accruals are estimated using the Jones (1991) model. EPSFREQ is the number of management earnings forecasts in a year. DQ is the Chen et al. (2015) disaggregation quality measure based on the level of disaggregation of financial data items in annual reports. FOG is a measure of financial statement readability computed by Li (2008), multiplied by negative one. The models include all control variables reported in Table 3 and defined in Appendix 1, for which results are not tabulated for brevity. Standard errors are clustered at the firm level
*, **, ***, denote significance at p < 0.10; p < 0.05, and p < 0.01, respectively
6.2 Management earnings forecasts and board meeting frequency
Next, we examine whether management earnings forecasts serve as a complementary signal for corporate directors. We single out firm years with at least one management forecast from the IBES database and separate management forecasts into two categories: Earnings forecasts that coincide with (match) the consensus analyst forecast at the time the management forecast is released and earnings forecasts that differ from the consensus analyst forecast (e.g., exceed or fall short the analyst forecast). The intuition of this analysis is that management earnings forecasts proxy for managers' private information about future earnings, while analyst consensus forecasts proxy for expected future earnings given the publicly available information set. In line with our hypothesis, we expect that complete alignment (matching) of management and analyst forecasts signals lower information asymmetry and a lower need for boards to seek more information through increased activity.
Table 6 displays regression results, investigating the link between the number of management forecasts that match or differ from the consensus analyst forecast in a given year and the corresponding number of board meetings during that year, separately in models 1 and 2 and together in model 3. The explanatory variables of interest in these regressions are defined as the total number of management forecasts coinciding with the consensus analyst forecast (#match), and the total number of management forecasts that deviate from the consensus during the year (#nomatch). In Model 4, we replicate the tests from Model 3 using Poisson regression methodology. Notably, our findings are consistent across all models. As hypothesized, the results reveal that boards tend to meet more frequently in years of more disagreement between management and analyst forecasts. This result suggests that boards actively seek additional information in less transparent information environments.9 The results in Table 6 corroborate earlier evidence about greater information seeking by boards when accounting information is of lower quality.10
Table 6
The relationship between manager and analyst forecast alignment with board meeting frequency
1
2
3
4
# Match
− 0.010***
− 0.009***
− 0.010***
(0.002)
(0.002)
(0.002)
# Nomatch
0.007***
0.004*
0.004*
(0.002)
(0.002)
(0.002)
Info Asymmetry
0.070***
0.071***
0.070***
0.081***
(0.009)
(0.010)
(0.009)
(0.010)
CEO CHAIR
− 0.024**
− 0.024**
− 0.023**
− 0.022*
(0.011)
(0.011)
(0.011)
(0.012)
Log Board Size
− 0.027
− 0.026
− 0.027
− 0.045
(0.032)
(0.032)
(0.032)
(0.035)
% NE directors
0.149***
0.149***
0.150***
0.160**
(0.057)
(0.057)
(0.057)
(0.067)
% Insider Ownership
0.007
0.011
0.009
0.043
(0.058)
(0.058)
(0.058)
(0.064)
Other Controls
Included
Included
Included
Included
F-stat/Chi-square
19.73
18.35
18.78
502.20
Adj. R-square
45.80%
45.70%
45.80%
Clusters
1,524
1,524
1,524
1,524
Observations
12,196
12,196
12,196
12,196
Models 1–3 are estimated using OLS methodology and model 3 is estimated using Poisson regression methodology. All models include firm and year fixed effects. The dependent variable is the natural logarithm of the number of board meetings in the OLS models and the number of board meetings in the Poisson model. # Match, # Nomatch is the number of management earnings forecasts each year that coincide with (match), or differ from the consensus analyst forecast at the time the management forecast is released, respectively. The models include all control variables reported in Table 3 and defined in Appendix 1, for which results are not tabulated for brevity. Standard errors are clustered at the firm level in all OLS models and are heteroskedasticity robust in the Poisson model
*, **, ***, denote significance at p < 0.10; p < 0.05, and p < 0.01, respectively
6.3 Financial restatements, internal controls weaknesses, and board meeting frequency
We extend our analysis to accounting restatements and disclosed material weaknesses in internal controls as additional robustness tests of how accounting quality relates to board activity. Our study of restatements and internal control weaknesses is motivated by the elusiveness of capturing accounting quality in empirical accounting studies, and is intended to provide some comfort against the inherent measurement bias, providing an attempt at triangulation of our results. An advantage of restatements and material weaknesses over other proxies of reporting quality is that an external source has identified a problem with financial statement quality, thus avoiding reliance on a model to measure reporting quality (Dechow et al. 2010). Both events are generally considered to be associated with lower accounting quality. Accordingly, and in line with our hypothesis, we expect board meetings to be higher in the post-restatement period and in years of reported material weakness.
