What Drives Corporate Governance Quality in Emerging African Economies? Evidence from Ghana

What Drives Corporate Governance Quality in Emerging African Economies? Evidence from Ghana Andrews Owusu* Coventry University, UK ABSTRACT This paper investigates the determinants of good corporate governance in Ghanaian listed firms for the study period 2000 to 2009. We develop a corporate governance index and its six sub-indices to measure firm-level governance quality. Annual reports of the Ghanaian listed firms are scored according to their governance disclosure during the whole period, 2000-2009 and pre 2003 and post 2003 introduction of the Ghanaian Code. A panel data analytical framework was used to find the determinants of strong corporate governance in Ghanaian listed firms. The empirical results over the whole period, 2000-2009 show statistically significant and positive relationship between external financing needs, firm size, institutional shareholdings and governance quality measured by the Ghanaian corporate governance index. While we find no relationship between external financing needs, director shareholding, firm performance measures and governance quality pre 2003, the result is positive and statistically significant for post 2003. The most consistent relationship we find concerns firm size and institutional shareholdings. However, growth opportunities appear to have no association with strong governance in Ghanaian firms. JEL Classification: G34, M40, M41, M48, Keywords: Firm Attributes, Firm-Level Governance Quality, Governance Index, Ghana Corresponding Author s Email Address: a.owusu@coventry.ac.uk INTRODUCTION This paper investigates the determinants of good corporate governance in the Ghanaian listed firms for the study period 2000 to 2009. Even though corporate governance has become increasingly important in realigning the shareholder and manager interest resulting from the separation of ownership and control (Berle and Means, 1932), relatively little is known about how corporate governance has evolved and the determinants of firm-level governance quality in emerging African economy like Ghana. As in Adda and Hinson (2006), there have been some inconsistencies and weaknesses in the regulation of firms in Ghana, suggesting that the level of legal protection offered to shareholders to prevent managers from expropriation of their wealth is relatively weak. Given that firm-level governance quality matter more in countries with weak legal systems (Metrick and Ishii, 2002; Klapper and Love, 2004), the implementation of good corporate governance can be seen as equilibrium response to the weak legal environment in which the Ghanaian firms operate. Fundamental to the implementation of good corporate governance is the agency theory which has identified some governance mechanisms to realign the shareholder and manager interests in anticipation of maximising shareholder wealth. Consequently, the Security and Exchange Commission Ghana (SECG) has become very anxious about governance issues and in 2003 introduced corporate governance guidelines on best practices (hereafter the Ghanaian Code) with which all firms were encouraged to comply. This was in line with the implementation of codes in many countries globally, for example, UK, South Africa, US, among others. The Ghanaian Code is also reflected the importance attached to corporate governance by the international organisations such as the Organisation for Economic Corporate and Development (OECD) and the Commonwealth Association for Corporate Governance (CACG). However, there is no study to date in Ghana that has analysed the governance quality and its determinants in Ghanaian firms. This study fills the above gap in the literature. Ghana is a unique context to investigate what drives firm-level governance quality because institutions and governance structures are considered weak relatively to the developed market counterparts (Gurgler et al, 2003). Given the international corporate governance debate which revolves around whether governance codes should be mandatory or voluntary (Aguilera and Cuervo- Cazurra, 2009), the introduction of the Ghanaian Code in 2003 provides a testing ground to investigate whether governance quality changes from pre 2003 to post 2003 might have been impacted by the changes in firm attributes for the first time. This paper contributes to the corporate governance discussion in the following ways. First, the paper provides the first direct evidence of contextual changes of governance practices across Ghanaian listed firms 57

from 2000 to 2009 by developing a comprehensive corporate governance index as a proxy to firm-level governance quality. Second, and apart from prior cross countries studies (Klapper and Love, 2004; Durnev and Kim, 2005), relatively few studies that have attempted to examine the determinants of firm-level governance quality in specific countries have focused on the Canadian (Anand et al, 2006), the US (Khanchel, 2007) and the Brazilian (Da Silveira et al, 2007) listed firms. This paper presents the first direct evidence across the Ghanaian listed firms attributes that are related to their governance quality. Third, the paper provides the first direct evidence of the determinants of firm-level governance quality that focuses on the pre and post introduction of the Ghanaian Code in 2003. This is particularly important because it helps to find whether the Ghanaian firms governance quality before (2000-2002) and after (2004-2009) the introduction of the Ghanaian Code can be explained by their attributes during the sub-periods. Anand et al (2006), Khanchel (2007) and Da Silveira et al (2007) do not address this problem. While we find no relationship between external financing needs, director shareholding, firm performance measures and governance quality pre 2003, the results is positive and statistically significant for post 2003. The most consistent relationship we find concerns firm size and institutional shareholdings. Our results over the whole period, 2000-2009 show that external financing needs, firm size and institutional shareholdings explain some of the changes in firm-level governance quality measured by the Ghanaian corporate governance index. The rest of the paper is structured as follows. Section 2 discusses the literature in relation to the Ghanaian corporate governance framework and the determinants of governance