Audit Quality and Firm Performance: Evidence from Botswana and Uganda

Auditing has its roots in the private sector where there is a concern of fraud through theft and misappropriation of assets. The importance of auditing lies in its perceived role in detecting fraud, errors and irregularities in financial statements. Auditing bolsters confidence and creditability of the financial statements which is needed for improving performance as users rely on them to make investment decisions. The audit of financial statements is an essential tool in maintaining an efficient market environment by reducing information asymmetries. Fraudulent behaviour occurs significant problems of agency arise resulting from a weak governance system. Thus, fraudulent financial statements can be seen as a problem of information asymmetry (Magnanelli, Nasta, and Pirolo, 2017). Xin, Zhou, and Hu (2018) found that companies that engaged in fraud experienced performance deterioration. In a study of high-profile corporate failures, Soltani (2014) found that a lack of audit quality is one of the reasons for financial and corporate scandals. The findings of the study are supported by other studies (Umar, Erlina, and Fauziah, 2019; Magnanelli et. al, 2017). The recent scandals of corporate failures such as Patisserie Valerie in the UK (2018), Steinhoff and KPMG in South Africa (2018), Kingdom Bank Africa Limited and Choppies Limited in Botswana (2015, 2018 respectively); Crane Bank in Uganda (2018) demonstrates the need of increased scrutiny of financial statements.

In all the cases mentioned above, evidence of fraud occurred despite unqualified audit opinion expressed of the financial statements by auditors. Due to these and past corporate failures such as Enron, WorldCom and Tesco, there has been an increased focus on studies on audit quality. Following this path of research, this study analysed the effect of audit quality on the financial performance of listed companies from two developing countries, Botswana, and Uganda. This study will address an empirical gap of capital markets of sub-Saharan Africa. The world bank has recognised the need for the development of domestic capital markets in addressing developmental challenges and hence has brought to the fore, issues of investor protection. Therefore, this study also addresses the need for reform in strengthening investor protection and the regulatory framework for public offerings in the African capital markets. Audit quality and financial performance are at the core of these issues as audited financial statements are a tool for ensuring the safeguarding of shareholder's investment. The following two specific objectives are addressed: a. The impact of auditor size on audit quality and; b. The impact of audit fees on audit quality. The next section of this study is a literature review developing our hypothesis that addressed the main aim of this paper, followed by a methodology section explaining the variables that were analysed. The final two sections are the discussion of the results and the conclusion.

In this study, the agency theory (Jensen and Meckling, 1976) was applied as a theoretical framework. Agency theory encapsulates the problem of owner versus agent and has been used extensively in the finance and accounting literature. Specifically, it has been used to explain the relationship between external auditor performance and function (Adams, 1994). The theory postulates that problems arise when interests are misaligned and where informational asymmetry exists between the agent and the owner. The main contention is that agents will make potentially prejudicial and onerous decisions to shareholders in order to benefit themselves. This type of opportunistic behaviour can lead to poor financial performance. The information asymmetry that exists between principal and agent requires a redress in order to improve information about company performance. External audits act as a monitoring tool that reduces information asymmetry. Therefore, the greater the information asymmetry, the higher the demand for higher quality audits and vice versa (Farouk and Hassan, 2014; Gunn, Hallman, Li and Pittman, 2017).

The agency theory was especially useful for this study as both Botswana and Uganda follow the Anglo-American corporate governance model where corporate governance is based on the agency relationship (Fooladi and Shukor, 2012). Corporate governance mechanisms such as audit quality aim to align the interests of agents with shareholders, thereby increasing firm performance (Jensen and Meckling, 1976). Given the lack of infrastructural development in the capital markets of developing countries such as those in Africa (Capkun, Collins and Jeanjean, 2015), it is essential to assess how auditing affects firm performance. For example, developing countries tend to have weak enforcement of laws leading to inadequate investor protection and concentrated ownership (Roussow, 2005; Berglof and Von Thadden, 1999). Thus, investors have to seek assurance on the reliability of the financial statements elsewhere. The issues above demonstrate the importance of auditing as a corporate governance mechanism in minimising the agent shareholder conflict thereby giving assurance that firm value will be increased (Tahinakis and Samarinas, 2016; Popović, Tošković,Majstorović,BrkanlićandKatić,2015;FrancaandCorina,2014;TürkânUğurDâi, 2014).