To identify a sample of restatements, we use the Audit Analytics database between September 1st, 2004, and December 31st, 2017. We broadly follow Ettredge et al. (2012) in our sample screening process and focus on firms that (1) only restated earnings once during the sample period to avoid the clouding effects of intertemporal comparisons when a single firm announces multiple restatements; (2) announced adverse restatements (i.e., overstatements); (3) have at least one observation in both the restatement and post-restatement period. In addition, following Ettredge et al. (2012), we delete all firm years in the pre-overstatement period. Finally, to account for the increasing proportion of immaterial or economically insignificant restatements (Hennes et al., 2008), we use an indicator variable that takes the value of one for non-reliance restatements signified by an 8-K item 402 report and zero otherwise. We identify 528 firms that make one-time adverse financial restatements (4220 firm-year observations).
Models 1 and 2 of Table 7 report the results of this relation; model 1 uses both the sample of restating and non-restating firms, and model 2 focuses on the sample of restating firms. The key variables of interest are the post-restatement variable which takes a value of one in the post-restatement years and zero otherwise, and the interaction between the post-restatement period and the non-reliance indicator variable. In both models, the post-restatement variable is unrelated to board meeting frequency. However, in line with expectations and the notion that material restatements undermine the board's confidence in accounting information, the sub-sample of material restatements is followed by an increase in board meeting frequency. This finding is consistent with greater information-seeking by boards in the years following material financial restatements to compensate for the lower accounting quality signaled by restatements.11
Table 7
Board meeting frequency around financial restatements and material internal control weaknesses
Financial restatements
Int. control weaknesses
E(Sign)
1
2
3
4
Post-overstatement
+
− 0.028
− 0.035
(0.021)
(0.025)
Post-overstatement × Non Reliance
+
0.134***
0.128***
(0.035)
(0.032)
ICMW
+
0.057***
0.059***
(0.013)
(0.019)
ICMW_fixed
–
− 0.044***
− 0.038*
(0.013)
(0.021)
Info Asymmetry
+
0.082***
0.079***
0.091***
0.105***
(0.006)
(0.014)
(0.006)
(0.014)
CEO Chair
–
− 0.045***
− 0.043*
− 0.015
− 0.036*
(0.009)
(0.023)
(0.009)
(0.021)
Log Board Size
–
− 0.023
− 0.036
− 0.053**
− 0.071
(0.026)
(0.066)
(0.025)
(0.073)
% NE directors
+
0.114**
0.168
0.203***
0.111
(0.046)
(0.134)
(0.055)
(0.107)
% Insider Ownership
–
− 0.052
− 0.247*
− 0.034
− 0.192**
(0.043)
(0.139)
(0.045)
(0.192)
Other controls
Included
Included
Included
Included
F-statistic
20.33
5.36
20.98
5.95
Adj. R-squared
46.70%
49.10%
49.0%
47.1%
Clusters
1,843
528
2,313
465
Observations
21,410
4,220
22,379
4,506
The models are estimated using OLS methodology and include firm and year fixed effects. Models 1 and 3 are estimated on the full sample with available data, and models 2 and 4 on firm-years with at least one restatement and material weakness in internal controls, respectively. The dependent variable is the natural logarithm of the number of board meetings in each firm-year. The post-restatement variable takes a value of one in the post-restatement years and zero otherwise. The non-reliance variable is set to one for firms that announced a material restatement by filing an 8-K report, and zero otherwise. ICMW equals one if the firm disclosed a 404 internal control weakness in the current or previous year and zero otherwise. ICMW_fixed is an interaction term between the ICMW indicator variable and a second indicator that equals one if the firm subsequently received an unqualified SOX 404 opinion, and zero otherwise. The model includes all control variables reported in Tables 3 and 4 and defined in Appendix 1, for which results are not tabulated for brevity. Standard errors are clustered at the firm level
*, **, ***, denote significance at p < 0.10; p < 0.05, and p < 0.01, respectively
Next, we examine the link between section 404 material weaknesses in internal controls and board meeting frequency. We use Audit Analytics to extract the sample firms disclosing at least one SOX 404 internal control material weakness from November 15th, 2004 to December 31st, 2017 (n = 465). We define our variables of interest following Ashbaugh-Skaife et al. (2008). Specifically, we assign a value of one to firms that received an adverse SOX 404 opinion in the current or previous year and zero otherwise (ICMW). Further, we assign a value of one to firms that remediated their internal control weaknesses if they subsequently received an unqualified SOX 404 opinion and zero otherwise (ICMW_fixed). Ashbaugh-Skaife et al. (2008) document lower accruals quality in firms that report an internal control weakness relative to control firms and improvements in accruals quality in firms that resolved their internal control problems relative to firms that continue to have weak internal controls. Accordingly, we expect board meetings to be higher in firms that report a material weakness but lower in firms that resolve their internal control problems.
