Corporate Governance and Cost of Capital: Evidence from Emerging Market

This study used a researcher self-constructed corporate governance index as a proxy to measure the firm-level corporate governance compliance and disclosure with the 2002 Pakistani Code of Corporate Governance, to examine the relationship between corporate governance and cost of capital. We found a negative and significant association between the Pakistani Corporate Governance Index (PCGI) and block ownership with the firm-level cost of capital. On average, better-governed Pakistani listed firms tend to be associated with a lower cost of capital than their poorly governed counterparts are. As an emerging market, good corporate governance practices are mainly related to minimise corporate failure and assist firms in attracting capital at a lower cost.


Introduction
The 1997 Asian financial crises were an evolving landscape for Asian policymakers and companies. Several institutional and policy weaknesses were uncovered by these crises and led to numerous economic reforms in the region. According to Demise (2006) regulations and guidelines have been legislated in developing countries with the support of international organisations such as the World Bank and Organization for Economic Cooperation and Development (OECD). The Pakistan Stock Exchanges have not been spared these significant reforms in the way companies are managed and controlled, which have swept across the world in recent times. Corporate governance reforms were the most critical part of those reforms that were aimed to restore investors' confidence. With respect to adopting corporate governance codes, and as the case with most developing countries, Pakistan issued its corporate governance code in March 2002, which is regarded as an essential development for corporate governance reforms. This corporate governance code has been established by the combined efforts of the Securities and Exchange Commission of Pakistan (SECP) and the Institute of Chartered Accountants of Pakistan (ICAP). The requirements of the code are comprehensively influenced by UK corporate governance style (Tariq and Abbas 2013). The code has a series of governance provisions that are focused on three main areas, including better disclosure, strengthening of internal control systems, and reforms of the board of directors, with the concern of making it accountable to the stockholders. Claessens and Yurtoglu (2013) find that firms with better corporate governance gain more accessible access to financing and better performance. The critical question is whether adopting similar corporate governance provisions from developed countries can effectively assist Pakistani firms to increase their firm value by reducing their cost of capital. The level of compliance is calculated by using the yearly average of PCGI. The year-wise increase in the level of compliance is also presented in Figure 1. The blue bar of the graph shows a constant increase in the level of compliance over the sample period. However, the increase in the level of compliance varies from year to year. For instance, in the beginning, years, there is a rise of 10 to 30%, which decreases to less than 5% in the last years. This is because most of the firms started following the corporate governance (CG) standards.
On the one hand-contrary to the Berle and Means model of separation of ownership and control-foremost Pakistani firms' ownership structure bears a resemblance to a concentrated family ownership structure. Arguably, this concentrated ownership structure of Pakistani firms is different from those of the Anglo-American structure of dispersed ownership. The corporate governance mechanisms formulated by following markets with dispersed ownership structure may not offer the right remedy to the governance issues for a market with concentrated ownership. Therefore, this study may provide exciting and different findings than those from the Anglo-American countries.
Second, Pakistan's constitution requires that all laws conform to Islam. Although the fiduciary duties set by the Security and Exchange Commission of Pakistan are initially based on Anglo-American common law and shareholding model of corporate governance, more importantly, they must also conform to Islamic business ethics (Ibrahim 2006;Arslan and Abidin 2019). In this regard, strong Islamic notions are incorporated in Pakistani corporate governance code, such as accountability, transparency, and responsibility. These can have important implications for the level of CG compliance and disclosure (Abu-Tapanjeh 2009;Ahmad 2011;Ilyas and Jan 2017).
Third, Pakistan has adopted the Anglo-American model to improve corporate governance standards in its corporate sector. The agency problem is expected to be different in developing countries like Pakistan, due to the nature of ownership structures, where the conflict of interests is between minority (outsider) and majority (insider) shareholders instead of managers and shareholders as is the case in UK and US (Bozec and Bozec 2011). Therefore, this study sheds light on whether the adaptation of commonly accepted corporate governance standards, as proposed by Anglo-American countries, can improve the corporate governance practices in emerging economies like Pakistan.