Different stakeholders define audit quality in different ways based on their utility. For example, users of financial statements deem high-quality audit to be one that precludes significant inaccuracies in the financial statement. On the other hand, society and the audit firm may deem a high-quality audit as one where the firm can successfully withstand litigation. According to the Financial Reporting Council (FRC), there is no agreed-upon definition of audit quality that can be used as a standard against which performance can be assessed (FRC, 2006). In reiteration, the International Organisation of Securities Commission (IOSO, 2009) reported that due to differences in stakeholder perceptions, it is challenging to define audit quality. Because of this difficulty, academicians and regulators have sought to determine audit quality with a composite framework. There are broad areas identified that overlap between regulator and academic audit quality frameworks (Francis, 2011; IAASB, 2013; Knechel, Krishnan, Pevzner, Shefchik and Velury, 2013; FRC, 2008). Hu (2015) put forth a framework that combines the regulator and academic viewpoints. The article identified three key drivers of audit quality, including input, output and context of the audit and suggested measurements. This study will pursue only a few of the measurements from the framework. Justification for selection of the measurements is made hereunder.

We began with a discussion of ex-post audit quality measurements which are components of the output drivers identified in Hu's Framework. The first issues are restatement and litigation measures. An earlier study by Palmrose (1988) of audit firms indicated that the number of litigations is an indicator of the quality of the firm. Also, in evaluating audit quality, the number of restatements occurring in the financial statements can be used as an indicator that the financial statements were not presented accurately in the first place. A comprehensive data set over a 35-year period in America revealed an average of 28 lawsuits per annum, indicating a 0.28% annual audit failure rate. It was also found that the number of successful lawsuits was less than 50% of the filed lawsuits (Palmore, 2000). Given the small number of audit failures, it is difficult to infer that measurement as a proxy for audit quality (Francis, 2004). Addressing the matter of financial statement restatements, Francis (2004) found that the restatements surveyed over five years of American listed firms were simple, straightforward, retrospective estimation adjustments that were, in some cases, initiated by the auditors. In the rest of the cases, the restatements did not lead to any business failure suggesting that restatements are inadequate as an indicator for audit failure.

Having determined that audit quality inferred upon by audit failures are quite intermittent incidences, in this study, we turn to other drivers for determining audit quality. Audit size is a measurement encompassing all three drivers of audit quality, and so it is selected on that basis. Furthermore, we opted to consider audit fees as a proxy encompassing the context driver. Having selected the proxies as mentioned above, all three areas of audit quality drivers are addressed in this study.

The audit market has evolved into a dual market in which there a few large dominant firms and that are perceived to deliver high-quality audits due to reputational risk in case of audit failure. Inversely, there are smaller, less prominent audit firms whose audits are perceived as being of lesser quality (Sirois, Marmousez and Simunic, 2016; Knechel, Niemi and Zerni, 2013). The lesser quality of the smaller firms lends credence to the use of Big4 as a determinant of audit quality as also suggested by empirical studies (for example, Krishnan, Ma and Yan, 2016). The big 4 measure is especially useful for this study due to its effectiveness as a corporate governance mechanism in countries characterised by a weak legal environment such as those of developing countries (Khlif and Samaha, 2016). The Big4 measure for audit size is therefore adopted in this study. Regarding audit fees, they are measured following the standards published by the Public Company Accounting Oversight Board, which states that fees for professional services are necessary to perform an audit or review including services rendered for the audit of the company's annual financial statements.

Antecedents regarding auditor size point to a positive relationship with firm performance. Specifically, big 4 audit companies tend to improve firm performance compared with non-big 4 audit firms (Alzoubi, 2018; Garven and Taylor, 2015; Kouaib and Jarboui, 2014; Lin and Hwang, 2010; Vander Bauwhede, Willekens and Gaeremynck, 2003). The improved performance is because big 4 audit firms have more experience in auditing publicly listed companies, better quality human resources, and the ability to handle complex audits (Sayyar Basiruddin, Rasid and Elhabib, 2015; Francis and Yu, 2009;). In a study of Malaysian firms Jusoh, Ahmad and Omar, (2013), found that firms that were audited by big 4 audit firms had a positive and significant relationship with firm performance similar to studies by Farouk and Hassan (2014) of Nigerian firms; Bouaziz (2012) of Tunisian firms; Phan, Lai, Le and Tran (2020) of Hanoi stock exchange; Eshitemi and Omwenga (2017) of Nairobi parastatals; Mustafa and Muhammad (2018) of Nigerian listed oil and gas companies. However, some studies revealed a significantly negative relationship between auditor size and firm performance, such as Elewa and El-Hadded (2019) in a study of Egyptian companies; Aledwan, Yaseen and Alkubisi (2015) in Jordanian cement companies. Based on this literature, the following is posited:

