Journal of Economics, Finance and Business Analytics
2024; 2(1): 34-44
http://www.quantresearchpublishing.com/ /index.php/jefba
ISSN:3006-0745 (Online)
Causal Effect Relationship between Audit Committee
Composition and Earnings Management among Listed
Manufacturing Firms in Kenya: A Panel Fixed Effect Model
Analysis
Mark Okeyo Okeyo and Winnie Nyamute
Finance and Accounting Department, University of Nairobi, Nairobi, Kenya
To cite this article:
Okeyo, M. & Nyamute, W. (2024). Causal effect Relationship between Audit Committee Composition and Earnings Management among
Listed Manufacturing Firms in Kenya: A Panel Fixed Effect Model Analysis. Journal of Economics, Finance and Business Analytics, 2 (1),
34 - 44
Received: 05 January 2024; Accepted: 25 January 2024; Published: 31 January 2024
Abstract: This project aimed to examine the effect of Audit Committee (AC) composition on Earnings Management (EM)
among manufacturing firms listed in Kenya. The components of AC targeted included independence, gender diversity,
expertise, age and size. The study relied on agency, stewardship and stakeholder theories. The project was based on descriptive
design where all the nine manufacturing firms listed at the NSE were used in the study. The research adopted annual secondary
data extracted from annual reports of the manufacturing firms. The data covered the period from eight years from 2013 to 2020
and was recorded on data collection sheets. The data in the data collection sheets were entered into a Microsoft Excel sheet.
The data was panel in nature covering seven years and nine firms giving rise to 72 observations. The prepared data was
exported to STATA version 15 before analysis was carried out. Measures of central tendency and dispersal including standard
deviation, mean, kurtosis and skewness were generated to identify outliers and general distribution of data. Tests of OLS
regression assumptions were carried out to enable the choice of the most suitable model of analysis. Finally, panel data
regression was adopted. The coefficient of determination was 0.7531 implying that Audit committee composition explains
75.31% of the total variation in earnings management. The study further revealed that AC Independence, gender diversity and
expertise had a significant inverse effect on EM. AC size had a direct and significant effect on EM. The effect of AC age
diversity on EM was direct and significant. The study concluded that AC composition had a critical role in minimising EM
among the firms studied.
Keywords: Audit Committee Composition, Earnings Management, Audit Committee Independence, Audit Committee
Gender Diversity, Audit Committee Expertise, Audit Committee Size, Audit Committee Age Diversity.
1. Introduction
Globally, in an effort towards improving board governance, sub-board committees are established and delegated with
various functions. A key subcommittee charged with ensuring the quality of books of accounts is the Audit Committee (Raimo,
Vitolla, Marrone & Rubino, 2021). The link between Audit Committee (AC) composition and Earnings Management (EM)
has been examined in the empirical literature with aspects of AC composition examined including Audit Committee
independence, gender diversity, experience, size and age diversity. The link subsisting between AC independence and EM has
often been inverse in most studies (Abubakar, Usmam. Anuforo, Alhaji, 2021). Additionally, the link between gender diversity
of the AC and EM has been inverse in various studies (Sudarman & Hidayat, 2019. Literature has also tended to establish an
inverse link subsisting between AC experience and EM (Kapkiyai, Cheboi & Komen, 2020). Finally, the causal effect link
between AC size and EM has tended to be positive in various studies (Isa & Farouk, 2018). AC composition is a collective
term that refers to AC independence, gender diversity, expertise, size and age (Moses, 2019). AC independence is the
condition that enables the committee to perform its duties objectively (Isa & Farouk, 2018). AC gender diversity is several
male or female directors within the committee as a proportion of the AC size (Chijoke-Mgbame, Boateng & Mgbame, 2020).
Journal of Economics, Finance and Business Analytics 2024; 2(1): 34-44 35
AC expertise is relevant auditing, accounting and financial technical skills possessed by the members of the AC (Zalata,
Tauringana & Tingbani, 2018). AC size is the enumeration of the directors serving in the AC. Finally, AC age diversity
represents the average age of the directors serving in the AC (Juhmani, 2017).
The composition of the AC determines its effectiveness (Juhmani, 2017). "It is paramount that the AC members be qualified
and have adequate academic and work experience, especially in the area of finance, auditing and accounts. There ought to be
diversity in the AC to improve the quality of output and decisions arrived at the AC meetings (Waweru, 2018). To enhance
accountability and objectivity, the firm should ensure that the AC is properly constituted in terms of independence, experience,
leadership and overall business knowledge (Hastuti, Setiawan & Widagdo, 2020). An effective AC is critical in assuring that
financial reports are not compromised in terms of manipulation of books of accounts in what is referred to as EM (Al-Absy,
Ismail & Chandren, 2018).
Various authorities have defined EM; however, there is no single definition of the concept that is generally adopted. El Diri
(2017) defined EM as a management discretion action to report financial statements that depict a picture of stable financial
performance. In a related definition, Kim, Kim and Zhou (2017) defined EM as conscious decisions taken by those preparing
financial statements and reports according to generally accepted accounting principles (GAAPs) to present an acceptable
financial performance to users of accounting information. In addition, Moratis and van Egmond (2018) define EM as the
manipulation of actual financial records by taking advantage of loopholes in the GAAPs. The minimization of EM is very
critical given that the executive in most cases makes decisions that are contrary to what shareholders would wish (Kjærland,
Haugdal, Søndergaard & Vågslid, 2020). Managers often take advantage of managerial discretion in earnings reporting
especially where controls are not adequate leading to EM. Such actions of EM lead to financial reports that are misleading the
users of accounting information (Harris, Karl & Lawrence, 2019). EM results from differing and contrasting interests of the
executive and the common stockholders. Therefore, it has become necessary for shareholders to incur monitoring costs by
employing directors who sit in the AC to ensure that the financial reports are a true reflection of the status of books of accounts
(Jordaan, De Klerk & De Villiers, 2018).
The composition of AC among the boards of listed manufacturing firms is very essential in the minimization of EM. The
empirical literature existing globally has consistently established that EM among listed companies hinges on some critical
aspects of AC composition including independence, expertise, gender diversity, size and age. The link existing between AC
independence and EM has often been inverse in most studies (Abubakar, Usmam. Anuforo & Alhaji, 2021). Additionally, the
causal effect link between AC gender diversity and EM has been inverse in various studies (Sudarman & Hidayat, 2019).