quality at the firm level. Section 3 focuses on the hypothesis development. Section 4 describes the selection of sample and research design. Section 5 presents the empirical results, while section 6 concludes the paper. BACKGROUND AND FIRM LEVEL GOVERNANCE QUALITY In Ghana, the recognition of corporate governance dates back to July 1963, when the Companies Code was enacted to govern the formation and operation of Ghanaian firms. Its provisions are largely based on the English Common Law, and notably, similar to the UK Companies Act 1948 but makes room for additional regulation of firms subject to special regulation (Adda and Hinson, 2006). The Ghanaian Code introduced in 2003 was the first attempt to make official corporate governance guidelines on best practices not backed by the force of law of which the Ghanaian firms were encouraged to comply with its provisions. In this respect, firms listed on the Ghana Stock Exchange (GSE) are expected to comply with the recommended governance provisions by the code in order to promote good corporate governance. Some of the key provisions in the code are separating the posts of the Chief Executive Officer (CEO) and the Chairman of the board of directors, limiting the board size to between eight and 16 directors, having at least one-third of the total membership of the board to be independent (outside) directors and at any event not less than two directors, appointment of a finance director to be responsible for the finance function, and the establishment of separate board committees to be responsible for audit and executive remuneration. Other provisions call for improvement of the relationship between shareholders and managers as well as provisions on financial affairs, auditing and disclosure practices. However, the Ghanaian Code failed to recommend for the establishment of a nomination committee as experienced globally. Unlike the UK and South Africa who relied heavily on the formation of committees to promote the highest standard of corporate governance in their various jurisdictions, the Ghanaian Code was introduced by SECG in 2003, with the principles applying to all corporate bodies approved or licensed as stock exchanges, dealers and investment advisers. In particular, the Code charged companies to adapt to their specific circumstances provided the spirit of the principles underlying the practices is maintained. This is similar to what Cadbury Report (1992) and King Report I (1994) first recommended in the UK and South Africa. Consistent with the UK and the South African approach to corporate governance associated with comply or explain philosophy, the Ghanaian Code also mandated companies to disclose their compliance with the code or provide explanation if any in their annual report. Although several studies have been carried out in Ghana recently to investigate the importance of firmlevel governance mechanisms to corporate performance (Kyereboah-Coleman and Biekpe 2006a; 2006b; Abor and Biekpe, 2007; Kyereboah-Coleman and Amidu, 2008; Kyereboah-Coleman and Osei, 2008; Isshaq et al, 2009), it is still not clear what drives these firms to improve their governance quality. Unlike Tsamenyi et al (2007) and Bokpin and Isshaq (2009) who focused on disclosure and transparency index, the firm-level governance quality in this study cut across a set of governance mechanisms recommended by the Ghanaian Code. This will then be developed into a Ghanaian corporate governance index (GCGI) to represent a firm-level governance quality before and after the introduction of the Ghanaian Code for our empirical analysis. 58

Firm-Level Governance Quality Proceedings of the Australia-Middle East Conference on Business and Social Sciences 2016, Dubai We directly measure the firm-level governance quality by developing a GCGI. To date, two main approaches of developing a corporate governance index have been employed to measure the firm-level governance quality in the existing literature. First, and based on the rating agencies 1 ranking methodologies, several North American researchers (e. g. Gompers et al, 2003; Klapper and Love, 2004; Durnev and Kim, 2005; Brown and Caylor, 2006; Erthugul and Hedge, 2009) have relied on these agencies database for the development of a corporate governance index as a direct measure of firm-level governance quality. However, the real practical problem with the governance data from these rating agencies is that no comparable dataset are available in many other countries and in particular Ghana. In addition, the subjectivity of the independent professionals involved in the rating of corporate governance practices may account for error and bias towards the ranking of the firms. The second approach is based on the researcher-developed governance index to measure firm-level governance quality from the publicly available source of corporate governance information (e.g. Khanchel, 2007; Garay and Gonzalez, 2008, Bassen et al, 2008; Price et al, 2011; Ntim et al, 2012; Abraham et al, 2015) or based on questionnaire survey by the researchers (e.g. Drobetz et al, 2004; Beiner et al, 2006). Similar to the rating agencies, this approach also has some limitations regarding the researcher being vulnerable to judgemental error and bias towards the development of a corporate governance index (Core, 2001). In this study, we employ the researcher-developed governance index to measure a firm-level governance quality because there is no corporate governance data readily available from any rating agencies in Ghana. In addition, the rating agencies methodologies are standardised in such a way that it might not be applicable to the Ghanaian situation. As in Garay and Gonzalez (2008) and Bassen et al (2008), all of the corporate governance information used to measure firm-level governance quality comes from the annual reports of the selected Ghanaian listed firms. Following Gompers et al (2003) and Garay and Gonzalez (2008), we developed the GCGI by assigning 1 point to each aspect of compliance with the Ghanaian Code provisions complied and disclosed in the firm s annual report or 0 otherwise (see Appendix 1 for the questions used to develop the GCGI). With this scoring method, and given the 36 binary objective questions, a firm s developed governance index in a particular financial year end can vary between 0 and 36, with 0 indicating perfect non-compliance and 36 indicating complete