Mean of PCGI (%)
Level of Compliance (%) Increase in level of Compliance (%) Figure 1. The compliance level with Pakistani Corporate Governance Index (PCGI) based on the full sample. Note on the authors' calculation: The level of compliance is calculated by using the yearly average of PCGI. The year-wise increase in the level of compliance is also presented in Figure 1. The blue bar of the graph shows a constant increase in the level of compliance over the sample period. However, the increase in the level of compliance varies from year to year. For instance, in the beginning, years, there is a rise of 10 to 30%, which decreases to less than 5% in the last years. This is because most of the firms started following the corporate governance (CG) standards.
Pakistan makes it unique for this study due to some reasons. First, like most countries in the developing world, Pakistani companies have controlling shareholders in the form of family ownership. This provides the controlling shareholders with both the incentive in the case of low cash flow rights and opportunity in the case of high free cash flows to expropriate outsider minority shareholders (Bozec and Laurin 2008). Similarly, strong corporate governance and investor protection found in the developed countries are believed to be more effective, as compared to Asian countries (Dyck and Zingales 2004). Notably, the Pakistani corporate setting shares some level of similarities and differences with the UK corporate environment.
On the one hand-contrary to the Berle and Means model of separation of ownership and control-foremost Pakistani firms' ownership structure bears a resemblance to a concentrated family ownership structure. Arguably, this concentrated ownership structure of Pakistani firms is different from those of the Anglo-American structure of dispersed ownership. The corporate governance mechanisms formulated by following markets with dispersed ownership structure may not offer the right remedy to the governance issues for a market with concentrated ownership. Therefore, this study may provide exciting and different findings than those from the Anglo-American countries.
Second, Pakistan's constitution requires that all laws conform to Islam. Although the fiduciary duties set by the Security and Exchange Commission of Pakistan are initially based on Anglo-American common law and shareholding model of corporate governance, more importantly, they must also conform to Islamic business ethics (Ibrahim 2006;Arslan and Abidin 2019). In this regard, strong Islamic notions are incorporated in Pakistani corporate governance code, such as accountability, transparency, and responsibility. These can have important implications for the level of CG compliance and disclosure (Abu-Tapanjeh 2009;Ahmad 2011;Ilyas and Jan 2017).
Third, Pakistan has adopted the Anglo-American model to improve corporate governance standards in its corporate sector. The agency problem is expected to be different in developing countries like Pakistan, due to the nature of ownership structures, where the conflict of interests is between minority (outsider) and majority (insider) shareholders instead of managers and shareholders as is the case in UK and US (Bozec and Bozec 2011). Therefore, this study sheds light on whether the adaptation of commonly accepted corporate governance standards, as proposed by Anglo-American countries, can improve the corporate governance practices in emerging economies like Pakistan.

Firm-Level Corporate Governance and Cost of Capital
Theoretically, corporate governance encompasses different mechanisms that can assure creditors and shareholders of the firm on a return on their investments (Shleifer and Vishny 1997). In the case of most developing countries, when firms have controlling shareholders (see Claessen et al. 2000;Faccio and Lang 2002), corporate governance mitigate agency problems between insiders shareholders and outside investors, including both creditors and minority shareholders. Insider shareholders enjoy the control of the firm's operation by having a large portion of voting rights and, therefore, may expropriate outside investors, including minority shareholders and creditors (Cumming et al. 2019;La Porta et al. 2002). In this context, good corporate governance practices are intended to safeguard minority shareholders and creditors among other outside investors against the expropriation of controlling shareholders (Ilyas and Jan 2017). Arguably, when investors feel protected, they are motivated to participate in the capital market more active and are more likely willing to pay more for such firms' securities. Firms can enjoy a lower cost of raising capital, which in turn raises the firms' value. Thus, we postulate the first hypothesis: Hypothesis 1. There is a statistically significant and negative relationship between firm-level corporate governance and firms' cost of capital.

Ownership Structures and Cost of Capital (COC)
In this subsection, the hypotheses of different ownership variables are developed with firms' cost of capital. The ownership variables include director ownership, institutional ownership, government ownership, block ownership, and foreign ownership.