Literature has also tended to establish an inverse causal effect link between AC experience and EM (Kapkiyai, Cheboi &
Komen, 2020). Finally, the causal effect link between EM and AC size has tended to be positive in various studies (Isa &
Farouk, 2018).
There are nine listed manufacturing and allied companies in Kenya including Carbacid Investments Ltd, East African
Breweries Ltd, Mumias Sugar, British American Tobacco Kenya, Flame Tree Group Holdings Ltd, Unga Group Ltd, B.O.C
Kenya Limited, Eveready East Africa Ltd and Kenya Orchards Ltd (NSE, 2020). The firms have ACs of varying composition
including independence, gender diversity, size, expertise and age diversity. In the list of the listed manufacturing companies,
Mumias Sugar halted most of its operations after a series of loss-making that could have been concealed by EM carried out
before 2013 when the company was reporting profitability (Birgen & Bogonko, 2018). Eveready East Africa Ltd too has been
reporting loss making. The government has significant shareholding in the two companies hence their failure in recent times is
partly because of systematic failures by the Board of management especially the AC leading to the concealment of poor
performance for some time before their poor performance became known (Sanghani, 2014).
Globally, in Indonesian firms, Siagian and Siregar (2018) evaluated the influence of AC financial skills on EM. The study
revealed that AC financial expertise had a minor effect on EM. In Malaysia, Ghaleb, Al-Duais and Hashed (2021) evaluated
the contribution of AC chairs' legal expertise to Real EM in energy and utilities companies. The study revealed that AC chairs
with legal expertise was directly associated with Real EM practices. In the Netherlands, Mardessi and Fourati (2020) examined
the association between the quality of financial reporting and AC for companies with stock floated at the Amsterdam Stock
Exchange from 2010 to 2017. The study revealed that the possibility of the AC to lower real EM was based on AC gender
diversity. In Nigeria, Abubakar, Usman, Anuforo, and Alhaji (2021) evaluated the influence of AC attributes on EM. The
finding shows that AC size prevents managers' activities in earnings manipulations. In addition, AC financial expertise was
useful in curtailing earnings manipulation practices. Locally in Kenya, Kariuki and Aluoch (2020) evaluated the contribution
of AC size on the financial reporting quality in firms listed at the NSE. The study concluded that there was a direct impact of
AC size on the quality of financial reports.”Kapkiyai, Cheboi, and Komen (2020) examined the causal effect link between the
effectiveness of audit committees and EM practices in listed firms at the Nairobi Securities Exchange (NSE). The study used a
panel regression model with results showing that an effective AC aids in reducing EM. Further, having more meetings, and
more directors who are independent and AC with adequate financial skills reduces EM. Kapkiyai, Cheboi and Komen (2020)
excluded AC characteristics like AC gender diversity and age. The current study went a step further by seeking an answer to
the research question, what is the effect of AC composition on earning management among listed manufacturing firms in
Kenya?
36 Okeyo & Nyamute (2024): Causal effect Relationship between Audit Committee Composition and Earnings Management
2. Literature Review
2.1 Theoretical Review
2.1.1 Agency Theory
Mitnick (1973) advanced and Jensen and Meckling (1976) through the theory the advanced explain that an agency relationship
exists where one party (principal) contracts the services of another party (agent) to represent it in a business transaction for
some fees such that the principal delegates some authority and responsibility to the agent (Jensen & Meckling, 1976). In
addition, agency conflict emerges when the principal' (Shareholder) and agent' (Executive) interests conflict hence the
managers take actions that may not be in congruence with shareholders worth (Dalton et al., 2007). To minimise agency
conflict, the principal (shareholders) may take certain actions that involve putting in place control mechanisms. Such control
mechanisms are often called agency costs (Jensen & Meckling, 1976). The major agency costs include the cost of employing
directors to serve on the board and monitor the work of the executive. The cost of contracting an external auditor to examine
books of accounts and ensure they are a true reflection of the financial status of the firm. Notwithstanding its strengths, agency
costs are an added cost in the operation of the firm hence eating into the profits of the shareholders (Wiseman,
Cuevas‐Rodríguez & Gomez‐Mejia, 2012). In addition, always acting in the best interest of the shareholders ignores other
stakeholders who have a stake in the firm like the government, employees, customers, competitors, suppliers and the public
(Bosse & Phillips, 2016). The adherence to shareholders' interests may also distort and interfere with other equally important
affairs like strategic focus, and investment plans hence minimising commitment to the creation of economic value (Eisenhardt,
1989). The theory is applicable in the study as it informs the variable AC composition and EM. To minimise the manager's
discretionary accruals where managers manipulate books of accounts to present an acceptable financial performance. Listed
firms employ directors to serve on the board of directors of which some are expected to serve in the AC to help in evaluating
books of accounts and control within the organization to minimise EM by the executive. The AC is expected to monitor the
work of the executive and ensure financial reports represent the true and fair view of the status of the firm."
2.1.2 Stakeholder Theory
The theory proposed by Freeman (1984) posits that firms exist to meet the interests of those having an interest in the operations
and future of the organization and not just the shareholders. The attention of the theory is to those stakeholders that can hinder
or improve the chances of the business to survive and thrive in the business environment (Phillips et al., 2003). The stakeholder
theory also focuses on the responsibility of the executive to take care of the interests of all key stakeholders. Post, Preston and
Sachs (2002), stated that stakeholders are individuals groups or other organizations that have a direct interest in the operation
of the firm and are affected and can affect the operation, actions, goals and policies of the firm in question. Major stakeholders
may include creditors, owners, directors, employees, government, unions, suppliers and the local community where the firm
gets inputs and sells output (Kaczmarek, Kimino & Pye, 2014). To meet the concerns and interests that are often competing as
well as conflicting, the firm needs to find the right balance where the critical segment of the stakeholders are considered first
before the other less important stakeholders (Bridoux & Stoelhorst, 2014). Nonetheless, the theory has received its share of
criticism, especially regarding the fact that it is not practical to meet the needs of all stakeholders without endangering the
shareholders worth (Blattberg (2004). In addition, the theory has been criticised for not fitting the capitalist world where the
interest of the owners of capital (shareholders) comes first (Mansell, 2013). In summary, the theory supports agency theory in
that it extends the boundary of agency theory from just focusing on shareholders to looking into other stakeholders' interests
too. The theory underpins corporate governance where the company is expected to balance the different divergent interests of
various stakeholders of the firm. The company is expected to operate under the dictates of the corporate governance principles
suggesting that a company should have directors serving in the AC to ensure that managers do not manipulate cooks of
accounts. This is done to ensure that financial reports are of high quality and present the fair and true economic status of the
firm such that different stakeholders can rely on them for various purposes and decisions.