Similar to restatements, we estimate two models linking internal control weaknesses to board meeting frequency; model 3 of Table 7 uses both the sample of firms with internal control weaknesses and control firms and model 4 focuses on the sample of firms with internal control weaknesses. As expected, internal control weaknesses are associated with higher board activity consistent with greater information-seeking by boards when accounting quality is low. Notably, this effect is significantly lower in firms that subsequently remedied their internal control problems.12 The evidence presented in models 3 and 4 is consistent with our hypothesis and suggests that boards meet more frequently in firms with low accruals quality. Admittedly, the former effect is also consistent with the alternative explanation that boards meet more frequently to address the internal weakness problem per se. However, while internal control problems are likely discussed at the board level, it is mainly the responsibility of the audit committees to oversee and resolve these problems. Thus, we posit that the increase in board activity following the revelation for internal control weaknesses reflects, at least in part, an effort by boards to compensate for the lack of reliable accounting information.
6.4 Endogeneity concerns and alternative explanations
Our results provide evidence of an association between earnings quality and board meetings that is robust to alternative measures of earnings quality and a large number of control variables. In the absence of a natural experiment, we rely on the variation that is orthogonal to the determinants of earnings quality and board meeting frequency. However, as is common in accounting research, we cannot rule out endogeneity as an alternative explanation for the results. Therefore, it is possible that omitted variables that are correlated with earnings quality and board activity may drive the results. One potential correlated omitted variable is the identity of the CEO. A growing body of work suggests that CEO characteristics such as ability, reputation, and overconfidence drive earnings quality, and thus the identity of the CEO may be an important omitted variable in explaining the results (e.g., Ahmed and Duellman 2013; Demerjan et al. 2013). To alleviate this concern, we use CEO names to create CEO-firm identifiers and re-estimate all our models using CEO-firm fixed effects. We thus control for time-invariant CEO-firm unobserved heterogeneity that may drive both earnings quality and board meetings. Our results continue to hold after controlling for CEO-firm fixed effects.
Further, our results hold when we introduce additional controls for institutional ownership, the issue of debt or equity, dividend payments, sales growth, restructuring, capital intensity, intangibles intensity, the length of the operating cycle, the length of the investment cycle, foreign operations, industry concentration, CEO turnovers, accounting performance, and industry times year fixed effects. In addition, to alleviate concerns that the first post-SOX years drive the results, given that the effect of SOX is likely to be most pronounced on governance mechanisms, we replicate our baseline model results after excluding fiscal years 2002 to 2005 from our main analysis.
A particular concern is that the observed relation between earnings quality and board meetings flows from board meetings to earnings quality, not vice versa. It is, therefore, plausible that more active boards, as a proxy for the board's monitoring effectiveness, correlate with the mechanisms that affect earnings properties. However, it seems counter-intuitive to argue that the negative association between board meeting frequency and discretionary accruals quality stems from more active boards negatively affecting the reporting process per se.