From a managerial signalling perspective, Bebchuk and Weisbach (2010) argue that the directors have more information about the firms, compared to outsiders (minority shareholders and creditors). Therefore, it is more likely that the executives can use the firms' statistics for the personal interests that shift risk to, rather than share risk with, outside shareholders (Demsetz and Lehn 1985), which, in turn, may increase the information asymmetry problem between directors and outside investors (minority shareholders and creditors). Therefore, it is likely that the firm with higher director ownership can have a higher cost of borrowing and a negative impact on profitability. It can be argued that director ownership may worsen the agency problem, as the outsider and insider can have conflicting interests (Demsetz and Lehn 1985). The second hypothesis is as follows: Hypothesis 2. There is a positive and statistically significant association between director ownership and firms' COC.
A limited number of studies only provide evidence on the relationship between institutional ownership and one component of the cost of capital. Institutional investors usually have a higher monitoring power, and it has been suggested that they can play a crucial role by forcing managers to make decisions in the best interest of shareholders (Shleifer and Vishny 1986). Crutchley et al. (1999) argue that institutional investors can have an impact on firms' capital structure.
Theoretically, monitoring can be beneficial to reduce the agency cost by minimising the conflicts of directors and investors (Solomon et al. 2003;Jensen and Meckling 1979). Arguably, intuitional investors with a significant shareholding are proposed as important CG mechanism for three main reasons (Diamond and Verrecchia 1991;Donnelly and Mulcahy 2008). First, having a considerable portion of shareholding and voting power permits them to take necessary actions (Donnelly and Mulcahy 2008).
Second, institutional investors have resources and capabilities to have more information than minority shareholders (Smith 1976). Third, with better knowledge and expertise, they can evaluate the firm's decisions and can interpret the disclosed information in annual reports (Chung et al. 2003;Bos and Donker 2004;Khan 2016;Elshandidy and Neri 2015). Thus, it is expected that institutional ownership can increase firm value by decreasing the firm's cost of capital. Therefore, the following is predicted: Hypothesis 3. Institutional ownership and firms' COC are significantly negatively associated.
From the resources dependence theory perspective, firms with higher government ownership can easily access financing from the government (Eng and Mak 2003). Arguably, firms may take the benefit of higher government cost of capital and, in turn, may increase the firm value. Similarly, (Siebels and Knyphausen-Aufseß 2012) argue that government ownership may not affect the managers due to their aligned interests with other corporate owners. Individually, executives may strive for improvement in the firm performance to improve and protect their reputation (Conyon and He 2011). In contrast, (Eng and Mak 2003) argue that higher state-owned firms may origin the agency problem. In this regard, government ownership may cause intervention in firms' operations, which may bring about weak corporate governance practices (Konijn et al. 2011). For example, the government may employ directors and a CEO irrespective of qualification (Cornett et al. 2010;Tsamenyi et al. 2007). Hence, we expect the following: Hypothesis 4. Firms' COC and government ownership are significantly and negatively associated.
The dominance of majority shareholders in publically traded firms demonstrates the willingness to accept risk by minority shareholders. (Bozec et al. 2014) argue that such risks are accepted by minority shareholders based on compensation. High-risk results in higher cost of capital for firms. Arguably, a higher cost of capital means a higher rate of return for investors that can be a form of compensation to them. Hence, it can be argued that block ownership is expected to have a more direct link with the cost of capital rather than financial performance and firm value, mainly as value is not only affected by risk but also by the firm's growth opportunities (Hail and Leuz 2006).
Empirical studies report mixed evidence in the relationship of block holders and firm-level cost of capital. For instance, Bozec et al. (2014) report significant empirical evidence of a positive correlation between excess control and the weighted average cost of capital. Similarly, Elston and Rondi (2007) report empirical evidence that concentrated inside ownership is significantly and positively associated with the firm cost of capital for Italian firms while having no significant relationship between the variables for German firms. In contrast, Pham et al. (2012) report significant empirical evidence of a negative relationship between concentrated ownership and the weighted average cost of capital.

Hypothesis 5.
There is a statistically significant association between block ownership and firms' COC.