2.2 Empirical Review
Kapkiyai, Cheboi, and Komen (2020) examined the causal effect link between the effectiveness of audit committees on EM
practices in listed firms at the NSE. The study adopted longitudinal data for 13 years from 2004 to 2017. The research adopted
panel regression with results showing that an effective AC aids in reducing EM. Further, having more meetings, more directors
that are independent and AC with adequate financial expertise reduces EM. Kapkiyai, Cheboi and Komen (2020) excluded
other AC attributes such as gender diversity of the AC as well as the age of the AC. The current study goes a step further by
examining the role of AC gender and age diversity on EM. In a study in Nigeria, Abubakar, Usman, Anuforo and Alhaji (2021)
evaluated the causal effect link subsisting between EM and AC attributes. The sample size was seventy-two firms that were
listed in the securities exchange market studied from 2014 to 2018. The study adopted the PCSEs model. The finding shows
that AC size prevents managers' activities in earnings manipulations. In addition, the result establishes that AC independence
Journal of Economics, Finance and Business Analytics 2024; 2(1): 34-44 37
presence on the AC controls managers' opportunistic behaviour. Abubakar, Usman, Anuforo and Alhaji (2021) were carried out
in Nigeria and hence may not be wholesomely adopted in Kenyan firms due to different operating environments.
In an examination of Indonesian firms, Siagian and Siregar (2018) evaluated the relationship existing between the financial
expertise of AC and EM. The sample size consisted of three hundred and eighty-four observations for the period 2012- 2014.
The research used Random Effect Model (REM) regression with results revealing that the financial expertise of AC had a
minor inverse influence on EM. Siagian and Siregar (2018) did not establish any major impact of AC financial expertise on
EM contrary to general expectations. Besides, the study concentrated on one aspect of AC composition and a study examining
other aspects of AC composition will expand the applicability of the study. Sudarman and Hidaya (2019) evaluated the causal
effect link subsisting between gender diversity in the AC and EM among Indonesian firms that had floated shares. The study
covered all the listed firms with data collected between 2013 and 2017. The study adopted multivariate regression analysis.
The findings revealed that AC gender diversity contributed to minimising EM. Sudarman and Hidaya (2019) focused on
gender diversity as an aspect of AC composition while ignoring other aspects. Another study comprehensively examining
aspects of AC composition on EM is therefore critical.
Setiawan, Phua, Chee and Trinugroho (2020) evaluated the causal effect link between EM AC characteristics among firms.
The research examined three hundred and ninety-three Indonesian organizations that had listed their common stock on the
stock exchange market. Data was collected between 2006 and 2010 period. The research used REM and FEM where results
showed that AC attributes had a major impact on EM practice. Further, AC gender diversity had an inverse influence on EM.
However, AC meetings frequency and expertise had a direct influence on EM. In addition, AC size and independence did not
show a significant effect on EM. However, Setiawan, Phua, Chee and Trinugroho (2020) find a direct effect of financial skills
and meeting frequency on EM contrary to theoretical expectations. Ghaleb, Al-Duais and Hashed (2021) analysed the causal
effect link between AC chairs’ legal expertise and Real EM in Malaysian energy and utilities companies between 2013 and
2018. The study was a sample of two hundred and twenty-nine companies. The study finds that AC chairs with legal skills are
significantly and positively associated with Real EM practices. The findings are inconsistent with those other studies and
another study ought to be done to clear out the inconsistency.
In the Netherlands, among firms listed on the Amsterdam Stock Exchange, Mardessi and Fourati (2020) examined the
relationship between the quality of financial reporting and AC for the period 2010 to 2017. The research used a panel
regression model. The findings showed that the capability of AC to reduce real EM was based on AC gender diversity.
However, Mardessi and Fourati (2020) was limited to gender diversity and another ought to be carried out AC attributes such
as the financial experience of the AC. Dakhlallh, Rashid, Abdullah, and Shehab (2021) examined the influence of board
structure on Real EM and the moderator influence of AC independence on their link. Using panel data from public companies
in Jordan from 2009 to 2018, the study showed that the structure of the board affects the decisions of companies to manipulate
reported earnings." Board independence exaggerates the incidence of all Real EM. CEO duality only exaggerates Real EM. AC
independence weakens the influence of the board size to curb all real EM proxies and AC independence enhances the ability of
the board to curb all real EM. AC independence enhances the impact of CEO duality on curbing real EM. Dakhlallh, Rashid,
Abdullah and Shehab (2021) adopted AC independence as a moderating variable. Therefore, a study examining the direct
influence of AC independence on EM is necessary.
Kariuki and Aluoch (2020) evaluated the contribution of AC size to the quality of financial reporting among firms listed at
the NSE. The study was a census of all the 62 listed firms used as a unit of analysis. Annual data spanning 5 years from 2014
to 2018 was used in the study. The research concluded financial expertise of the AC had a direct influence on the quality of
financial reports for listed firms at the NSE. The study focused on only one aspect of AC composition. In a study of Australian
firms, Mollik, Mir, Monir, McIver and Bepari (2020) evaluated the direction of causation between EM and the effectiveness of
AC. The data was collected from 2006 to 2009 considered as the GFC period. The research adopted FEM where the findings
showed that improved audit quality reduced EM in the PCP; however, during GFC AC did not influence EM. Moreover, the
expertise of AC did not influence EM. Mollik, Mir, Monir, McIver and Bepari (2020) reported future research should examine
the effects the effect of other AC characteristics during and before global shocks.
3. Methodology
3.1 Research Design
A descriptive research design was adopted in this study. Descriptive designs are used in studies that do not influence the
environment in which the variables are interacting but rather report on ex post facto relationships after variables have
interacted in their natural environment (Kothari, 2004). Descriptive design is also concerned with establishing the causal effect
relationship after the collection and analysis of relevant data using scientific methods. The design enabled the researcher to
collect secondary annual data regarding AC compositions and EM to evaluate the causal effect link between AC compositions
and EM.