Nevertheless, to partially address the challenge of correct identification and the direction of causality, we test our hypotheses using a dynamic panel estimator (Arellano and Bond 1991; Blundell and Bond 1998), which allows the inclusion of the lagged dependent variable to control reverse causality while avoiding dynamic panel bias (Nickell 1981). The Arellano and Bond (1991) estimators use first differences to remove time-invariant and heterogeneity and remove unobserved heterogeneity by instrumenting endogenous regressions by their past realizations. For this analysis, we use the absolute value of performance-matched abnormal accruals (ABS(AAadj)) to measure accruals quality.13
Table 8 presents results for our main tests. Model 1 is estimated using the two-step difference generalized method of moments (GMM; Arellano and Bond 1991), and model 2 using the two-step system GMM (Blundel and Bond 1998). Both models use the Windmeijer (2005) correction of standard errors and allow earnings quality to be correlated with past realizations of the error term. We use one and two-period lags of the dependent variable and all the explanatory variables from Table 3. In both models, the Hansen J-statistic for over-identifying restrictions suggests that the instruments are uncorrelated with the error term, and the Arellano and Bond (1991) serial correlation test rejects second-order serial correlation in the first-difference residuals. The reported coefficients confirm our prior results.
Table 8
GMM Regressions – Results from main tests
1
2
ln Board Meetings
0.512***
0.624**
(0.155)
(0.270)
Ln Board Meetings
0.017
0.007
(0.036)
(0.066)
AQ
− 0.095***
− 0.104***
(0.022)
(0.035)
Info Asymmetry
0.026**
0.032
(0.012)
(0.028)
CEO CHAIR
0.013
− 0.006
(0.011)
(0.020)
Log Board Size
− 0.146***
0.016
(0.051)
(0.058)
% NE directors
0.001
0.205***
(0.082)
(0.074)
% Insider Ownership
− 0.060
0.225
(0.151)
(0.194)
Other controls
Included
Included
AR (1) p-value
0.000
0.006
AR (2) p-value
0.282
0.399
Hansen J test of overidentifying restrictions (p-value)
0.225
0.552
Observations
31,755
34,855
The models are estimated using the generalized method of moments (GMM) and include year fixed effects. Model 1 is the Arellano and Bond (1991) two-step Difference GMM, and model 2 is the Blundel and Bond (1998) two-step system GMM. The dependent variable is the natural logarithm of the number of board meetings in each firm-year. AQ is the absolute value of the difference between the firm's discretionary accruals and its closest firm in terms of return on assets within the same two-digit industry multiplied by negative one. Discretionary accruals are estimated using the Jones (1991) model(ABS(AAadj)). The models include all control variables reported in Table 3 and defined in Appendix 1, for which results are not tabulated for brevity. Standard errors are reported in parentheses and are corrected using the Windmeijer (2005) correction process
*, **, ***, denote significance at p < 0.10; p < 0.05, and p < 0.01, respectively
Table 9 presents results using the two step-system GMM methodology in four panels, corresponding to our further tests that were presented in Tables 4, 5, 6 and 7 respectively: the significance of the interactive terms between accruals quality and non-executive directors, insider ownership, and information asymmetry, respectively (panel A), four alternative financial reporting quality measures (panel B), manager and analyst forecast alignment (panel C), and financial restatements and internal control weaknesses (panel D). The results reinforce the moderating effect of non-executive directors, robustness across earnings quality measures, matching analyst and manager forecasts, and the effect of non-reliance restatements and material weaknesses. The effects of non-matching forecasts and remedied internal control weaknesses (ICMW-fixed), that were previously marginally significant, retain their negative sign but lose that significance, whereas the previously insignificant information asymmetry interactive term becomes significantly negative, as would be expected, in the GMM analysis. Taken as a whole, the GMM results are largely consistent with earlier findings.