compliance. The corporate governance provisions included in the GCGI are solely based on the Ghanaian Code with six broad set of corporate governance best practices that Ghanaian listed firms are required to comply or provide explanation for non-compliance in their annual report. In this respect, the average of the six comprehensive sub-indices for each firm constitutes the developed GCGI. These sub-indices include: (1) board composition; (2) audit committee; (3) remuneration committee; (4) shareholder rights; (5) financial affairs and auditing; and (6) disclosure practices. Although, this paper is similar to that of Anand et al (2006), Khanchel (2007) and Da Silveira et al (2007), this is the first study that measures firm-level governance quality based on the provisions of code of best practices on corporate governance applicable to a particular country s business operational environment. There is significant literature that shows firm-level governance quality measured by a corporate governance index improve overtime, rather than weaken, particularly after the introduction of code of best practice on corporate governance. For example, significant improvements in firm-level governance quality over time have been found by Henry (2008), Price et al (2011), Ntim et al (2012) and Abraham et al (2015) in Australia, Mexico, South Africa and India, respectively. There is also evidence of unsatisfactory firm-level governance quality with sluggish increase in the overall corporate governance index in Brazil (Da Silveira et al, 2007) where the formal adoption of code of best practice on corporate governance was not considered. Finally, survey evidence such as Owusu and Weir (2013) found that 51% of the directors of the Ghanaian listed firms believed that the standard of corporate governance has improved after the introduction of the Ghanaian Code. Given the significant evidence that shows improvement in firm-level governance quality over time, it is important to investigate the potential determinants of firm-level governance quality in Ghana. HYPOTHESIS DEVELOPMENT Prior to developing our hypotheses, we examine how the firm-level governance quality based on the Ghanaian Code criteria has improved over time. In Figure 1, we plot the percentages of compliance with the Ghanaian Code from 2000 to 2009. The figure indicates that the Ghanaian listed firms have considerably improved their 1 The agencies are IRRC - Investor Responsibility Research Centre; CLSA - Credit Lyonnais Securities Asia; ISS - Institutional Shareholder Services; TCL - The Corporate Library; GMI Governance Metrics International; and S&P Standards & Poor s. 59

governance quality over time with the firms complying with 73% of the Ghanaian Code provisions in 2009 compared to 53% in 2000. This increase in compliance with the Ghanaian Code is in line with prior studies (Henry, 2008; Price et al, 2011; Ntim et al; 2012, Abraham et al, 2015) who reported a significant increase in compliance with their country specific codes of best practices among Australian, Mexican, South African and Indian listed firms, respectively. In contrast, Gomper et al (2003) using the US governance data reported a stable compliance with corporate governance among listed firms over a long period of time. The changes in Ghana provide a powerful setting to investigate why firms within the same contractual environment tend to voluntarily choose different corporate governance practices. FIGURE 1: PERCENTAGES OF COMPLIANCE WITH THE GHANAIAN CODE, 2000-2009 La Porta et al (1998) note that better investor protection increases investors readiness to provide financing which should be reflected in lower costs of capital. Klapper and Love (2004) build on this theory by demonstrating that the main goal of corporate governance is to improve investors confidence of receiving greater returns on their investment. Recent studies such as Khanchel (2007) and Da Silveira et al (2007) show a statistically significant and positive relationship between external financing needs and firm-level governance quality. This suggests that the Ghanaian firms with greater future external financing needs are expected to improve their corporate governance quality. Assuming that greater compliance with the Ghanaian Code indicates better firm-level governance quality, we expect firms with greater external financing needs to improve their firm-level governance quality leading to hypothesis 1. Hypothesis 1: Greater external financing needs should lead to better firm-level governance quality. Better growth opportunities do not necessarily indicate better firm-level governance quality. For example, Khanchel (2007) find growth opportunities to have no significant impact on firm-level governance quality. However, Da Silveira et al (2007) show significant positive relationship between growth and firm-level governance quality, suggesting that firms with better growth prospects will need to raise external funding for expansion and may deem it necessary to improve their corporate governance quality to lower costs of capital. The same could be true for Ghanaian firms, which is why we investigate operational improvements in addition to external financing needs by investigating sales growth impact on firm-level governance quality. Our expectation is that firms with growth opportunities are likely to improve their firm-level governance quality leading to hypothesis 2. Hypothesis 2: Better growth opportunities should lead to better firm-level governance quality. Klapper and Love (2004) note that the effect of firm size on corporate governance quality is ambiguous, such that larger firms may have greater agency costs because it is difficult to monitor their activities or due to the free cash flows argument by Jensen (1986), and therefore they need to adopt better corporate governance practices in order to mitigate the agency problems. In contrast, smaller firms may have greater growth opportunities, and therefore will need to raise more external financing. Thus, this can motivate them to improve their corporate governance quality in order to lower their costs of capital. Given that both arguments provide incentive for greater firm-level governance quality, we expect firm size to influence corporate governance quality leading to Hypothesis 3. 60