3.2 Study Population
38 Okeyo & Nyamute (2024): Causal effect Relationship between Audit Committee Composition and Earnings Management
The research targeted all manufacturing firms listed at the NSE as of 31 December 2020. There were nine (9) listed
manufacturing firms in NSE (www.nse.co.ke). The study was a census of all listed manufacturing firms at the NSE therefore
no sampling was carried out with data covering the period of eight years from 2013-2020. Listed manufacturing firms were
targeted since most EM happens in firms that have major inventory and manufacturing firms are a classic example of firms
with the majority of resources tied in receivables and inventory. Additionally, listed manufacturing firms do publish audited
financial statements hence they have ready secondary data.
3.3 Data Collection
The research adopted panel data extracted from annual reports of the manufacturing firms. The data was specifically
extracted from the financial statements and statement of corporate governance. Regarding AC composition, the specific data
collected included the number of members in the AC, the number of independent non-executive directors in the AC, the
number of female members of the AC, number of AC members with finance/accounts/auditing experience and qualification.
Regarding the EM, the specific data collected included net income from total assets, operations, cash flow from operating
activities, debtors and net property, plant and equipment of all listed manufacturing companies. The data covered the period
from eight years from 2013 to 2020 and was recorded on data collection sheets.
3.4 Data Analysis
The data in the data collection sheets will be entered into a Microsoft Excel sheet. The data in the Excel sheet were
examined for completeness before the study variables were computed. The data was panel in nature covering seven years and
nine firms giving rise to 72 observations. The prepared data was exported to STATA version 15 before analysis where
measures of central tendency and dispersion such as standard deviation, mean, Kurtosis and skewness were generated to
identify outliers and general distribution of data. Diagnostic tests were carried out to enable choice of the most suitable model
of analysis. Finally, panel data regression was adopted. The panel data regression model adopted is presented in equation [1].
EMit = β0+ β1Ind1it + β2GD2it+ β3Expertise3it+ β4size 4it+ β5Age5it+ ɛit ...................................................................................[1]
Where:
EM = Earnings Management is measured by discretionary accruals (the difference between total accruals and non-
discretionary accruals).
Ind = Audit committee independence is measured by the ratio of independent non-executive directors in the committee to the
size of the audit committee.
GD = Gender diversity is measured by the ratio of female directors in the audit committee to total membership of the Audit
Committee.
Expertise = Audit committee expertise is measured by the ratio of membership of the committee having finance/accounts and
auditing training and experience to the total membership of the audit committee.
size = Audit committee size is measured as the number of members of the audit committee.
Age =Audit committee age diversity is measured as the average age of the members of the Audit Committee.
β0 is the intercept term, β1- β5 are the coefficient of independent variables.
ɛ =Error term, i= firm 1, 2, 3...9 and t = Time period 2012, 2013, 2014.............................2020
3.5 Diagnostic Test
Diagnostic test was carried out before inferential analysis to examine the robustness of the regression model. The diagnostic
tests were carried out to ensure classical least squares assumptions were not violated. The test included heteroscedasticity,
autocorrelation, normality, multicollinearity and unit root diagnostic tests. A normality test is carried out to ensure data points
are normally distributed as given by mean equaling to median (Kothari, 2004). The study adopted Kurtosis and skewness to
examine the normality of data such that when the skewness statistic is zero and Kurtosis is 3 then the data is perfectly normal.
Further, the Shapiro-Wilk test was adopted to test the normality of the data where non-normal data have a p-value less
than .05 level of significance. According to Gujarati (2003), data is said to be homoscedastic if the error term depicts constant
variance and mean. This means that the residuals are dispersed evenly around the mean with constant variance on either side of
the mean. The absence of homoscedasticity is referred to as heteroscedasticity where the error terms are non-constant.
Heteroscedasticity results in spurious regression where the standard errors are over-identified and misleading. The study used
the Modified Wald test for the presence of group heteroscedasticity. The study concluded the presence of homoscedasticity if
the p-value generated is greater than 0.05 level of significance. In the presence of heteroscedasticity alone, the study can use
robust standard errors. According to Kothari (2004), multicollinearity exists where the explanatory variables used in the study
Journal of Economics, Finance and Business Analytics 2024; 2(1): 34-44 39
are highly correlated among themselves and with the error terms. Cooper and Schindler (2006) noted that the presence of
multicollinearity results in inflated parameter estimates. The research adopted variance Inflation Factor (VIF) to examine the
presence of multicollinearity. VIF values greater than 5 signify the presence of multicollinearity (Kothari, 2004).
Autocorrelation is said to exist only and only if residuals in one period are highly correlated with error terms in successive
periods (Gujarati, 2008). The study adopted the Wooldridge Drukker test to evaluate the presence of autocorrelation where a
probability value greater than 0.05 will be taken to imply the absence of autocorrelation. In case the researcher establishes the
presence of autocorrelation of order one alone, the researcher will adopt panel-correlated standard errors (PCSEs) (Wooldridge,
2013).
3.6 Random or Fixed Effects
The research employed the Hausman test to decide on the suitability of random effects and fixed effects models.
Wooldridge (2013) asserts that the Random effects model is preferred since the fixed effects model is only efficient in
producing acceptable estimates when the data being analyzed suffers from correlation issues. Further, the fixed effects model
may not be most appropriate if there is little variability of variables across time (Allison, 2009). If the P-value is greater than a
5% level of significance, a random effects model should be used. To confirm the significance of the effect of AC composition
on EM, the study used an F-test where a p-value less than 5% level of significance signifies a significant effect. The p-values
associated with the coefficients of the independent variables were used to test the significance of the effect of each explanatory
variable on EM. The p-values less than 5% level of significance show a significant effect of the individual explanatory variable
on the dependent variable.
4. Results
4.1 Descriptive Analysis
The study examined the distribution of the explanatory and outcome variables in terms of standard deviation, mean,
minimum and maximum. The purpose of descriptive analysis was to establish the general nature of the data to be used for
inferential analysis. The findings are presented in Table 1.
Table 1: Summary of Descriptive Statistics
Note: Gender Diversity (GD), Age diversity (Age), independence (Ind), Expertise (Expertise), size (size) and Earnings Management (EM).