Table 9
GMM Regressions—results from further tests
Panel A: Moderators of the relationship between accruals quality with board meeting frequency
1
2
Ln Board Meetingst−1
0.398***
0.667***
(0.143)
(0.115)
Ln Board Meetingst−2
0.057
0.103
(0.037)
(0.087)
AQ
− 0.033*
− 0.790*
(0.020)
(0.427)
AQ × % NE directors
− 0.306*
(0.183)
AQ × % Insider Ownership
0.023
(0.115)
AQ × Info Asymmetry
− 0.078***
(0.023)
AQ × SOX
− 1.498***
(0.522)
Other controls
Included
Included
AR (1) p-value
0.000
0.000
AR (2) p-value
0.829
0.783
Hansen J test of overidentifying restrictions (p-value)
0.389
0.208
Observations
34,855
34,855
Panel B: Alternative measures of financial reporting quality
1
2
3
4
ABS(AAadj)
EPSFREQ
DQ
FOG
Ln Board Meetingst−1
0.624**
0.311
0.678***
0.485*
(0.270)
(0.256)
(0.234)
(0.255)
Ln Board Meetingst−2
0.007
0.077
− 0.018
0.021
(0.066)
(0.066)
(0.060)
(0.066)
FRQ
− 0.104***
− 0.004*
− 0.322***
− 0.009***
(0.035)
(0.003)
(0.108)
(0.003)
Other controls
Included
Included
Included
Included
AR (1) p-value
0.006
0.035
0.001
0.003
AR (2) p-value
0.399
0.869
0.199
0.574
Hansen J test of overidentifying restrictions (p-value)
0.552
0.320
0.675
0.287
Observations
34,855
35,427
35,089
19,223
Panel C: Manager and analyst forecast alignment
1
2
3
Ln Board Meetingst-1
0.165**
0.113*
0.117*
(0.066)
(0.065)
(0.062)
Ln Board Meetingst-2
0.086***
0.092***
0.094***
(0.024)
(0.024)
(0.024)
# Match
− 0.007***
− 0.008***
(0.003)
(0.003)
# Nomatch
0.001
− 0.002
(0.003)
(0.003)
Other Controls
Included
Included
Included
AR (1) p-value
0.000
0.000
0.000
AR (2) p-value
0.374
0.135
0.121
Hansen J test of overidntifying restrictions (p-value)
0.741
0.572
0.665
Observations
12,043
12,043
12,043
Panel D: Financial restatements and material internal control weaknesses
1
2
Ln Board Meetingst−1
0.014
0.277***
(0.140)
(0.088)
Ln Board Meetingst−2
0.054
0.050**
(0.039)
(0.025)
Post
− 0.613*
(0.352)
Post × Non Reliance
1.199***
(0.460)
ICMW
0.166*
(0.101)
ICMW_FIXED
− 0.096
(0.136)
Other controls
Included
Included
AR (1) p-value
0.004
0.000
AR (2) p-value
0.226
0.791
Hansen J test of overidntifying restrictions (p-value)
0.692
0.573
Observations
20,175
21,806
The models are estimated using the Blundel and Bond (1998) two-step system generalized method of moments (GMM) and year fixed effects. The dependent variable is the natural logarithm of the number of board meetings in each firm year. AQ is the absolute value of the difference between the firm's discretionary accruals and its closest firm in terms of return on assets within the same two-digit industry multiplied by negative one. Discretionary accruals are estimated using the Jones (1991) model (ABS(AAadj)). The models include all control variables reported in Table 3 and defined in Appendix 1, for which results are not tabulated for brevity. Standard errors are reported in parentheses and are corrected using the Windmeijer (2005) correction process.
*, **, ***, denote significance at p < 0.10; p < 0.05, and p < 0.01, respectively.
Our paper has largely assumed that low accounting quality signals a low quality of hard accounting information. An alternative possibility is that low accounting quality is a sign of other potential business problems/challenges facing the firm, leading the board to meet more often to address these challenges. To address this possibility, we identify and control for a number of important business situations that may lead to more frequent board meetings, examining the incremental value of accounting quality in explaining board activity beyond these situations. The results are presented in Table 10. First, model 1 examines the effect of negative prior stock returns, capital expenditures, the disposal of property plant and equipment, the issue of long term debt or new equity, a CEO change, and dividend and share repurchase payouts on board meeting frequency. As expected, each of these situations is associated with a higher number of board meetings. Importantly, accounting quality has negative and significant effect on the number of board meetings beyond these controls.