A firm's choice of issuing debt or equity to finance their activities can be affected by foreign investors. Theoretically, information asymmetry is relatively higher among foreign investors because of language and distance (Huafang and Jianguo 2007). Higher foreign ownership may lead to debt financing as a governance mechanism; thus, it may force firms to issue debt over equity (Phung and Le 2013). Additionally, firms may prefer debt rather than equity as they may take advantage of foreign investors' relationship and reputation to have easy access to international capital markets, which will usually provide a lower cost of borrowing and, thus, lower cost of capital. As a result, it the following can be argued: Hypothesis 6. Firms' COC and foreign ownership are significantly and negatively associated.

Data Sample and Summary Statistics
The sample used in analysing the CG compliance level (PCGI) and its impact on the cost of capital (COC) was made up of the Karachi Stock Exchange (KSE) listed firms. The majority of KSE listed firm's annual reports became publicly available in 2003 with required CG information after the issuance of Pakistani CG code in 2002. This makes it possible to gather data from 2003, when the code was effectively implemented, and firms started to publish their annual reports. The sample ends in 2013, as it was the most recent year with available data at the time of data's being manually collected.
To be included in the sample of this study, a firm has to meet two conditions. First, the firms' eleven-year annual reports from 2003 to 2013, inclusive, must be available. Second, its corresponding eleven-year financial and stock market information had to be available. In this study, the financial industry is not included in the final sample for three main reasons. First, financial firms have a different capital structure than those of nonfinancial firms, which may have an impact on firm value (Lim and Wang 2007;Ali Shah and Butt 2009). Second, financial firms have been suggested to be heavily regulated. In the case of Pakistan, financial firms are required to comply with more regulations than their industrial counterparts do. We ended up with 160 firms for the period 2003 to 2013 and with 1760 firm-year observations.
There are three types of data being used in this study, namely (i) corporate governance variables, (ii) financial variables, and (ii) Stock Market variables. First, using a content analysis approach, corporate governance variables were manually collected from the annual reports of the sampled firms. These annual reports were collected from different sources: Rest of World Filings of the Perfect Information Database, the companies' website, and the KSE website. Firms' annual reports that were not available in the above sources were obtained from the SECP head office in Islamabad, Pakistan. Second, the data on financial variables of 130 firms were collected from Datastream, while the data for the remaining 30 firms were collected from Balance Sheet Analyses of State Bank of Pakistan's publication. Sampled firms' monthly stock prices, Government of Pakistan T-Bill rates, and Market indices variables constitute the third type of data used in this study, which were collected from Datastream. Missing or insufficient data related to the Company's monthly stock prices, Government of Pakistan T-Bill rates, and market indices data were collected from the website of the business recorder. Table 1 presents summary statistics. The minimum of PCGI is 0.00, and the maximum is 97.18, while the mean score of the index is 54.23 for 1760 firm-year observations. There is a relatively large variation in the CG compliance among Pakistani listed firms, as shown by a standard deviation of 33.57. The findings are in line with the previous corporate governance literature (e.g., Ntim et al. 2012;Henry 2008), indicating that corporate governance standards improve over time. The mean of director ownership is 20.88%, with a minimum of 0% and a maximum of 98%. The average of director ownership is relatively high among Pakistani listed firms 1 (Samaha et al. 2012;Henry 2008). Appendix A presents the definitions of variables. COC denotes the cost of capital, COD denotes the cost of debt, COE denotes the cost of equity, PCGI denotes the Pakistani Corporate Governance Index, DOWNP represents director ownership, IOWNP represents institutional ownership, GOWNP represents government ownership, BOWNP represents block ownership, FOWNP represents foreign ownership, BIG4 represents the audit firm size, BSZ represents the size of the board of directors, BGEN represents board diversity on the basis of gender, BNAT represents board diversity on the basis of nationality, LTA represents firm size as the log of total assets, ROE represents return on equity as a measure of profitability, SALESG represents growth opportunities, LVG represents leverage, and β represents Beta-a measure of risk.