Table 1 presents the descriptive statistical analysis. AC gender diversity was measured by the ratio of female directors in
the audit committee to the total number of members of the audit committee. The mean gender diversity was 0.44 implying that
there was an average of 44% of the directors in the audit committee being female. The standard deviation showed that the AC
diversity of individual firms was spread around the mean with about 0.16 points. A minimum of 0.18 gives the firm that had
the lowest AC gender diversity while a maximum of 0.84 gives the firm that had the highest AC gender diversity. Audit
committee age diversity was measured by the average age of the directors in the AC. The mean age was 54.5 with a standard
deviation of 4.1 around the mean age. The minimum of 44 shows the firm that had the youngest audit committee while the
maximum of 63 shows the firm that had the oldest AC. AC independence was measured by the ratio of independent non-
executive directors in the AC to the total audit committee size. The mean audit committee independence was 0.67 implying
that about 67.1% of the audit committee of listed manufacturing firms in Kenya were comprised of independent non-executive
directors. The minimum captured the firm that had the lowest audit committee independence at 23% and the maximum
presented the firm that had the highest Audit committee independence at 90%. Audit committee expertise was measured by the
ratio of members of the AC possessing finance, accounting and auditing training and experience to the total audit committee
size. The mean AC expertise was 0.18 implying that about 18% of the AC membership were those directors that possessed
finance, accounting and auditing training and experience. The standard deviation showed that the audit committee expertise on
individual firms was spread around the overall mean by about 19%. The minimum expertise presents the firm that had the
GD Age Ind Expertise size EM
Mean 0.44505 54.5435 0.67116 0.1861 3.8254 2,530,681
Standard Deviation 0.16957 4.11494 0.17237 0.19333 0.63601 231,924
Kurtosis 3.16686 4.79457 3.2825 3.1374 2.83586 3.639546592
Skewness 0.16059 0.77472 -0.5155 0.44661 0.16007 0.121889848
Minimum 0.1806 44 0.23 0.1 3 83,635
Maximum 0.8412 63 0.9 0.63636 7 13,274,959
Count 72 72 72 72 72 72
40 Okeyo & Nyamute (2024): Causal effect Relationship between Audit Committee Composition and Earnings Management
lowest number of experts in finance, accounting and auditing at 10% while the maximum presents the firm that had the highest
number of experts in finance, accounting and auditing in its audit committee at 0.63%.
Audit committee size was measured by the number of directors in the AC. The mean audit committee was 3.82 implying
most audit committees had about 4 directors. The standard deviation showed that the AC sizes of individual firms were spread
around the mean by about 0.63. The minimum showed that the firm with the smallest size audit committee had three (3)
members while the maximum showed that the firm with the biggest size audit committee had seven (7) members. Finally, EM
was measured by discretionary accruals. The mean EM was about Ksh. 2.53 million with a standard deviation of Ksh.231
thousand. The minimum EM was Ksh.83.6 thousand while the maximum EM was Ksh.13.2 million. The Kurtosis values for all
variables were around three (3) and the skewness was around zero (0) hence the variables were normally distributed.
4.2 Diagnostic Tests
The study adopted Kurtosis and skewness to examine the normality of data such that when the skewness statistic is zero
and Kurtosis is 3 then the data is perfectly normal. Further, the Shapiro-Wilk test was adopted to test the normality of the data
where non-normal data have a p-value less than a .05 level of significance. The finding revealed almost all variables depicted
normal distribution with only age being slightly skewed but generally normal. The study used the Modified Wald test for the
presence of group heteroscedasticity. The study would conclude the presence of homoscedasticity if the p-value generated is
greater than 0.05 level of significance. In the presence of heteroscedasticity alone, the study can use robust standard errors. The
finding showed that the p-value was greater than 0.05 (p= 0.0551) implying homoscedasticity. The research adopted and
variance Inflation Factor (VIF) to examine the presence of multicollinearity. VIF values greater than 5 signify the presence of
multicollinearity (Kothari, 2004). The results revealed that there was no problem of multicollinearity given that all the VIF
values were less than 5 and the mean VIF was also less than 5. The study adopted the Wooldridge Drukker test to evaluate the
presence of autocorrelation where a probability value greater than 0.05 is taken to imply the absence of autocorrelation. The
study revealed that there was no problem with serial correlation given the p-value was greater than 0.05 level of significance.
The research also employed the Hausman test to decide on the suitability of random effects and fixed effects models.
Wooldridge (2013) asserts that the Random effects model is preferred since the fixed effects model is only efficient in
producing acceptable estimates when the data being analyzed suffers from correlation issues. If the P-value is greater than a
5% level of significance, a random effects model should be used. The results showed that the fixed effect model (FEM) was
more appropriate given the p-value was less than 0.05 level of significance.
4.3 Regression Analysis
The study adopted a fixed effect model to examine the effect of AC composition on EM among listed manufacturing firms in
Kenya. The study adopted a fixed effect model given that the Hausman test showed that FEM was more efficient than the REM
given that the p-value was less than 0.05. The regression output consisted of coefficient of determination, F-test and t-test.
