The Effect of Ownership Structure and Audit Quality on Firm Performance

The study examined the effect of ownership structure and audit quality on firm performance of listed companies in Ghana. The research employed a quantitative research approach; secondary data was extracted from various annual reports and financial statements of the selected companies. The target population was all 42 listed companies on the Ghana Stock Exchange. The sample size was 20 companies selected from all industries. The study period was 2013-2018 resulted in 160 firm-yearly empirical observations. The study used return on asset (ROA) and return on equity (ROE) as the performance measure. Ownership structure was measured using managerial ownership and institutional ownership, audit quality was also measured with the auditor’s reputation, audit committee size and audit committee independence. The control variables used were board size and firm size. The researcher found a weak positive correlation between institutional and managerial ownership and firm performance. Moreover, there was a positive effect of audit quality on firm performance. It implies that the engagement of the services of the Big 4 audit firms has an incremental effect on firm performance. Audit committee size posited a positive effect on firm performance whereas audit committee independence was seen to harm firm performance. Similarly, board independence showed a positive effect on ROE and a negative effect on ROA. Board size, however, indicated a positive effect on firm performance. The researcher recommended the pressing need of diversifying sharehold-ings in firms as a sweetener to attract more skills and expertise among shareholders that can be tapped to enhance the performance of firms. However, managers should be protected from unnecessary shareholding meddling.


Introduction
The ownership structure of every business or company is viewed as a structured and productive solution to the sharing of risk and incentive problems (Thomsen and Pedersen, 2000). Singh and Davidson (2003) opined that the firm ownership framework is/are considered among the key internal mechanisms and structures of corporate governance and corporate finance. Scholars such as Bruton et al., (2010), Douma et al., (2006) have examined the diverse forms of ownership structure in different states and countries particularly on managerial, foreign, and domestic, institutional, and individual shareholders. An auditor is task with the responsibility of averting, disclosing, and reporting of unlawful acts and fraud in an organization (Oluwagbemiga, 2010). This debate has heightened by the collapse of both large and small corporations worldwide. The independence of auditors is to ensure the enhancement of the financial reporting quality by increasing the effectiveness and efficiency of the process of auditing and making sure an auditor is not too acquainted with the client to not imperil their integrity thus diminishing the opinion of their independence (Tobi et al, 2016). Users usually have confidence in the audited financial statements when auditors are independent. The exploration resulted that, varied forms of ownership may positively or negatively impact the financial performance of a firm when all things are being equal. Almajali et al., (2012) posits that ownership structure has a linear effect on the stockholders and investment in a firm that however is a factor of a firm's financial performance. Matoke and Omwenga (2016), indicated that there is a positive and insignificant effect of audit quality on financial performance and the greater the auditor's independence, the higher the propensity of the firm making substantial net profit margins. Elewa (2019) postulated that audit quality has an insignificant impact on return on asset and return on equity of firms. Heugens et al., (2009) and Grosfeld (2006) conceptualize that ownership structure can be defined from two perspectives notably, ownership concentration and ownership identity. According to Jiang et al., (2011) and Pedersen and Thomsen, (1999) ownership concentration specify shares of the biggest owner influenced by monitoring cost and absolute risk i.e. the size of the firm (larger size firms attract handsome capital funds and huge value-to-sale rates) factored by the Herfindahl index, which, indicates the percentage shareholding. Kim (2011) stated in his report that, foreign ownership is characterized by the number of shares controlled by all foreign investors that is been cut down by the total number of shares outstanding for that firm at the end of an accounting period expressed as a percentage.