The mean of institutional ownership is 10.70%, with a minimum of 0 and a maximum of 95%, revealing that there is a substantial variation in this variable. However, this average institutional ownership is consistent with some of the previous studies. For instance, Aggarwal et al. (2011) report average institutional ownership of 8%, 8%, and 9% in Greece, Hong Kong, and New Zealand, respectively. On the other hand, Chung and Zhang (2011) report over 50% of institutional ownership among US firms.
Concerning government ownership, the average is 6.39% with a minimum of 0 and a maximum of 95%, revealing that the Pakistani government relatively holds a high percentage of firms' share and is expected to have an impact on the willingness of firms to comply with CG provisions. The average of block ownership is 55.45%, with a minimum of 0 and a maximum of 99.80%, revealing a higher level of ownership concentration among Pakistani listed firms. The high level of block ownership may suggest a low CG compliance and disclosure, as it is expected that the market for control may not be working well, as compared with a low concentration of ownership.
Regarding foreign ownership, its mean is 9.97%, with a minimum of 0 and a maximum of 93%, with a standard deviation of 21.62%. This may suggest that the presence of foreign ownership can have an important role in improving the CG standards among Pakistani listed firms. This is supported by a correlation coefficient. Appendix C shows the correlation matrix in more detail.

Empirical Model
The impact of the level of CG compliance and its relationship with COC for Pakistani listed firms is estimated by the following ordinary least square model: where i and t subscript represents firm and time, respectively. COC is the cost of capital calculated by the weighted average cost of capital. PCGI is the Pakistani Corporate Governance Index. DOWNP is the percentage of shares owned by directors, and IOWNP is the percentage of shares owned by institutions. GOWNP indicates the percentage of shares owned by the government. BOWNP represents the percentage of shares owned by shareholders with at least 5%. FOWNP accounts for the percentage of shares owned by the foreigner. A set of control variables includes firm size (LTA), profitability (ROE), sales growth (SALESG), leverage (LEVG), capital expenditure (CETA), and Beta (β). Beta (β) measures the Beta of the firm by using a regression of stock return to market returns. Industry and year fixed effects are included in all models; ε is the error term.

Main Findings
In this study, the impact of the level of corporate governance compliance and its relationship with the cost of capital for Pakistani listed firms were investigated. As shown in Table 2, we find that the coefficient on PCGI is negative and statistically significant at 5% level, suggesting that firms with a high level of corporate governance standards have a lower cost of capital. Our finding is consistent with our first hypothesis. In the case of most developing countries, when firms have controlling shareholders (see Claessen et al. 2000;Faccio and Lang 2002), corporate governance mitigates agency problems between insider shareholders and outside investors, including both creditors and minority shareholders. Similarly, good corporate governance practices in Pakistan are intended to safeguard minority shareholders and creditors among other outside investors against the expropriation of controlling shareholders. Table 2 shows the findings of the influence of ownership variables on firms' cost of capital. First, the coefficient on director ownership is positive and statistically significant, suggesting that firms with a high level of director ownership have a higher cost of capital. Our finding is in line with the second hypothesis, and the prediction of agency theory a higher level of director ownership may worsen agency problems (Demsetz and Lehn 1985). In a similar vein, it has been suggested that higher director shareholdings may make a firm more vulnerable to collusion between directors and firm management (Vafeas and Theodorou 1998;Konijn et al. 2011). In this regard, Bennedsen and Wolfenzon (2000) argue that one of the three ways by which multiple block-holders can influence firm value is that they can use their power to form a coalition to expropriate value at the expense of other shareholders.
Second, the coefficient on institutional ownership on the cost of capital is positive and statistically insignificant, meaning that the percentage of institutional ownership has no explanatory power in explaining the variation in the firm-level cost of capital. This is contrary to the formulated third hypothesis. Theoretically, the relationship between institutional ownership and cost of capital being negative can be useful, as monitoring can be beneficial in reducing the conflicts of interest between investors and directors (Jensen and Meckling 1979;Solomon et al. 2003). However, the current study does not lend empirical support to the CG literature in regard to Pakistan (e.g., Bhojraj and Sengupta 2003;Piot and Missonier-Piera 2009), and it documents a negative relationship between institutional ownership and firm-level cost of capital.  (BIG4), size of the board of directors (BSZ), board diversity on the basis of Gender (BGEN), firm size as log of total assets (LTA), profitability (ROE), growth opportunities (SALESG), leverage (LVG) and systematic risk (β). Parameter estimates were obtained by OLS estimation (panel least squares). The year 2003 and Auto industry were excluded from the analysis in order to avoid the dummy variable trap. The asterisks *, **, and *** denote the 10%, 5%, and 1% level of significance, respectively.