Hypothesis 3: Firm size should lead to greater firm-level governance quality Khanchel (2007) find performance to have no significant impact on firm-level governance quality, evidence supported by Da Silveira et al (2007). In contrast, Shabbir (2008) note that firms tend to behave in an opportunistic manner regarding the adoption of good corporate governance practices. The author shows that firms become more compliant when their prior period stock market performance declines and less compliant when there are improvements in their operating performance. This suggests that better firm performance affect corporate governance quality to decline as there will be no incentives for firms to improve their governance quality. It is therefore interesting to establish whether better performance is associated with greater corporate governance quality across Ghanaian firms. In Ghanaian setting, I expect firms with better performance to have a negative impact on their corporate governance quality leading to hypothesis 4. Hypothesis 4: Better performance should lead to lower firm-level governance quality Sheifer and Vishny (1986) note that institutional investors by virtue of their large stockholdings will have the incentives to monitor firms, which, if effective, will go a long way to improve corporate governance quality. Jensen and Meckling (1976) argue that the agency problems can be mitigated when managers have ownership interest in a particular firm. This suggests that managerial ownership helps to align managers and shareholders interests which could lead to improve firm-level governance quality. Durnev and Kim (2005) develop a theoretical model to analyse the potential determinants of corporate governance quality and find corroborating evidence that ownership concentration drives firms to adopt better corporate governance practices. In contrast, Anand et al (2006) find that the presence of majority shareholder or executive blockholder is negatively related with better corporate governance standards. Recently, Khanchel (2007) finds a positive and statistically significant association between directors and officers ownership or institutional ownership and corporate governance quality, evidence supported by Da Silveira et al (2007) who also find large blockholders to have a positive impact on better corporate governance practices. They however find family ownership to have a negative impact on corporate governance quality. In general, the relationship between various ownership structures and firm-level governance quality is ambiguous. However, and following the work of Khanchel (2007) and Da Silveira et al (2007), we expect institutional shareholdings or director shareholdings to have a positive impact on firm-level governance quality among Ghanaian firms leading to hypothesis 5. Hypothesis 5: Institutional shareholdings or director shareholdings should lead to greater firm-level governance quality SELECTION OF SAMPLE AND RESEARCH DESIGN Our sample of listed firms covers the period 2000 to 2009 in Ghana. We first downloaded the list of listed firms from GSE website. In the second stage, we verified whether all firms filed their annual reports over the sample period with the Ghana s stock exchange each year. As a result, the firms listed during each year under consideration were included in the sample based on the availability of their annual reports. Given our objective of investigating issues concerning firm attributes and corporate governance quality before and after the introduction of the Ghanaian Code in 2003, we attempt to collect data for each firm for three years to the introduction and six years after the introduction. This provides us a ten year window surrounding the introduction of the Ghanaian Code. As noted earlier, governance data was collected from each firm s annual report filed with the GSE from 2000 to 2009. All firm attributes data are from the 2005 and 2010 GSE Factbooks, which contain the summarised financial statement and ownership data. Our final sample of 35 out of the 38 listed firms as of December 2009 represents 92 percent of all listed firms currently traded on the GSE. As a result, we arrived at 283 firm year observation over the period 2000 to 2009. Table 2 provides a breakdown of the industrial composition of our sample. The major concentration of firms is in the finance industry (11 firms) followed by the manufacturing industry (9 firms), with the remaining firms being spread across 5 other industry classifications. For each individual analysis we use the total observations (283), and then group the sample into pre 2003 and post 2003 introduction of the Ghanaian Code. Given the introduction of the Ghanaian Code, 2003 was used as a dividing year and therefore excluded from the pre and post analyses, which means that the firm year observation varies based on the analysis of the sub-periods. 61

TABLE 1: BREAKDOWN OF THE GHANAIAN FIRMS BY INDUSTRIAL SECTORS Industrial Number of firms % of each sector Finance 11 31.4 Distribution 8 23.9 ICT 2 5.7 Manufacturing 9 25.7 Mining 2 5.7 Food & Beverages 2 5.7 Agriculture 1 2.8 Total 35 100 Similar to Anand et al (2006), Khanchel (2007) and Da Silveira et al (2007), we examine the relationship between firm attributes and governance quality. The measures of firm attributes we use are: external financing needs (FINANCE) defined as the ratio of total debt to capital employed, where capital employed is the sum of total debt and equity; growth opportunities (GROWTH) defined as the percentage of the difference between the current year s of sales and previous year s of sales divided by the previous year s of sales of each firm; firm size (SIZE) defined as the natural log of the book value of a firm s total assets; institutional shareholdings (INSTHOLD) defined as the proportion of shares held by institutional shareholders in excess of 3% of total shareholding and director shareholdings (DIRHOLD) defined as the proportion of shares held by board of directors to the total shareholdings. The measures of firm performance 2 we use are return on assets (ROA) defined as operating profit after tax divided by the book value of total assets; return on equity (ROE) defined as operating profit after tax divided by the book value of equity; and Tobin s Q (Q-ratio) defined as the market value of total assets divided by the book value of total assets, where the market value of total assets is measured by the market value of equity plus the book value of total assets minus the book value of equity (Gompers et al, 2003; Klapper and Love, 2004; Garay and Gonzalez, 2008). We also include two control variables, i.e. firm age (AGE) defined as the number of years since a particular firm s incorporation to the end of 2009 and year dummy variables. Given the panel nature of the data and the need to investigate the impact of the changing firm attributes on corporate governance quality over multiple-year timeframe, we employ panel data regression model as a method of estimation which provides a means of controlling for unobserved firm heterogeneity over the sample period. This is particularly important because we are interested in analysing the impact of firm attributes that varies overtime on corporate governance quality. To ensure consistent and efficient results, we use Breusch and Pagan (1980) Lagrange Multiplier test to choose between pooled OLS and the alternative random or fixed effects models; and the Hausman (1978) specification test to distinguish between random and fixed effects regression models. The Hausman test allows us to compare the fixed and random effects estimates coefficients (Wooldridge, 2008) to determine which model is appropriate for the empirical analysis. A central assumption in random effects estimation is the assumption that the random effects are uncorrelated with the explanatory variables and therefore the decision should be as follows: if Prob>chi2 is not significant (p-value >0.10), then random effects regression model is appropriate. If this is < 0.05 (i.e. significant), then firms fixed effects regression model is appropriate. Using the GCGI as the total corporate governance quality measure in equation 1, the Hausman test gave X 2 of 58.09 (p-value = 0.000), suggesting that the fixed effects regression model is appropriate for the empirical analysis. The fixed effects regression model we initially use to investigate the relationship between firm attributes and corporate governance quality of the Ghanaian firms is in the following general form: 2 According to Black et al (2006), insiders (management) and outsiders (investors) value firm performance differently and therefore we considered both accounting-based and market-based performance measures for our empirical analysis. Also, Samia et al (2011) are of the view that ROA and ROE are short-term performance measures, whereas Tobin s Q represents long-term performance measure. 62

where GCGI it is the dependent variable; is the overall intercept; is a set of firm attributes of interest, j, firm i in year t; is a set of firm specific control variables, k, for firm i in year t; where k = 1 to m; is a vetor of 9 dummy variables representing the 10 sample years; is the firm specific fixed effects, consisting of a vector of 34 dummy variables to represent the 35 sample firms; and is the unobserved error component. In what follows, we perform similar equation to (1) for the sub-indices of the GCGI as dependent variables as well as the pre 2003 and post 2003 firm attributes-corporate governance quality relationship. EMPIRICAL RESULTS Descriptive Statistics Table 2 provides the summary statistics of firm characteristics variables for the Ghanaian firms during the whole period and the pre and the post sub-periods. The results in Panel A of Table 2 show that the sample firms have an overall average size of 6.49 during the whole period (2000-2009), suggesting that they are small size firms. The annual growth and external financing needs are 0.09 and 0.269, respectively. The institutional investors have a strong presence in the Ghanaian firms with an average ownership of 75.96 percent of the total shareholdings. we also calculate director shareholdings with an average of 8.59 of the total shareholdings. The average firm performance measures are 5.70, 18.67 and 1.13 for return on assets, return on equity and Tobin s Q, respectively, and the average firm age is around 33 years. Panel B of Table 2 contains firm characteristics variables for pre 2003 (2000-2002) and post 2003 (2004-2009) introduction of the Ghanaian Code. The sample firms size experienced marginal 4% average change during the pre and post sub-periods. Whereas the Ghanaian firms external financing needs on average increased from 0.218 to 0.296, a change of 38%, growth opportunities experienced significant decrease from 0.196 to 0.018, a change of 91%. Although institutional shareholdings experienced an average downwards change of 3%, the directors invested in more shares after the introduction of the Ghanaian Code with an overwhelming average change of 214%. The firm performance measures on average experienced 63%, 37% and 27% downwards changes for return on assets, return on equity and Tobin s Q, respectively. This suggests that all the firm performance measures on average declined during post 2003 sub-period. As expected, the firm age experience 0.5% change during the pre and post sub-periods. TABLE 2: SUMMARY STATISTICS OF THE GHANAIAN FIRMS CHARACTERISTICS FINANCE is the external financing needs, GROWTH is the growth opportunity, SIZE is the firm size, INSTHOLD is institutional shareholdings, DIRHOLD is the director shareholdings, ROA is the return on assets, ROE is the return on equity, Q-ratio is the Tobin s Q and AGE is the firm age. Mean Median Std. Deviation Minimum Maximum Panel A: Overall period from 2000 to 2009 firm characteristics variables FINANCE 0.269 0.223 0.261 0.124 0.746 GROWTH 0.0900 0.1207 0.51716-0.5700 1.3600 SIZE 6.4980 6.5427 1.32470 4.2300 8.8300 INSTHOLD % 75.96 73.65 13.82 27.00 95.00 DIRHOLD % 8.59 0.10 18.55 0.00 87.00 ROA 5.6996 4.4010 11.32180-9.74 31.59 ROE 18.6668 16.3414 39.76865-31.8801 68.25 Q-ratio 1.1285 0.7249 1.67413-0.0612 7.0100 AGE 32.78 13.9900 5.3 60.5 63