Table 10
Business decisions
1
2
AQ
− 0.004***
0.000
(0.001)
(0.002)
Negative returns t
0.030***
0.042***
(0.004)
(0.009)
Investmentst+1
0.159***
0.217***
(0.032)
(0.068)
Disposalst+1
0.503*
1.409**
(0.260)
(0.549)
External financet+1
0.025**
0.018
(0.013)
(0.025)
CEO changet
0.132***
0.161***
(0.007)
(0.015)
Payoutt+1
0.105**
0.223***
(0.043)
(0.080)
AQ × Negative returnst
− 0.002*
(0.001)
AQ × Investmentst+1
− 0.011
(0.010)
AQ × Disposalst+1
− 0.176**
(0.088)
AQ × External financet+1
0.001
(0.004)
AQ × CEO changet
− 0.006**
(0.002)
AQ × Payoutt+1
− 0.024**
(0.012)
Other controls
Included
Included
F-stat
48.53
38.07
Adj. R-square
48.50%
48.50%
Clusters
2,644
2,644
Observations
30,419
30,419
The dependent variable is the natural logarithm of the number of board meetings in each firm year. Investments in the sum of capital expenditures and acqusision expenditures scaled by total assets. Disposals is the total value of property, plan and equipment sold during a year scaled by assets. External finance is the sum of long-term debt issuance and the sale of common and preferred stock scaled by assets. CEO change takes the value of 1 if there is a CEO change in the year and zero otherwise. Payout is the sum of dividends and share repurchases scaled by assets. The models include all control variables reported in Table 3 and defined in Appendix 1, for which results are not tabulated for brevity. Standard errors are clustered at the firm level
*, **, ***, denote significance at p < 0.10, p < 0.05, and p < 0.01, respectively
Model 2 interacts each of these controls with accruals quality. The main insight is that fewer board meetings would be necessary to deal with these decisions when accruals quality is high. Mostly consistent with expectations, we find that accruals quality mitigates the effect of business decisions on board meetings in the presence of negative stock returns, and a CEO change, and when the firm engages in asset disposals and larger corporate payouts. Together, the evidence from Table 10 provides some assurance that accounting quality has an effect on board meetings even in the presence of various business problems / challenges.14 In sum, the robustness tests reported in this section are largely consistent with earlier findings and provide added confidence about our interpretation of the results. However, notwithstanding such test results, we refrain from drawing strong conclusions about causality.15
7 Conclusions
Accounting information constitutes a key source of hard information on which boards of directors rely in carrying out their advisory and monitoring tasks. In this paper we hypothesize that the quality of accounting information influences the board's information-gathering and decision-making operations, as revealed by the frequency of board meetings. We posit that boards are more active when financial reporting quality, as proxied by accruals quality, is low. Studying 2,721 Russell 3000 firms between 1996 and 2017 we find empirical evidence that is consistent with predictions. In further tests we find a moderating role of outside directors, ownership structure, and SOX in the relation between earnings quality and the demand for soft information by boards. Our inferences are corroborated by results from alternative tests linking board meeting frequency to management earnings forecasts, the incidence of financial restatements, and material weaknesses in internal controls. The study's main inferences withstand additional tests on several alternative accounting quality measures. Last, to address identification concerns against the backdrop of our largely association-based evidence, we complement our inferences using Generalized Method of Moments methodology.
Our study contributes to extant knowledge in different ways. Academically, it adds to the finance literature on the determinants of board structure by documenting the relevance of financial statement properties in explaining board meeting frequency. It also contributes to a growing body of accounting research positing that earnings quality affects corporate governance outcomes. For practice, this study provides a useful perspective to board chairpersons and CEOs who are responsible for formally informing the board by highlighting the value of high-quality accounting information for directors, and raises the issue of structuring board meeting days and surrounding activities so as to facilitate the exposure of directors to high-quality firm-specific information. For market participants, the results bring board meeting frequency to the forefront as a useful, potentially understudied dimension of board operations.
Getting inside the black box of board activity in general, and board meetings in particular remains a major challenge for students of boards of directors. In-depth field research analyzing the various forms of soft information to which directors are exposed, and their complementary role vis-à-vis accounting information, would help advance our understanding of the link between information quality and director effectiveness. Ultimately, in addition to the well-documented importance of director qualifications, affiliation, and incentives, corporate directors need access to high-quality information. Our study takes a useful step in this direction by bringing to the forefront the relationship between hard accounting information and complementary soft information in board workings.