Table 2 presented the fixed effect model where the overall coefficient of determination (R2) was 0.7531 implying that AC
composition explains 75.31% of the total variation in EM. The remaining variation of 24.69% was explained by variables that
were not part of the study. The p-value associated with F-test was less than 0.05 implying that the AC composition had a
significant effect on EM among the listed manufacturing firms in Kenya. Further, the study sought to examine the effect of
individual components of AC composition on EM. The effect of AC Independence had a significant inverse effect on EM (β1 =
-1.633, t= - 4.88, p= 0.000< 0.05). The study revealed that the effect of AC gender diversity on EM was inverse and
statistically significant (β2 = -1.27, t= 3.29, p= 0.005< 0.05). AC expertise had a significant inverse effect on EM (β3= -1.083,
t= -2.62, p= 0.047<0.05). The study also revealed that AC size had a direct but not statistically significant effect on EM
4= .1539, t= 0.82, p= 0.414> 0.05). The effect of AC age diversity on EM was direct and statistically significant (β5 =
0.5168, t= 3.78, p= 0.002<0.05). Finally, the intercept term had a coefficient of β0 = .3556 implying that when the explanatory
variables are held constant at zero (0), EM was .3556. The model was thus estimated as:
EMit = .3556- 1.633Ind it - 1.27GDit – 1.083 Expertise it+ .1539 size it+ .5168Age it……………………………………….[2]
Journal of Economics, Finance and Business Analytics 2024; 2(1): 34-44 41
Table 2: Fixed Effect Model
5. Discussion
The study sought to establish the effect of Audit committee composition on earnings management among listed
manufacturing firms in Kenya. The overall coefficient of determination (R2) was 0.7531 implying that AC composition
explains 75.31% of the total variation in EM. The remaining variation of 24.69% is explained by other variables, not within the
scope of this study. Further, the p-value associated with F-test was less than 0.05 implying that the AC composition had a
significant effect on EM among the listed manufacturing firms in Kenya. The finding agrees with Setiawan, Phua, Chee and
Trinugroho (2020) who showed that AC attributes had a major impact on EM practice. Regarding the effect of each aspect of
AC composition on EM, the study adopted the t-test. The study established AC Independence had a significant inverse effect
on EM 1 = -1.633, t= - 4.88, p= 0.000< 0.05). The study showed that for every one-unit increase in AC independence, EM
was reduced by 1.633 units. The inverse relationship implies that AC independence is responsible for reduced earnings
management among the firms. Having more independent non-executive directors in the AC means that they will be objective in
monitoring the accounting controls in place to reduce the chances of manipulation of books of accounts by the executive and
employees working under them. The findings are in agreement with Kapkiyai, Cheboi, and Komen (2020) showed that having
more independent AC reduces EM. Abubakar, Usman, Anuforo and Alhaji (2021) revealed that AC independence controls
managers' opportunistic behaviour of manipulation of books of accounts. The study also revealed that the effect of AC gender
diversity on EM was inverse and statistically significant (β2 = -1.27, t= 3.29, p= 0.005< 0.05). A one-unit increase in gender
diversity leads to a 1.27-unit decrease in EM. Having more female directors in the AC is associated with improved stewardship
over the resources of the firm. Female directors are more conservative hence they tend to encourage the adoption of
conservative accounting concepts and principles that are associated with tight controls over resources hence reduced earning
72
Group variable: var10 Number of groups = 9
ID
Number of groups
= 9
9
R-sq: Obs per group:
8
8.0
8
F(5,58) = 70.81
corr(u_i, Xb) = -0.0417 Prob > F = 0.000
EM Coef. Std. Err. t P>|t| [95% Conf. Interval]
GD -1.276986 .387749 -3.29 0.005 -4.114673 4.668645
Age .516822 .136835 3.78 0.002 0.100421 1.566777
Ind -1.633233 .334666 -4.88 0.000 -1.948878 4.315344
Expertise -1.083274 .413216 -2.62 0.047 -2.757463 3.924015
size .153982 .186811 0.82 0.414 -.2214302 .5293943
_cons .3556423 4.687305 0.08 0.940 -3.063851 9.775135
sigma_u 1.1794627
sigma_e .80386545
rho .68282062 (fraction of variance due to u_i)
F test that all u_i=0: F(8, 58) = 6.18 Prob > F = 0.0000
42 Okeyo & Nyamute (2024): Causal effect Relationship between Audit Committee Composition and Earnings Management
management opportunities. The finding is supported by Sudarman and Hidaya (2019) who revealed that AC gender diversity
contributed to minimising EM. Mardessi and Fourati (2020) also showed that the capability of AC to reduce real EM was
based on AC gender diversity. The study also revealed that AC expertise had a significant inverse effect on EM (β3= -1.083, t=
-2.62, p= 0.047<0.05). A one-unit increase in AC expertise was associated with a 1.083 reduction in EM. Having more AC
members with expertise in finance, accounting, and auditing was advantageous to the committee as they could review audit
reports with knowledge. Experienced AC members can review various controls in place in the firm to identify weaknesses to
make recommendations for their strengthening. The strengthened controls reduce the loopholes that are exploited by executives
and employees working under them in manipulating books of accounts. The findings agree with Siagian and Siregar (2018)
revealing that the financial expertise of AC had an inverse influence on EM. Kariuki and Aluoch (2020) also revealed that the
financial expertise of the AC had a direct influence on the quality of financial reports for listed firms at the Nairobi Securities
Exchange. However, Mollik, Mir, Monir, McIver and Bepari (2020) had contrary findings showing that the expertise of AC did
not influence EM.
The study also revealed that AC size had a direct but not statistically significant effect on EM (β4= .1539, t= 0.82, p= 0.414>
0.05). A one-unit increase in AC size was associated with an increase in EM by .1539. The direct effect implies that having a
larger than necessary AC size is associated with poor deliberations at the audit committee hence encouraging. However, the
effect was not significant implying that the size of the AC was not a major factor when other aspects of AC composition were
as expected. The finding conflicts with Kapkiyai, Cheboi, and Komen (2020) who showed that having more directors reduces
EM. Abubakar, Usman, Anuforo and Alhaji (2021) also showed that AC size prevents managers' activities in earnings
manipulations. The research showed that the effect of AC age diversity on EM was direct and statistically significant (β5 =
0.5168, t= 3.78, p= 0.002<0.05). A one-unit increase in AC average age results in increased earnings management. The positive
relationship could imply that the increased average age of the AC beyond a given point hinders their ability to minimise EM.
Having very old directors may lead to increased chances of EM management by the executive as the directors in AC with
advanced age may not manage to rigorously monitor internal controls as well as books of accounts.
6. Conclusions
6.1 Conclusion
The study noted that AC composition had a critical role in minimising EM among the listed manufacturing firms in Kenya.
Further, the research noted that having more independent non-executive directors in the AC means that they will be objective
in monitoring the accounting controls in place to reduce the chances of manipulation of books of accounts by the executive and
employees working under them. The study also held that having more female directors in the AC is associated with improved
stewardship over the resources of the firm. Female directors are more conservative hence they tend to encourage the adoption
of conservative accounting concepts and principles that are associated with tight controls over resources hence reduced earning
management opportunities. The study also noted that having more AC members with expertise in finance, accounting, and
auditing was advantageous to the committee as they could review audit reports and various internal controls in place at the firm
to identify weaknesses to make recommendations for their strengthening. The strengthened controls reduce the loopholes that
are exploited by executives and employees working under them in manipulating books of accounts. The study is also of the
view that having larger than necessary AC is associated with poor deliberations at the audit committee hence encouraging EM.
Finally, the researcher thought that having very old directors may lead to increased EM management by the executive as the
directors of advanced age may not manage to rigorously monitor internal controls as well as books of accounts.