Panel B: Firm characteristics variables for Pre 2003 and Post 2003 Ghanaian Code Pre 2003 (2000-2002) Post 2003 (2004-2009) Change% FINANCE 0.218 0.296 38% GROWTH 0.1959 0.0180 (91%) SIZE 6.3065 6.5439 4% INSTHOLD % 74.33 72.03 (3%) DIRHOLD % 3.35 10.53 214% ROA 10.1279 3.7046 (63%) ROE 20.8554 13.0486 (37%) Q-ratio 1.4082 1.0333 (27%) AGE 32.66 32.81 0.5% Total observation 65 193 Table 3 presents the summary statistics of governance variables for the Ghanaian firms during the whole period and the pre and the post sub-periods. Panel A of Table 2 indicates that the overall firm-level governance quality measured by the Ghanaian corporate governance index is 68.16 percent on average. Firms whose governance quality is the worst have an index of 39 percent and the firms that have the best achieved 100 percent compliance level. The financial affairs and auditing sub-index is 90.93 percent on average and it is the highest firm-level governance quality among the six sub-indices, while the remuneration committee sub-index experienced the lowest of 33.81 percent. The board composition, audit committee, shareholder rights and disclosure sub-indices achieved firm-level governance quality of 65.37 percent, 60.78 percent, 74.26 percent and 87.87 percent, respectively. This suggests that the overall firm-level governance quality in Ghana (68 percent) is on average reasonably higher than 61 percent reported among South African firms (Ntim et al, 2012). In Panel B of Table 3, we investigate the pre and post introduction of the Ghanaian Code firm-level governance quality. As set out in Panel B, the Ghanaian firm-level governance quality on average experience 30% change during the sub-periods with pre 2003 achieving an average of 56.19 percent and 73.16 percent during post 2003. The audit committee sub-index experienced the highest increase from 26.15 percent during pre 2003 to 75.23 percent post 2003, a change of 188%. Although the board composition sub-index experience downwards change of 2%, the remuneration committee, shareholder rights, financial affair and auditing and disclosure sub-indices all experienced substantial increase with a change of 23 percent, 10 percent, 26 percent, respectively. This suggests that the Ghanaian firm-level governance quality experienced significant change during the sub-periods and it is important to investigate whether any of the firm attributes has any impact on the changed governance quality. TABLE 3: SUMMARY STATISTICS OF GOVERNANCE VARIABLES FOR THE GHANAIAN FIRMS, 2000-2009 BOARD is the board composition index, AUCOM is the audit committee index, RECOM is the remuneration committee index, SHOLD is the shareholder rights index, FAA is the financial affairs and auditing index, DISC is the disclosure index and the GCGI is the Ghanaian corporate governance index. Mean Median Std. Dev Minimum Maximum Panel A: Overall period from 2000 to 2009 governance variables (%) BOARD 65.37 66.67 16.54 33.33 100.00 AUCOM 60.78 83.33 41.22 0.00 100.00 RECOM 33.81 16.67 29.67 16.67 100.00 SHOLD 74.26 83.33 11.41 50.00 100.00 FAA 90.93 100.00 12.64 50.00 100.00 DISC 87.87 100.00 16.13 50.00 100.00 GCGI 68.16 72.22 15.08 39.00 100.00 64

Panel B: Governance variables for the Pre 2003 and Post 2003 Ghanaian Code (%) Pre 2003 Post 2003 Change % BOARD 65.38 64.16 (2%) AUCOM 26.15 75.23 188% RECOM 29.23 36.01 23% SHOLD 69.23 76.08 10% FAA 76.67 96.63 26% DISC 74.10 93.35 26% GCGI 56.19 73.16 30% Total observation 65 193 Results Table 4 reports the correlation matrix and it is used to investigate the relationship between all variables within the analysis. High collinearity exists between the GCGI and its sub-indices (i.e. BOARD, AUCOM, RECOM, SHOLD, FAA and DISC). Also, high collinearity exists between AUCOM and FAA & DISC. To avoid the problem of multicollinearity, these variables were included in separate regression models in our empirical analysis. Table 5 presents the results for the firm attributes-governance quality relationship during the whole period. A positive coefficient indicates high governance quality and a negative one low governance quality. As shown in model 1, the external financing needs is found to be statistically significant and positively related to the overall governance quality measured by the Ghanaian corporate governance index, suggesting that hypothesis 1 is supported. This is also consistent with Khanchel (2007) and Da Silveira et al (2007) who argue that firms with greater future financing needs turn to improve their firm-level governance quality. Investigating the constituents of the Ghanaian corporate governance index suggests that the board composition (model 2), audit committee (model 3), shareholder rights (model 5) and disclosure (model 7) sub-indices drive the relationship. Contrary to hypothesis 2, growth opportunities is found to have no association with the overall governance quality (model 1), evidence consistent with Khanchel (2007) who found similar results across the US firms. The results also suggest that firm size is associated with higher overall governance quality (model 1) and the relationship is influenced by audit committee (model 3), remuneration committee (model 4) and disclosure (7) sub-indices. This evidence provides support for hypothesis 3 and is in line with the findings of Khanchel (2007) in the US firms. Although the director shareholdings has no relationship with the overall governance quality (model 1), there is strong relationship between board composition (model 5) and shareholder rights (model 2) sub-indices. The institutional shareholdings is positive and statistically significant (model 1) and the strong relationship is influenced by board composition (model 2), audit committee (model 3), remuneration committee (model 4) and disclosure (model 7) sub-indices. All the firm performance measures are found to have no association with the overall governance quality measured by an index and its sun-indices (models 1-7). These results support the work of Khanchel (2007) and Da Silveira et al (2007) who find their performance measures to have no impact on their index. 65

FINANCE GROWTH SIZE INSTHOLD DIRHOLD ROA ROE QRATIO AGE BOARD AUCOM RECOM SHOLD FAA DISC GCGI Proceedings of the Australia-Middle East Conference on Business and Social Sciences 2016, Dubai Table 4: Pearson correlation matrix of the firm attributes and corporate governance variables FINANCE 1.0000 GROWTH 0.0387 1.0000 SIZE 0.1969 0.1439 1.0000 INSTHOLD 0.0017 0.1147 0.0135 1.0000 DIRHOLD 0.0016 0.1050 0.1348-0.1400 1.0000 ROA 0.2141 0.1344-0.0663 0.1389 0.2302 1.0000 ROE 0.2162 0.1608 0.0271 0.1248-0.2223 0.2846 1.0000 QRATIO 0.313 0.0584-0.1115 0.2767-0.1982 0.1766 0.1843 1.0000 AGE 0.1894 0.0515-0.0525 0.0823 0.1783 0.1439 0.0052-0.0449 1.0000 BOARD 0.1366 0.0276 0.1348 0.1106 0.3625 0.0472 0.1679 0.1566-0.0724 1.0000 AUCOM 0.1504 0.0704 0.0393 0.1230 0.367 0.0378 0.0314 0.0181 0.2138 0.1940 1.0000 RECOM 0.2161 0.0267 0.0932 0.0542-0.1605 0.0338 0.0943 0.0830 0.0080 0.3375 0.4292 1.0000 SHOLD 0.1051 0.0063 0.0241 0.1866 0.2452 0.1182 0.0979 0.0373 0.0923 0.2859 0.2295 0.1639 1.0000 FAA 0.1349 0.1584 0.0184-0.1810-0.0562-0.0925-0.0965-0.0885 0.0480 0.1356 0.5891 0.1086 0.3974 1.0000 DISC 0.0382 0.1284 0.0345 0.1697 0.0242-0.0222-0.0536-0.0847 0.1145 0.2534 0.7323 0.3228 0.3848 0.7002 1.0000 GCGI 0.1773 0.0640 0.0179 0.0855 0.1293 0.0127 0.0654 0.0318 0.1015 0.4812 0.8729 0.6789 0.4603 0.6429 0.8009 1.0000 Notes: The table indicates Pearson s correlation coefficients. FINANCE is the external financing needs, GROWTH is the growth opportunity, SIZE is the firm size, INSTHOLD is institutional shareholding, DIRHOLD is the director shareholdings, ROA is the return on assets, ROE is the return on equity, Q-ratio is the Tobin s Q and AGE is the firm age. BOARD is the board composition index, AUCOM is the audit committee index, RECOM is the remuneration committee index, SHOLD is the shareholder rights index, FAA is the financial affairs and auditing index, DISC is the disclosure index and the GCGI is the Ghanaian corporate governance index. 66