Declarations
Conflict of interest
This manuscript has not been published and is not under consideration for publication elsewhere. We wish to confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome. All authors have approved the manuscript and agree with its submission to the Review of Quantitative Finance and Accounting.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The number of board meetings held annually excluding actions by written consent
CEO Chair
One if the firm's CEO chairs the board, zero otherwise
Board Size
Total number of directors serving on the board
% NE Directors
Fraction of non-executive directors serving on the board
% Insider Ownership
Fraction of common stock beneficially owned by officers and directors as a group
Earnings Precision
AQ
The standard deviation of the firm-level residuals from the Dechow and Dichev (2002) model during years t−4 to t, multiplied by negative one. The model is a regression of working capital accruals on lagged, current, and future cash flows plus the change in revenue and PPE. All variables are scaled by average total assets. The model is estimated cross-sectionally for each industry with at least 10 observations in a given year based on two-digit SIC classification
Earnings precision-alternative
ABS(AAadj)
The absolute value of the difference between the firm's discretionary accruals and its closest firm in terms of return on assets within the same two-digit industry multiplied by negative one. Discretionary accruals are estimated using the Jones (1991) model. The model is estimated cross-sectionally for each two-digit SIC with at least 10 observations in a year
EPSFREQ
The number of annual management earnings forecasts during a fiscal year
DQ
The Chen et al. (2015) measure of financial data disaggregation in the annual report
FOG
Financial statement readability as computed by Li (2008), multiplied by negative one
Information asymmetry
Std. Dev. Returns
The standard deviation of daily returns during the fiscal year
Idiosyncatic Risk
The standard deviation of the residual returns from a market model regression of daily returns on the value weighted market returns during the fiscal year
Bid-Ask Spread
The average of the daily bid-ask spread over a year scaled by the midpoint between bid and ask prices
Info Asymmetry
The average of the standardized values of the standard deviation of daily stock returns, the bid-ask spread, and idiosyncratic risk
Firm characteristics
Stock Return
The cumulative monthly stock return during the fiscal year
R&D
The Research and Development expense scaled by total assets
# of Analysts
The number of analysts that issued an annual earnings forecast for the firm during the year
Book Leverage
The book value of short-term and long-term debt scaled by the book value of assets
Acquisitions
The value of acquisitions (AQC) reported by Compustat, scaled by the book value of assets
Firm Age
The number of years between a firm's current fiscal year and its first fiscal year-end
# of Segments
The number of business segments reported in the Compustat Segments database
Assets ($milion)
The book value of assets
# Employees (thousand)
The total number of firm employees
Innate factors of earnings quality
Std. Dev. CFO
Standard deviation of cash flow from operations scaled by total assets from years t−4 to t
Std. Dev. Sales
Standard deviation of sales scaled by average total assets from years t−4 to t
Operating Cycle
The sum of receivables to sales plus inventory to cost of goods sold multiplied by 360
Consistent with this notion, Kim and Oh (2023) find evidence of insider trading activity by outside directors based on the information they receive around board meeting dates.
We consider board meeting frequency to be a reasonable albeit imperfect proxy for information acquisition and decision-making operations at the board level. Admittedly, other relevant but unobservable inputs are pertinent including informal contact among directors, visits to the firm’s facilities by directors, and the duration of and topics discussed in each board meeting.
In line with the empirical findings, the vast majority of proxy statements of publicly traded firms highlight the vital role of board meetings in facilitating the boards’ oversight responsibilities. In addition, several nominating committees consider the directors’ ability to participate in board meetings and past meetings attendance for election or re-election decisions.
The change in working capital accruals is defined as the change in current assets less the change in current liabilities less the change in cash plus the change in short term debt in current liabilities: ΔWCA = ΔCA − ΔCL − ΔCASH + ΔSTDEBT.
In robustness tests (untabulated), our results hold when we introduce additional controls for institutional ownership, the issue of debt or equity, dividend payments, sales growth, restructuring, capital intensity, intangibles intensity, investment cycle, operating cycle, foreign operations, industry concentration, CEO turnovers, and accounting performance. To preserve model parsimony, we do not include these additional controls in our main specification. Finally, our results hold when we include CEO-firm fixed effects to control for time-invariant CEO-firm heterogeneity, and industry × year fixed effects to control for macro-economic conditions that affects all firms in a given industry and year.