Literature Review
Ongore (2011) examines the relation between a firm's ownership structure and its performance. His study focuses on 42 out of 54 firms that are listed in Kenya. His research categorizes ownership structure into state ownership, foreign ownership, diffuse ownership, corporate ownership, insider ownership and ownership concentration. The study considers return on equity, return on assets and dividend yield as indicators for measuring the performance of a firm. The research employs the Pearson's correlation and logit regression to analyze how the different variables relate. The results of the analysis show that ownership concentration and state ownership are negatively linked to the performance of a firm. His research shows that government-owned business typically does poorly because they are characterized by tribalism, unnecessary bureaucracy, nepotism, and favoritism, no respect for rules and regulations and influence from the politicians. The research demonstrates once again that when the firm's ownership is left in a few hands of around five or less major shareholders, there is the likelihood of the shareholders exerting much pressure on management in terms of monitoring. This therefore hinders the willingness of executives to be imaginative and innovative and this usually results in the firm performing poorly. However, corporate ownership, insider ownership and foreign ownership are all positively related to performance of a company. The research therefore implies that companies should go in for diffuse ownership rather than concentrated ownership since that is a way of reducing the pressure exerted by the major shareholders on managers and also to attract new shareholders with diverse skills and knowledge. He also suggests that firms should also be interested in insider ownership because managers become much more committed to the firm if they own shares in the business. Again, he recommends that the government should advocate for partnership with private individuals and institutions to own a part of the state-owned companies as this will help boost the performance of those companies. Tsegba and Ezi-herbert (2011) analyze the association a firm's performance and how its ownership is structured of Nigerian listed companies. The study focuses on 73 listed firms during the years of 2001 to 2007. Their research considers earnings per share and share price as performance indicators and categorizes ownership structure into insider ownership, concentrated ownership, dominant ownership and foreign ownership. They employ ordinary least squares to examine the association between the different variables. The results of their findings prove that there is no substantial correlation between concentrated ownership, foreign ownership, dominant ownership structures and the overall performance of a company. They in addition monitor that insider ownership is negatively associated with the overall performance of a firm. Their research suggests that corporate administrative bodies should reconsider foreign ownership, concentrated ownership, and dominant ownership structures as governance mechanisms since they do not have any significant relationship on the overall performance of a company. They again suggest that there should be monitoring by shareholders on their firms' management activities in the situation where the firm is characterized by insider ownership as the performance of companies is negatively affected by insider ownership.
Al-matari, Al-matari and Saif (2017) also analyze the association that exists between the performance of a company and how its ownership is being structured. Their study considers 81 organizations recorded on the Omani stock trade for the time of 2012 to 2014. Their study uses secondary data by extracting information from the yearly reports of those organizations. The paper categorizes ownership structure into foreign, governmental and institutional ownership and also uses return on assets to estimate the performance of a company. The study uses regression analysis to check the connection between the ownership structure and the overall performance of the firm. The findings of the research show a significant positive connection between governmental ownership and foreign ownership structures and the general performance of companies. Likewise, the study reveals that there is a positive connection between the general performance of a firm and institutional ownership structure but it was not substantial.

Sample and Data
The population of the study encompassed of all listed companies on the Ghana Stock Exchange. The target population is focused on listed firms due to data accessibility as they are all public companies and are authorized by regulations to publish their annual reports for accessibility by the general public as well as its members. Purposive sampling method was employed in the study for the selection of the companies' base on the following outlined conditions: (1) Comprised firms with complete annual report and was available to the researcher.
(2) Companies with full annual report but not for the period of study were accepted.
(3) Companies without information on any of the variables for a particular year was excluded despite having their annual reports published. Based on the above conditions, 20 companies were selected for the study. Information on the research variables were obtained from secondary data sources, in particular the analyzed firms' annual reports from the 2012-2018 periods. Which resulted in 160 firm-yearly empirical observations?

Dependent variable
The dependent variable as indicated above is firm performance.

Independent Variable
The independent variables will be ownership structure, audit quality and audit committee. Ownership structure was proxied by institutional and managerial ownership. Audit quality was also measured by external auditor reputation and audit committee by the size and independence of audit committee of various companies.

Control Variables
In testing whether the effect is caused by other variables, control variables are used to in testing this effect. Control variables warrant that the model is not subject to omitted variable bias. Firm size and board size were employed as the control variables.

Data Analysis
The data was employed in the analysis to critically estimate the effect of audit quality and audit committee on firm performance. Sayrs (1998) cited in Afriyie and Akotey (2015), explained panel data as a longitudinal data which encapsulates the observation of a given sample economic variables over a given period of time. Furthermore, (Hsiao,2003) emphasized that panel data is more effective in examining variables econometrically. He further stated that, panel data minimizes the extent of collinearity among the explanatory variables. Several researchers have employed similar method in evaluating the effect of certain economic variables on firms over time (Baldavoo and Nomlala, 2019; Akotey, Sackey, Amoah and Manso, 2013; Afriyie and Akotey, 2015). Therefore, this method is appropriate for the study. Where Yit represents the dependent variable of listed firm i at time t, α denotes the constant term. β1 and β2 represent the coefficients of the independent variables and control variables respectively, X1it, and X2it denotes the independent and control variable of firms i at time t and µit represent the stochastic disturbance term.