The analysis is further developed to investigate whether the changes in firm-level governance quality might have been impacted by the changes in the firm attributes. Table 6 provides pre 2003 and post 2003 analysis of the firm attributes impact on governance quality. As Panel A and Panel B of model 1 shows, firm size and institutional shareholdings have had consistent positive and significant impact on the overall governance quality during the sub-periods. However, and as shown in Panel A of model 1, the external financing needs, growth opportunities, director shareholdings and all the firm performance measures did not have any impact on the overall governance quality during pre 2003 sub-period. TABLE 5: FIXED EFFECTS REGRESSION RESULTS OF THE IMPACT OF FIRM ATTRIBUTES ON CORPORATE GOVERNANCE QUALITY, WHOLE PERIOD FINANCE is the external financing needs, GROWTH is the growth opportunity, SIZE is the firm size, INSTHOLD is institutional shareholdings, DIRHOLD is the director shareholdings, ROA is the return on assets, ROE is the return on equity, Q-ratio is the Tobin s Q and AGE is the firm age. BOARD is the board composition index, AUCOM is the audit committee index, RECOM is the remuneration committee index, SHOLD is the shareholder rights index, FAA is the financial affairs and auditing index, DISC is the disclosure index and the GCGI is the Ghanaian corporate governance index. The models provide t-statistics which are in parenthesis. Coefficients are on top of parenthesis. Year dummy and firm specific fixed effects dummy variables are included but their coefficients are not reported. ***, ** and * significant at 1, 5 and 10 percent, respectively. Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 GCGI BOARD AUCOM RECOM SHOLD FAA DISC FINANCE 0.094 0.101 0.168 0.023 0.062 0.087 0.054 (3.19)*** (4.30)*** (1.91)* (0.45) (2.54)** (2.65) (2.44)** GROWTH 1.606 0.027 0.127 0.409 0.887 0.588 0.671 (1.47) (1.23) (1.33) (1.36) (1.02) (1.23) (1.86)* SIZE 0.835 0.575 0.387 0.823 0.382 0.520 0.956 (2.90)*** (1.15) (3.98)*** (1.71)* (0.73) (1.17) (2.42)** INSTHOLD 0.004 0.046 0.534 0.543 0.115-0.049 0.090 (2.04)** (2.49)** (1.71)* (2.69)*** (1.16) (0.52) (1.75)* DIRHOLD 0.074 0.593 0.521-0.057 0.451-0.025 0.057 (0.78) (3.71)*** (0.87) (0.17) (2.70)*** (0.34) (0.59) ROA 0.115 0.233 0.077 0.019 0.087-0.134-0.020 (1.50) (0.91) (0.35) (0.15) (1.40) (1.50) (0.40) ROE 0.023 0.031 0.097 0.007 0.043-0.011-0.003 (1.05) (0.79) (1.53) (0.20) (0.42) (0.41) (0.11) QRATIO 0.254 0.241 0.026 0.187 0.192-0.131-0.364 (1.61) (0.88) (0.02) (0.27) (0.58) (0.27) (0.67) AGE 0.707-0.258 0.681 0.150 0.384 0.380 0.452 (5.36)*** (1.60) (9.43)*** (3.33)*** (3.19)*** (3.63)*** (3.20)*** _cons 33.228 77.620-222.448 22.567 40.822 70.918 67.573 (3.40)*** (4.46)*** (3.50)*** (2.15)** (4.26)** (3.90)*** (2.98)*** R 2 0.46 0.25 0.36 0.10 0.28 0.54 0.54 N 283 283 283 283 283 283 283 As expected, the increase in external financial needs and director shareholdings appear to have had a positive and statically significant impact on governance quality during post 2003. Also, all the firm performance measures are found to be statistically significant and positively related to the overall governance quality during post 2003, suggesting that firms become more compliant with the Ghanaian Code when their performance declines during post 2003. This evidence is consistent with Shabbir (2008) who found UK firms to become more compliant with the UK Code of corporate governance when their prior period performance declines. Although growth opportunity certainly lacks significant, these results suggest that the strong relationship reported during the whole period (exception to director shareholdings and firm performance measures) is generally influenced by the changes in firm attributes during the sub-periods. 67

TABLE 6: FIXED EFFECTS REGRESSION RESULTS OF THE IMPACT OF PRE AND POST FIRM ATTRIBUTES ON CORPORATE GOVERNANCE QUALITY FINANCE is the external financing needs, GROWTH is the growth opportunity, SIZE is the firm size, INSTHOLD is institutional shareholding, DIRHOLD is the director shareholdings, ROA is the return on assets, ROE is the return on equity, Q-ratio is the Tobin s Q and AGE is the firm age. BOARD is the board composition index, AUCOM is the audit committee index, RECOM is the remuneration committee index, SHOLD is the shareholder rights index, FAA is the financial affairs and auditing index, DISC is the disclosure index and the GCGI is the Ghanaian corporate governance index. The models provide t-statistics which are in parenthesis. Coefficients are on top of parenthesis. Year dummy and firm specific fixed effects dummy variables are included but their coefficients are not reported. ***, ** and * significant at 1, 5 and 10 percent, respectively Panel A:Pre 2003 (2000-2002) firm attributes impact on corporate governance quality Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 GCGI BOARD AUCOM RECOM SHOLD FAA DISC FINANCE 0.077 0.013 0.243 0.011 0.161 0.049 0.107 (0.99) (1.24) (1.19) (1.10) (0.68) (0.50) (0.85) GROWTH 0.498 0.152 0.633 0.354 0.198 0.070 0.054 (1.06) (0.08) (0.88) (1.14) (0.80) (1.32) (1.01) SIZE 0.733 0.391 0.363 0.437 0.069 0.069 0.090 (2.98)*** (1.93)* (3.35)** (2.09)** (0.60) (2.02)** (1.53) INSTHOLD 0.742 0.171 0.232 0.768 0.634 0.727 0.588 (2.42)** (2.59)** (1.94)* (1.23) (2.23)** (1.39) (0.87) DIRHOLD 0.431 0.198 0.158 1.172 0.702 0.171 0.014 (0.43) (0.29) (0.34) (0.80) (0.58) (0.96) (1.01) ROA 