Third, the coefficient on government ownership is positive and statistically insignificant, suggesting that there is no statistically significant association between government ownership and firms' COC. This finding shows that the level of government ownership has no explanatory power in explaining the variation in firm-level COC. Theoretically, this positive relationship between government ownership and COC is consistent with the prediction of agency theory. Eng and Mak (2003) argue that higher government ownership may cause agency problems where government ownership may lead to intervention in firms' operations, which may result in poor corporate governance practices (Konijn et al. 2011;Elshandidy and Neri 2015). For instance, the government may appoint the CEO and directors regardless of experience and qualification (Tsamenyi et al. 2007;Cornett et al. 2010).
Fourth, unlike the institutional and government ownership, the coefficient on block ownership is negative and statistically significant at 1% level, suggesting that there is a relationship between the block ownership and firm-level cost of capital. This shows that Pakistani firms with a higher level of block ownership have a lower cost of capital than those firms with a smaller percentage of block ownership. This is consistent with the prediction of agency theory in which the dominance of majority shareholders in publicly traded firms demonstrates that minority shareholders have the risk of expropriation. Bozec et al. (2014) argue that minority shareholders can accept such a risk as long as they are compensated. Empirically, this finding is in line with the previous literature (e.g., Pham et al. 2012) that provides empirical evidence of a negative relationship between ownership concentrations on the firm-level weighted average cost of capital.
Finally, the coefficient on foreign ownership is positive and statistically significant at the 1% level, indicating that there is a statistically significant and positive relationship between foreign ownership and firm-level cost of capital, inconsistent with the sixth hypothesis. This finding shows that Pakistani firms with a higher level of foreign investors have a higher cost of capital than those with less or no foreign investors. Theoretically, this positive relationship between foreign ownership and the cost of capital is consistent with the prediction of information asymmetry. This issue is relatively higher among foreign investors because of language and distance obstacles (Huafang and Jianguo 2007), which may lead to a higher cost of capital. Empirically, the finding of this positive relationship between foreign ownership with firm-level COC is in line with the prior literature (e.g., Boubakri et al. 2016).

Unweighted Index Versus Weighted Index
The current study responds to the literature in order to address the possibility that the main findings may be sensitive to the type of corporate governance index. Hence, a weighted corporate governance index instead of an unweighted corporate governance index is employed by assigning 20% weight to each sub-index of PCGI, whereas the unweighted corporate governance index has different weights assigned to each sub-index 2 . This procedure is line with prior studies (e.g., Beiner et al. 2006;Ntim et al. 2012) that used the same method to test whether their main findings are sensitive to the weighted corporate governance index (WPCGI) or not. Therefore, the PCGI in Equation (1) is replaced by the WPCGI. Table 3 shows a comparison for results using the unweighted index versus those using the weighted index. 2 The corporate governance index that is used in the current study to measure corporate governance compliance and disclosure among Pakistani listed firms consists of 70 corporate governance provisions divided into five sub-indices, which are equally weighted, but the number of corporate governance provisions are different in the five sub-indices and leads to different weights being assigned to each sub-index.  (BIG4), size of the board of directors (BSZ), board diversity on the basis of gender (BGEN), firm size as log of total assets (LTA), profitability (ROE), growth opportunities (SALESG), leverage (LVG) and systematic risk (β). Parameter estimates were obtained by OLS estimation (panel least squares). The year 2003 and Auto industry has been excluded from the analysis in order to avoid the dummy variable trap. The asterisks *, **, and *** denote the 10%, 5%, and 1% level of significance, respectively.