Model Specification
From equation (a) above, the following equations as models for my estimation to assess the effect of audit quality and audit committee on firm performance of listed firms on the Ghana Stock Exchange. Disturbance term also known as the residual has been fused to cater for any error that have the potency to affects the estimates as a result of unobservable events, human error, noisy financial data and also the need to be parsimonious.
According to (Eisenhart, 1947;Kreft and de Leeuw, 1998;Gelman, 2005;Akotey et al, 2013), there are two main models of panel data analysis. These are fixed effect and random effect model.
However, (Akotey et al, 2013), postulated that these two models are independent of each other. Fixed effect model is suitable when it is assumed that all time invariant events that might affect the independent variable are constant. On the other hand, when time invariant variables are interrelated at random, random effect model is used (Gelman, 2010). In order to decide which of the models is appropriate, the Durbin-Wu-Hausman (DWH) specification test is conducted (Hausman, 1978). To facilitate the test, two hypotheses are set: the null hypothesis (H0) and alternative hypothesis (H1).

H0 = Random model is appropriate H1 = Fixed model is appropriate
The thumb rule for the test state that, if the chisquare statistic (p-value) is greater than 5% significant level, the null hypothesis (H0) is accepted, implying that the random effect estimator was appropriate for modelling the data. On the other hand, if the chisquare statistic (p-value) is less than 5% significant level, the alternative hypothesis (H1) is accepted, meaning fixed effect model was used (Greene, 2008). The statistical software used in deriving the estimators of my analysis was (STATA 15.0).  Inferring from table 3 above, it can be observed that over the 6-year period, the maximum ROA is 0.9164096 representing 91.64096% after tax income to total assets and a minimum value of 32.81 losses in profit. Companies on the average earned a return of 4.62818% on their total assets employed during the period of consideration, and a standard deviation of 13.39% which indicate a marginal difference of best performing companies and non-performing ones. The maximum ROE is 1.879938 amounting to 187.9938% profit to the overall equity and a minimum value of 132.0982% loss. On the average, the companies earned a total return of 18.39% for shareholders on their investment. Nevertheless, the standard deviation of 37.67% significantly indicates that, there is a notable difference between performing and non-performing companies.

Descriptive Analysis
The maximum institutional ownership is 82% and a minimum of 1%. The average institutional ownership for the companies during the period of consideration is 41.96%, and a standard deviation of 24.68%. Averagely, the managerial ownership of the companies for the year of consideration is 29.28%, with a standard deviation of 18.16%. The maximum managerial ownership is 62% and lowest ownership of 1%.
The average audit quality is 0.9159664 representing 91.60% that means that more than half of the companies were audited by the big 4 auditing firms. In addition, the audit committee size is averaged with 4 members with a maximum and minimum member of 6 and 2 respectively. The average audit committee independence is 4 members. This means that on the average the proportion of external directors on the audit committee is 4 members for the companies during the period of consideration, with a maximum member of 9 and minimum member of 1. minimum values of 1billion and 120billion cedis respectively. With a standard deviation 34.79billion cedis suggests that there is a smaller difference among firms with respect to their

Correlation Analysis
Correlation basically measures the inter relationship among two or more variables. It determines the linear dependency among variables and the mutual relationship between variables. However, correlation associations do not mean causality i.e. if x affect y or vice versa.  Table 4.2 outlines the correlation analysis between ROA and ROE on the selected independent variables (institutional ownership, managerial ownership, audit quality, audit committee size and audit committee independence) and the control variables (board size and firm size). From the results, ROA is negatively and significantly related with ACIND (r = -0.22; p < 0.05) and BSIZE (r = -0.19; p < 0.05). However, ROA is positively and correlated with INSOWN (r = 0.07), MANOWN (r = 0.02), AUDQ (r = 0.03) and FSIZE (r = 0.10). On the other hand, ROA posits a negative correlation with ACSIZE (r = -0.1).
ROE is negatively and significantly correlated with INSOWN (r = -0.11; p < 0.05). Also, there is negative correlation between ROE and MANOWN. ROE is positively related with AUDQ, ACSIZE, ACIND, BSIZE, FSIZE with a correlation coefficient of (r = 0.06; r = 0.13; r = 0.10; r = 0.12; r = 0.20) respectively. Similarly, there exists a significant correlation among the independent variables and control variables. Also, the highest correlation coefficient of 0.75 is recorded between ACSIZE and ACIND, indicating that there is no substantial concern of multicollinearity among independent variables.  From the table above, null hypothesis (H0) cannot be rejected, implying that fixed effect model is not appropriate. This is simply because the prob > chi2 (0.9524) is greater than 5% significant level. Therefore, random effect model was used for the estimate.