H1= Auditor size is a significant determinant of firm performance

It is a consensus in the accounting literature that audit fees reflect efforts of auditors because the audit market is highly regulated, and the ability to earn rents is limited. In anticipation of more audit work, i.e. more extensive reviews and closer supervision of staff, audit firms charge a higher fee commensurate with the amount of work involved in the auditing process (Schelleman and Knechel, 2010; Krishnan, Sun, Wang and Yang, 2013). Therefore, audit fees are a signal to the market of the added credibility of the financial information, thereby increasing firm value. In this regard, there has been conflicting evidence. Eshlemen and Guo (2014) found that audit fees were indicative of more considerable efforts by the auditor. Similar findings by other authors such as Blankley, Hurtt and MacGregor (2013) and Asthana and Boon (2012) supported this. However, dissent exists where audit fees did not indicate more considerable auditing efforts (Lin, Lin and Chen, 2018; Krauß, Pronobis and Zülch, 2015; Choi, Kim and Zang, 2010) Donatella, Haraldsson and Tagesson (2019). These studies are consistent with a similar branch of literature that suggests that audit fees create an economic bond, a determinant of audit behaviour (Laitinen and Laitinen, 2018; Hoitash, Markelevich and Barragato, 2007). When this economic bond exists between auditors and client audit, fees are expected to have an inverse relationship with firm performance. Such is the case with studies by Mustafa and Muhammad, 2018; Sulong et al., 2013; Moutinho, Cerqueira and Brandao, 2012. However, Laitinen and Laitinen (2018) pointed out that the auditor-client economic bond was prevalent among non-Big 4 auditing firms, suggesting that protecting their reputations outweighed the value of any additional fees to be gained. In that regard, audit fees are expected to positively correlate with firms' performance, such as Sayyar et al. (2015).

Thus, the second hypothesis is:

H2= Audit fees are a significant determinant of firm performance

The basis for selection of the Botswana and Uganda for this study was first, their legacy of colonial inheritance. Both countries were former British colonies and were two of thirteen countries that developed a corporate governance code that borrows heavily from the British Anglo-American System. Thus, Botswana's and Uganda's corporate governance code is chiefly concerned with the principal-agent relationship and the associated rules and laws that regulate that relationship. The second basis for selection is in the comparison of the two countries as developmental states. Mbabazi and Taylor (2005) argue that developmental states are not defined by their economic performance but rather by their ideological underpinnings of development and whose resources are directed towards economic development. Botswana and Uganda are both considered prosperous countries in their own right, albeit with different outcomes. Both countries remain Ricardian economies, having achieved rapid post-colonial growth with no structural transformations of their economies. This success is attributable to developmental institutions whose purpose it was to foster tri- lateral partnerships amongst civil society, private companies and the state. This trio, constitutes prevailing contemporary governance of the countries, linking with the fundamental basis of selection. Botswana and Uganda are examples of two developmental African countries whose institutions (including financial) have led to growth. In summary, the selection of Botswana and Uganda as subjects of this study was justified and further supported by other studies (Sekitoleko, 2017; Ganamotse, Samuelsson, Abankwah, Anthony and Mphela, 2017; Onessimo, 2016; Kiiza, 2006; Mbabazi and Taylor, 2005)

Data was gathered from the financial statements of domestically listed companies from the Botswana Stock Exchange (BSE) and the Uganda Stock Exchange (USE) for the five years 2014-2018. The BSE had 26 listed companies by the end of December 2019, and Uganda had nine listed companies by the end of December 2019. Selection of company financial statements was based on two factors, firstly, that the company is registered during the five years under study and secondly, that the data is available for the companies. The final data set consisted of data from twenty-four companies, seventeen from Botswana and seven from Uganda. Table 1 depicts the demographics of the companies in thesample.

The model constructed was to determine the impact of audit quality on firm performance. As earlier alluded, audit quality was measured using auditor size and audit fees as proxies. Financial performance was measured using the return on assets (ROA) and Tobin's Q. These represented non-market and market measures of performance. ROA measures firm profitability as a proportion of net income to total firm assets, whereas TQ measures firm value as a proportion of market capitalisation of a firm to total firm assets. Additionally, the model uses four control variables, firm size, leverage and the book to market ratio. Table 2 below gives a summary of the variables and their measurement.

For this study, two models are used as below indicating the two dependent variables for measures of firm performance.