6.2 Recommendations
Given the inverse effect of AC impendence on EM, the study recommends that listed manufacturing firms ensure that their
AC have more independent non-executive directors to help encourage objective review of internal controls and financial
reports with a view of presenting fair and true financial status of the firm. The Nairobi Securities Exchange should also ensure
that listed manufacturing firms have well-constituted audit committees in terms of independence to minimise EM. Based on
the inverse relationship between gender diversity and EM, the study suggests that listed manufacturing firms incorporate more
female directors in their AC to minimise EM. Female directors tend to be conservative as regards to application of accounting
principles and practises hence lowering the chances of manipulation of the accounting information system of the firms. The
NSE should also closely monitor the election of the board of directors to ensure more female directors are incorporated by
various firms. The study also revealed an inverse effect of AC expertise on EM therefore necessitating the need for listed
manufacturing firms to ensure that each AC has an expert in finance accounting or auditing or a combination of any or all.
Audit committees having relevant expertise are critical in minimising the chances of manipulation of books of accounts by the
executive and employees working under them. The NSE should also ensure that listed firms have relevant experts on their
boards to minimise EM and ensure the stability of listed manufacturing firms. Further, the positive effect of Audit committee
size on EM makes it critical that listed manufacturing firms have the right number of AC members. Having too many members
Journal of Economics, Finance and Business Analytics 2024; 2(1): 34-44 43
in the AC may not necessarily translate to a reduction in EM. Optimal AC sizes may help improve the quality of deliberations
at the committees and policy recommendations thereof. The NSE aught also ensures that committees of the whole board for
listed firms are of the right size to improve the quality of corporate governance among firms.
Finally, given the positive effect of an average of the AC members on EM, the study recommends that listed manufacturing
firms not elect directors who are advanced in age to be in charge of the committee. The committee needs a blend of younger
and relatively experienced directors who are energetic and can rigorously monitor the accounting information system of the
firms and internal controls. The NSE should also consider setting age limits for members servicing critical committees of the
board.
6.3 Limitations
The study was limited to the 9 listed manufacturing firms at the NSE hence the findings are more relevant for adoption by
listed manufacturing firms. The findings may not be useful for non-manufacturing firms that possess different operating
environments. Non-listed manufacturing firms should apply the findings for policy purposes with caution. The study was also
limited to five aspects of audit committees including AC independence, gender diversity, expertise, size and age diversity.
Other aspects of AC composition such as audit AC nationality, and experience diversity were not within the scope of the study
hence the parameter estimates may differ when excluded elements of AC are added to the model. The parameter estimates
should thus be applied with caution by firms making decisions. The study also adopted discretionary accruals as the measure of
EM as suggested by Guo and Zhang, (2021). Even though widely used in various empirical studies, it may not capture all
aspects of EM management. The study was also limited to secondary data that may not capture all aspects of AC composition
and EM especially the qualitative aspects that need a composite of measures both quantitative and qualitative.
6.4 Areas for Further Studies
The breadth of the current study can be improved through various recommendations to future researchers. The current
study was limited to the 9 listed manufacturing firms in Kenya. The study recommends that future researchers extend the study
to non-manufacturing firms listed at the NSE. This will enhance the usefulness of the findings across listed firms regardless of
their operating environment. The current study was also limited to aspects of AC composition including AC independence,
gender diversity, expertise, size and age diversity. The study therefore recommends that future studies should study other
aspects of AC composition that were not within the scope of the current study such as AC nationality, and experience diversity
among others. This would help in generating parameter estimates that are more accurate and have added value to the current
study. The study adopted discretionary accruals as the measure of EM even though widely used in various empirical studies, it
may not capture all aspects of EM management. The study thus recommends that future researchers should adopt other proxies
of EM management to improve the dearth of empirical studies in EM. The study also recommends that future studies should
adopt primary tools of data collection for measuring AC composition and EM in addition to quantitative secondary data.
Conflicts of Interest
The authors declare no conflicts of interest.
References
[1] Abubakar, A. H., Usman, A., Anuforo, P. U., & Alhaji, B. Y. (2021). Audit Committee Attributes and Real Earnings Management in
Nigeria. Asian People Journal, 4(1), 84-92. https://doi.org/10.37231/apj.2021.4.1.254
[2] Al-Absy, M. S. M., Ismail, K. N. I. K., & Chandran, S. (2018). Accounting expertise in the audit committee and Earnings
Management. Business & Economic Horizons, 14(3), 45-56.
[3] Birgen, Y., & Bogonko, J. B. (2018). Effect of price interventions on the financial performance of Mumias Sugar Company.
International Academic Journal of Economics and Finance, 3(2), 292-308.
[4] Bosse, D. A., & Phillips, R. A. (2016). Agency theory and bounded self-interest. Academy of Management Review, 41(2), 276-
297. https://doi.org/10.5465/amr.2013.0420
[5] Bridoux, F., & Stoelhorst, J. W. (2014). Microfoundations for stakeholder theory: Managing stakeholders with heterogeneous motives.
Strategic Management Journal, 35(1), 107-125. https://doi.org/10.1002/smj.2089
[6] Chijoke-Mgbame, A. M., Boateng, A., & Mgbame, C. O. (2020, July). Board gender diversity, audit committee and financial
performance: evidence from Nigeria. In Accounting Forum, 44 (3) 262-286. https://doi.org/10.1080/01559982.2020.1766280
[7] Dakhlallh, M. M., Rashid, N., Abdullah, W. A. W., & Shehab, H. J. A. (2021). The moderate effect of Audit committee
independence on the board structure and real Earnings Management: Evidence from Jordan. Journal of Contemporary Issues in
Business and Government, 27(2), 123-133.
[8] Dalton, D. R., Hitt, M. A., Certo, S. T., & Dalton, C. M. (2007). The fundamental agency problem and its mitigation. Academy of
Management Annals, 1(1), 1-64. https://doi.org/10.1080/078559806
44 Okeyo & Nyamute (2024): Causal effect Relationship between Audit Committee Composition and Earnings Management
[9] Davis, J. H., Schoorman, F. D., & Donaldson, L. (2018). Toward a stewardship theory of management. In Business Ethics and
Strategy. Washington: Routledge.
[10] Donaldson, L., & Davis, J. H. (1991). Stewardship theory or agency theory: CEO governance and shareholder returns. Australian
Journal of Management, 16(1), 49-64.