ROA as dependent variable
The estimated model used for the study is indicated below:  Using random effect regression model with ROA as independent variable, the results are presented in the table below. The overall R-Square which indicates the variation of the ROA explained by the independent variables is 0.07. The lower R-Square value can be explained by the type of data set we used, panel data, which encompass a cross section of firms over a period of time. The p-value (0.3765) also indicates that all the variables are significant in the explaining ROA jointly.
From the estimates, it can be deduced that there is a positive effect of MANOWN, AUDQ, ACSIZE and FSIZE on ROA. This means that a marginal change in ROA will be predicted by a rise in the co-efficient of determination of the variables. This is represented as follows accordingly; 0.0004623, 0.045889, 0.0078058 and 0.0001075.
On the other hand, INNOWN, ACIND and BSIZE posit an inverse effect on ROA. This implies that, a marginal change in ROA can be explained by a fall in the variables. This is evidenced by the negative co-efficient of determination of the variables as indicated below; -0.0003119,-0.0015926 and -0.0169355. The p-value and the t-statistics of the individual variables which measures how significant each independent variable affect the ROA statistically all indicates that INSOWN, MANOWN, AUDQ, ACSIZE, ACIND, BSIZE, FSIZE are all not statistically significant individually in affecting ROA. This is because the p-values are more than all the confidence interval (1%, 5% and 10%) and the t-statistics are also less than 2 as the thumb rule indicates.

ROE as the dependent variable
The estimated model used for the study is indicated below:  097. This is because the p-values are more than all the confidence interval (1%, 5% and 10%) and the t-statistics are also less than 2 as the thumb rule indicates.

Discussion of findings
The study examined the effect of ownership structure and audit quality of firm performance in Ghana. From the results above, it can be inferred that there exist a weak positive correlation between institutional and managerial ownership and firm performance, which is supported by previous studies which establish a positive relationship between the variables in developed countries like ( Also, Audit quality, Audit committee size, Board size and firm size can predict ROE positively. However, institutional ownership, Managerial ownership, Audit quality, Audit committee size, Audit committee independence, Board size and Firm size negatively affect ROA. In addition, Institutional ownership, Managerial ownership and Audit committee independence negatively affected ROE. The p-value and the t-statistics of the individually variables which determine how important each independent variables affect ROA statistically all indicates that managerial ownership, institutional ownership, Audit committee size, Audit quality, Board size, Audit committee independence and Firm size are all not statistically significant individually in affecting ROA, for the reason being that the P-values are more than all the confidence level interval (1%, 5% and 10%) and the t-statistics are also less than 2 as the thumb rule establish.
According to outcome above, the null hypothesis will be accepted, implying that the fixed effect model is not appropriate. This is simply because the prob> chi2 (0.9524) is greater than 5% significant level. Hence, the random effect model was utilized for the estimate.
Added to above outcome, the exist moderating effect of audit quality and managerial ownershipfirm performance relationship. This might be attributed to the fact that more than 50% of the total firms sampled utilized the services of big 4 to review their annual report which ensured investors' confidence on the report and will willingly invest their resources with ease and without concern of future investment risks.

Conclusions
The conclusions that can be drawn from the findings in Ghana are that managerial ownership can offer direct economic motivation which will improve active monitoring by managers engagement. In addition to this, managerial ownership can align ownership and control via considerable shareholding ownership. Additionally, firms should increase the level of institutional ownership which will revamp their purchasing of audit services to ensure quality audit. Firms should notice that quality of audit rallies the quality of financial reporting, agency cost arising from managerial manipulation as a result of their self-centeredness motive are reduced and lessening in information asymmetry.