0.181 0.021 0.599 0.077 0.227 0.018 0.184 (1.24) (0.22) (1.22) (0.37) (1.31) (0.10) (0.81) ROE 0.022 0.005 0.055 0.002 0.090 0.030 0.026 (0.65) (0.21) (0.48) (0.04) (2.21)** (0.74) (0.49) QRATIO 0.222 2.629 0.302 0.992 0.729 0.580 0.558 (0.64) (2.06)** (0.36) (0.36) (0.32) (0.69) (0.19) AGE 0.742 0.964 0.170 0.320 0.352 0.737 0.011 (2.30)** (1.07) (0.26) (0.17) (2.10)** (1.69)* (1.81)* _cons 31.906 23.206 12.254 17.503 25.940 23.140 0.085 (3.62)*** (2.36)** (4.14)** (2.39)** (1.69)* (2.65)** (1.79)* R 2 0.55 0.16 0.50 0.31 0.33 0.41 0.31 Panel B: Post 2003 (2004-2009) firm attributes impact on corporate governance quality FINANCE 0.017 0.133 0.095 0.069 0.066 0.012 0.012 (2.96)*** (3.67)*** (1.03) (1.07) (2.02)** (1.46) (2.30)** GROWTH 0.004 0.390 0.424 0.528 0.011 0.160 1.098 (1.01) (1.57) (0.19) (0.34) (1.01) (0.25) (1.14) SIZE 0.544 0.984 1.521 0.666 0.658 0.493 0.894 (2.27)** (1.65)* (1.00) (0.63) (2.22)** (1.13) (2.37)** INSTHOLD 0.132 0.137 0.181 0.401 0.112 0.154 0.117 (2.14)** (0.85) (2.44)** (1.40) (1.77)* (1.31) (0.67) DIRHOLD 0.254 0.713 0.084 0.037 0.511 0.097 0.338 (2.25)** (4.54)*** (2.21)** (0.13) (3.61)*** (0.84) (1.98)** ROA 0.027 0.113 0.059 0.183 0.025 0.102 0.067 (2.41)** (1.27) (2.26)** (1.16) (0.31) (1.66)* (1.69)* ROE 0.017 0.041 0.007 0.103 0.025 0.012 0.030 (1.65)* (1.14) (2.07)** (1.67)* (0.79) (0.44) (0.51) QRATIO 0.529 0.136 0.849 0.835 0.596 0.062 0.082 (1.98)** (0.18) (0.44) (2.13)** (0.88) (0.11) (0.10) AGE 0.218 0.969 1.031 0.295 0.361 0.097 0.630 (0.91) (2.91)*** (1.21) (0.50) (3.53)*** (2.50)** (1.74)* _cons 35.231 26.062 11.333 16.129 22.415 17.759 18.345 (3.94)*** (3.45)*** (2.47)** (1.79)* (2.67)** (2.25)** (2.09)** R 2 0.29 0.21 0.24 0.36 0.26 0.28 0.17 N 193 193 193 193 193 193 193 68

Robustness Check One main concern in using fixed effects regression model is whether sectoral differences affect the firm-level governance quality. To check the robustness of our results, the dummy variables representing the 10 sample years (time effect) were excluded from the analysis. Instead, and in line with Gompers et al (2003), we replaced them with industry dummies to control for the sectoral effect. This procedure serves as sensitivity analyses which enable us to report more consistent and robust empirical results. As a result, we replicated Tables 5 and 6 and included 6 dummy variables to represent the 7 industries to control for sectoral effect. We find that under both specifications the results are qualitatively the same as what we reported earlier. CONCLUSIONS This paper investigates the determinants of good corporate governance in Ghana, focusing on possible differences in findings before and after 2003. We choose 2003 as the dividing time because it was the year that the Ghanaian Code was introduced. An important part of the code was for firms to adapt the practices to suit their own particular circumstance, suggesting that firms with for example future external financing needs will become more compliant (Khanchel, 2007). Given that there is no prior academic research in emerging African economies that has examined the determinants of firm-level governance quality, the objectives of this paper are largely notable. We directly measure a firm-level governance quality by developing a Ghanaian corporate governance index and its six sub-indices for the whole period (2000-2009) and the pre and post sub-periods. We then use a panel data regression framework to investigate the extent to which variations in governance quality can be explained by the changes in firm attributes. The empirical results over the whole period, 2000-2009 show statistically significant and positive relationship between external financing needs, firm size, institutional shareholdings and governance quality. However, we find a shift in the relationship between external financing needs and the overall governance quality after 2003. Prior to 2003, we document no relationship between external financing needs and total governance quality but we find a positive relationship between the two after 2003. We also find that director shareholdings and all the firm performance measures switches significant level following 2003, suggesting that firms with stronger director interest in the form of shares and declined performance measures actually improve their governance quality. The most consistent relationship we find concerns firm size and institutional shareholdings. We conclude that firms with high external financing needs, high director shareholdings and low performance turn to have strong governance. Our results are broadly in agreement with prior studies who reported that firms with greater financing needs usually have the incentive to improve their governance practices (Durnev and Kim, 2005; Khanchel, 2007). The study has important implications in understanding firm attributes that are associated with governance quality in general. The introduction of code of corporate governance in Ghana is of relevance here because firms in emerging African economies with future external financing needs would have to improve their governance quality that can be benchmarked before they can attract investments. The results of this study also support the importance of regulators and policy makers for providing clear code of corporate governance guidelines that firms are expected to disclose the extent of compliance in their annual report. There are some limitations to this study which must be considered when interpreting our results. This study was conducted on a sample of Ghanaian listed firms and therefore generalisation should be limited to this category. Also, our firm-level governance quality was based on comments that show compliance (equally weighted) with the Ghanaian Code recommendations and disclose in each firm s annual reports but does not assess the quality of information provided. That notwithstanding, this study has contributed to the existing literature by providing evidence of firm attributes that explain governance quality among firms representing 92 percent of the Ghanaian stock-market capitalisation. This study is an important first step in understanding firm attributes in explaining governance quality in an emerging African economy like Ghana where there is no evidence available. Future research may consider assigning weights to comments that show compliance with the Ghanaian Code provisions in assessing governance quality. 69