Admittedly, there are various reasons for differences in board meeting frequency across firms, including a provision for a minimum number of meetings annually in a firm’s by-laws, or certain events requiring special meetings. Including firm fixed effects in the model accounts for this possibility. Further, each firm’s by-law provisions may themselves be the result of self-selection on the basis of a firm’s governance structure and information environment (Gillan et al. 2011).
For economy in presentation, although we include all control variables in these and subsequent models, we do not tabulate coefficients and standard errors for the non-governance control variables in 4, 5, 6, 7 and 8. Similar to Tables 3 and 4 all such variables are included in the models with minimal variation in the results.
Apart from the above proxies of reporting quality, we have replicated our analysis using the absolute value of discretionary accruals estimated using the modified Jones model (Dechow et al. 1995), and a proxy for earnings timeliness measured as the R2 from firm-specific reverse earnings-returns regressions (Bushman et al. 2004). In addition to examining each alternative accounting quality proxy separately, we also replicate our results using a composite index for financial reporting quality. Specifically, we computed a financial reporting quality index as the standardized average of AQ, EPSFREQ, ABS(AAadj), and DQ.
Results are similar if we define the number of management forecasts in each of the two categories as the log of one plus the number of such forecasts We also repeated our main tests after including all observations in the regression, including those with no recorded management forecast in IBES, and assigning a value of zero where no management earnings forecast was issued. The results on this expanded sample are again very similar to the results reported in model 4 of Table 6. Finally, results are very similar when we include accruals quality as an additional explanatory variables.
The timing of "matching" is important, placing board meeting dates around forecast dates. That is, we would expect more frequent board meetings after there is a realized non-match between management and analyst forecasts. However, given that only the number of total board meetings in a year is disclosed, we are unable to place the timing of board meetings around forecasts. As a further empirical test to this end, we annualize both board meetings and management forecasts and, using lead and lag variables, we examine whether board meeting frequency is greater following a year when analyst and manager forecasts exhibit greater disparity, in line with a more opaque information environment. The empirical result is in line with this expectation.
Notably, when we exclude the material restatement interactive term, the post-restatement variable becomes positive and significant in model 1, but not model 2,. The results in models 1 and 2 also hold when we exclude the restatement year (i.e., controlling for the possibility the restatement year effect captures the board’s effort to deal with the restatement itself).
The incremental board meetings of the ICMW_fixed firms relative to control firms is given by the sum of the two coefficients. Consistent with evidence presented by Ashbaugh-Skaife et al. (2008), the summation of these two coefficients is not significantly different from zero (untabulated).
Our AQ measure of accruals quality is estimated using data points over time aggregated by firm, and do not lend itself to measuring changes from one year to the next. Our results are robust when we use the Chen et al. (2015) disaggregation quality measure, as a measure of financial statement quality.
The possibility contemporaneous business problems may alternatively drive the results is also pertinent in the interpretation of the negative AQ x SOX coefficient from Table 4; i.e., that the change in meeting frequency is related to a lower need to address business problems following improved corporate governance quality after SOX. Accordingly, in additional tests (not tabulated) we examined the moderating effect of SOX after controlling for the effect of other business decisions on board meetings. We did so by (i) including an AQ x SOX interactive term in model 1 of Table 10, and (ii) including three-way interactions of AQ x SOX x business decisions in model 2 of Table 10. The interactive term in (i) is negative and statistically significant, as are most interactions with individual business decisions in (ii), in line with our earlier evidence. These results provide some comfort that, even though business decisions are evidently incrementally important in explaining board meeting frequency, accounting information quality contributes towards explaining the intensity of board activity beyond such issues.
Our analysis in this paper is largely carried out at-equilibrium and does not explore out-of-equilibrium scenarios studying whether soft information can in fact compensate for inefficiencies in the production of hard accounting information. In the absence of such evidence we cannot make definitive policy recommendations mandating the exposure of corporate directors to soft firm-specific information. Nevertheless, our results are suggestive of and leave open the possibility that exposing the firm’s directors to a stimulating soft information environment, such as through meetings with senior firm administrators, employee representatives, investment officers and other firm stakeholders might enhance director effectiveness. Further, greater transparency surrounding board meetings at the end of each year, such as the disclosure of the specific dates in which meetings took place, director attendance of individual meetings and other rudimentary information, all subject to proprietary information cost constraints, could prove to be useful to market participants.