Model I

Model II

(1) Classic assumption test

Before running the regression analysis, the assumption of normality of data was tested. Table 3 depicts the descriptive statistics of the data. All the variables except Big4 had values between .000 and .783 and .312 and -1.014, which indicated no problems of skewness or kurtosis respectively. The normality of data was achieved after transforming the ROA, Qratio and BTM in line with studies of Jusoh, Ahmad, and Omar (2013) and Coakes, Steed, and Ong (2009). The other variables did not require transformation. The transformation of the data was a two-step process, firstly requiring ranking of the selected variable data set using a fractional rank and secondly, using the fractional rank to compute the normalised variable (Templeton, 2011). Due to abnormality of Big 4, it was dropped from the models and thus was not considered in further analysis.

After transformation, the histogram of standardised residuals indicated that the data contained approximately normally distributed errors, as did the normal P-P plot of standardised residuals, which showed points that were not completely on the line, but close.

An analysis of standard residuals was carried out, which showed that the data contained no outliers (Std. Residual Min = -2.895, Std. Residual Max = 2.124). Tests to see if the data met the assumption of collinearity indicated that multicollinearity was not a concern (LAuditFees, Tolerance = .165, VIF = 6.070; LEV, Tolerance = .792, VIF = 1.263; LTA, Tolerance=.151,VIF=6.643;NormBTM,Tolerance=.498,VIF=2.007).The datamet the assumption of independent errors (Durbin-Watson value = 1.084 and 1.036) for each of the two models

(2) Panel data regression analysis

Table 4 shows the correlation between the independent and control variables and the dependent variables NormROA and NormQratio. All correlations except between LEV and NormBTM are significant at the .01 level. LAuditFees shows a significant and negative correlation between the two dependent variables (-.781 and -.545 between NormQratio and NormROA, respectively). The other variables show correlations between .839 and -.844. Of the control variables, NormBTM has the only positive correlation with the dependent variables.

A multiple regression was carried out to assess if audit fees (LAuditFees), firms size (LTA), BTM and leverage (LEV)are predictors of financial performance i.e. NormROA and NormQratio. Table 5 depicts the summary of the model regression analysis. It was found that LAuditFees, NormBTM, LEV, LTA explain a significant amount of the variance in the value of ROA (F (4, 113) = 40.783, p < .05, R2 = .591, R2Adjusted = .576). Results also revealed that LAuditFees, NormBTM, LEV, LTA explain a significant amount of the variance in the value of NormQratio (F (4, 114) = 390.264, p < .05, R2 = .932, R2Adjusted =.930).

A further analysis of the individual level predictors shows that LAuditFees did not significantly predict NormROA or NormQratio (β = .120, p = .420; β = .061, p = .314). However, all other variables, i.e. the control variables revealed to be significant predictors of performance for both NormROA and NormQratio.

Based on the analysis of the data, H1, which stated that Auditor size is a significant determinant of firm performance can neither be confirmed nor rejected due to the data set violating assumptions of further parametric analysis. However, based on the results of the regression analysis, this study fails to reject the null hypothesis for H2, which stated that Audit fees are a significant determinant of firm performance.

The results of this study mimic those of Elewa and El-Hadded (2019) and Tanko and Polycarp (2019) that also found an insignificant effect of audit quality on performance. Tanko and Polycarp (2019) attributed the insignificance of the audit quality to political connectedness of companies. The study found that companies that were politically connected performed better than those that were not and the quality of the audit did not matter.

This study sought to determine the impact of audit quality on the financial performance of companies listed on the Botswana and Uganda stock markets. Regression analysis was conducted on a five-year panel data for 27 companies in total. In order to measure audit quality, auditor size and audit fees were used as proxies while financial performance was proxied by Return on Assets and Tobin's Q. Due to non-conformity of the auditor size data to tests of normality, the variable was dropped from the analysis, and therefore this study could not test the first hypothesis. The second hypothesis was rejected as the results revealed that Audit fees were statistically non-significant as a predictor for measures of both the non-market and market measures of performance.

Theoretical contributions of this study are with respect to empirical evidence from the stock markets of two African countries, precisely, evidence on the relationship between audit quality and firm performance. It was found that contrary to the overwhelming empirical studies on the matter, audit quality is not a significant predictor of financial performance and thus ineffective as a corporate governance mechanism. The findings of this study suggest that investors cannot rely on the performance of audits as an assurance mechanism. Policymakers should ensure that there is the enforcement of existing laws in cases were an auditing firm issues an unqualified audit opinion for companies that later collapse. Additionally, more emphasis can be placed on external mechanisms of governance, such as board composition and ownership structure.

The main limitation of this study is due to its sample size; two countries with small capital markets were sampled, which resulted in the analysis of 27 companies with 120 data points. Though this did not undermine the veracity of the findings; still, future studies can sample larger data sets from different or additional Sub-Saharan capital markets.