[11] Eisenhardt, K. M. (1989). Agency theory: An assessment and review. Academy of Management Review, 14(1), 57-74.
[12] Freeman, R. E. (1999). Divergent stakeholder theory. Academy of Management Review, 24(2), 233 236.
[13] Ghaleb, B. A. A., Al-Duais, S. D., & Hashed, A. A. (2021). Audit committee chair’s legal expertise and real activities
manipulation: Empirical evidence from Malaysian energy and utilities sectors. International Journal of Energy Economics and
Policy, 11(1), 65-78.
[14] Gujarati, D. N., Porter, D. C., & Gunasekar, S. (2012). Basic econometrics. Tata McGraw-Hill Education.
[15] Guo, K., & Zhang, A. (2021). Effect of Fair Value Measurement on Earnings Management of Chinese Real Estate Listed Companies.
Proceedings of Business and Economic Studies, 4(1)-23-47
[16] Harris, O., Karl, J. B., & Lawrence, E. (2019). CEO Compensation and Earnings Management: Does gender matter? Journal of
Business Research, 9(8), 1-14.
[17] Hastuti, S., Setiawan, D., & Widagdo, A. K. (2020). Substitution between accrual and real earnings management: The role of
independent commissioners and Audit committee. Jurnal Keuangan dan Perbankan, 24(2), 225-240.
[18] Isa, M. A., & Musa, F. (2018). The moderating effect of Audit committee on the relationship between board diversity and EM of
banks in Nigeria. Iranian Journal of Accounting, Auditing and Finance, 2(1),11-35.
[19] Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behaviour, agency costs and ownership structure. Journal
of financial economics, 3(4), 305-360.
[20] Jordaan, L. A., De Klerk, M., & De Villiers, C. J. (2018). Corporate social responsibility and earnings management of South African
companies. South African Journal of Economic and Management Sciences, 21(1), 1-13.
[21] Juhmani, O. (2017). AC characteristics and Earnings Management: The case of Bahrain. International Journal of Accounting and
Financial Reporting, 7(1), 11-31.
[22] Kaczmarek, S., Kimino, S., & Pye, A. (2014). Interlocking directorships and firm performance in highly regulated sectors: the
moderating impact of board diversity. Journal of Management & Governance, 18(2), 347-372.
[23] Kapkiyai, C., Cheboi, J., & Komen, J. (2020). Audit Committee Effectiveness and Earnings Management among Publicly Listed
Firms in Kenya. SEISENSE Journal of Management, 3(2), 31-44.
[24] Kim, J., Kim, Y., & Zhou, J. (2017). Languages and Earnings Management. Journal of Accounting and Economics, 63(2-3), 288-306.
[25] Kjærland, F., Haugdal, A. T., Søndergaard, A., & Vågslid, A. (2020). Corporate governance and Earnings Management in a Nordic
perspective: evidence from the Oslo Stock Exchange. Journal of Risk and Financial Management, 13(11), 256.
[26] Kothari, C. R. (2004). Research methodology: Methods and techniques. New Delhi: New Age International.
[27] Mansell, S. F. (2013). Capitalism, corporations and the social contract: A critique of stakeholder theory. Cambridge University Press.
[28] Mardessi, S. M., & Fourati, Y. M. (2020). The impact of Audit committee on real earnings management: Evidence from the
Netherlands. Corporate Governance, 2 (1), 33-46.
[29] Mitnick, B. M. (1975). The theory of agency. Public Choice, 24(1), 27-42.
[30] Mollik, Abu T.; Mir, Monir; McIver, Ron; and Bepari, Md. Khokan (2020). Effects of Audit Quality and Audit Committee
Characteristics on EM during the Global Financial Crisis: Evidence from Australia, Australasian Accounting, Business and Finance
Journal, 14(4),85-115. doi:10.14453/aabfj.v14i4.6
[31] Moratis, L., & van Egmond, M. (2018). Concealing social responsibility? Investigating the relationship between CSR, Earnings
Management and the effect of industry through quantitative analysis. International Journal of Corporate Social Responsibility, 3(1), 1-
13.
[32] Moses, T. (2019). Board Characteristics, Audit Committee Composition, And Financial Reporting in Nigeria. International Journal
of Innovative Social Sciences & Humanities Research, 7(1), 37 44.
[33] Post, J. E., Preston, L. E., & Sauter-Sachs, S. (2002). Redefining the corporation: Stakeholder management and organizational
wealth. Stanford University Press.
[34] Raimo, N., Vitolla, F., Marrone, A., & Rubino, M. (2021). Do Audit committee attributes influence integrated reporting quality?
An agency theory viewpoint. Business Strategy and the Environment, 30(1), 522-534.
[35] Sanghani, D. A. (2014). The effect of liquidity on the financial performance of non-financial companies listed at the Nairobi Securities
Exchange (Unpublished, Masters Thesis), University of Nairobi.
[36] Setiawan, D., Phua, L.K., Chee, H.K. and Trinugroho, I. (2020). The effect of Audit committee characteristics on Earnings
Management: The case of Indonesia, Afro-Asian Journal Finance and Accounting, 10 (4), 447463.
[37] Siagian, D., & Siregar, S. V. (2018). The effect of Audit committee financial expertise and relative status on Earnings Management:
Case of Indonesia. Jurnal Akuntansi, 22(3), 321-336.
[38] Sudarman, W. A., & Hidayat, W. (2019). Audit committee and Earnings Management: the role of gender. Jurnal Akuntansi, 23(3),
379-392.
[39] Waweru, N. (2018). AC characteristics, board ethnic diversity and Earnings Management: evidence from Kenya and Tanzania.
International Journal of Corporate Governance, 9(2), 149 174.
[40] Wiseman, R. M., Cuevas‐Rodríguez, G., & Gomez‐Mejia, L. R. (2012). Towards a social theory of agency. Journal of
Management Studies, 49(1), 202-222.
[41] Wooldridge, J. M. (2013). Correlated random effects panel data models. IZA Summer School in Labor Economics (http://www. iza.
org/conference_files/SUMS_2013/viewProgram.
[42] Zalata, A. M., Tauringana, V., & Tingbani, I. (2018). Audit committee financial expertise, gender, and EM: Does the gender of the
financial expert matter? International review of financial analysis, 5(5),170-183