Adjusted R-square is 0.540825 for unweighted index and 0.550872 for weighted index, suggesting that 54% and 55% variability in PCGI and WPCGI, are jointly explained by independent variables in Equation (1). Similarly, the F-statistic is 60.19378, using unweighted index, and 60.41580, using the weighted index, and both are statistically significant at 1% level. This suggests that both analyses are appropriate, and all the parameters in the analyses are jointly significant. Generally, the findings of both analyses are similar, as both predict the same sign of coefficient, magnitude of coefficient, and level of significance, either using PCGI or WPCGI.

Robustness Tests
The current study employs alternative proxies for the cost of capital in order to account for the possibility that the main findings are sensitive to different proxies. In particular, and in line with past studies (e.g., Pham et al. 2012), the cost of equity and cost of debt is used as an alternative cost of capitals' measurement. The cost of debt is considered for the cost of selecting debt covenants; therefore, it signals credit risk and agency problems (Kim 2018;Elshandidy and Neri 2015). Tables 4 and 5 report results for the cost of equity and cost of debt, respectively. We find consistent results with our main findings. Notes: Variables are defined as follows. Cost of capital (COC), Pakistani Corporate Governance Index (PCGI), director ownership (DOWNP), institutional ownership (IOWNP), government ownership (GOWNP), block ownership (BOWNP), foreign ownership (FOWNP), audit firm size (BIG4), size of the board of directors (BSZ), board diversity on the basis of gender (BGEN), firm size as log of total assets (LTA), profitability (ROE), growth opportunities (SALESG), leverage (LVG), and systematic risk (β). Parameter estimates were obtained by OLS estimation (panel least squares). The year 2003 and Auto industry were excluded from the analysis in order to avoid the dummy variable trap. The asterisks *, **, and *** denote the 10%, 5%, and 1% level of significance, respectively.   (BIG4), size of the board of directors (BSZ), board diversity on the basis of gender (BGEN), firm size as log of total assets (LTA), profitability (ROE), growth opportunities (SALESG), leverage (LVG), and systematic risk (β). Parameter estimates were obtained by OLS estimation (panel least squares). The year 2003 and Auto industry were excluded from the analysis in order to avoid the dummy variable trap. The asterisks *, **, and *** denote the 10%, 5%, and 1% level of significance, respectively.
We further employed lagged structure to examine the extent to which the main findings were affected by the endogeneity problem. Table 6 presents the results. Generally, the findings of both analyses are similar, as both analyses predict almost the same sign and magnitude of coefficient with the level of significance.  . The year 2003 and Auto industry were excluded from the analysis in order to avoid the dummy variable trap. The asterisks *, **, and *** denote the 10%, 5%, and 1% level of significance, respectively.

Two-Stage Least Squares Result
After carrying out the Durbin-Wu-Hausman exogeneity investigation, the current study rejects the null hypothesis of no endogeneity as the coefficient on P-PCGI is statistically significant (0.000) at 1% level of significance with PCGI. The finding of this investigation shows that the endogeneity problem exists. Therefore, following, the current study uses the Two Stage Least Square (2SLS) technique as robust to find out how far the findings are biased and inconsistent due to this problem.
2SLS is performed in two stages. In the first stage, the PCGI is regressed on four alternative CG variables, namely board diversity based on nationality, the number of non-executive directors on the board, the number of board of directors' meetings, and capital expenditure, besides, to controlling variables. The alternative CG variables' selection is based on literature (e.g., Ntim et al. 2012;Pham et al. 2012;Ntim and Soobaroyen 2013;Tariq and Abbas 2013). The equation below specifies this regression, where the predicted value of PCGI and residuals will be saved as P-PCGII and R-PCGI, respectively. The study accepts the P-PCGII as a valid instrumental variable as P-PCGII is significantly associated with PCGI and insignificantly related to R-PCGI. This decision is taken based on the correlation matrix that includes PCGI, P-PCGII, and R-PCGI.
where PCGI refers to the Pakistani CG index, and BNAT, NEXD, BFM, and CE are defined as board diversity based on nationality, the number of non-executive directors on the board, the number of board of directors' meetings, and capital expenditure, respectively. Control variables remain the same, as explained in Equation (2). In the second stage, Equation (2) is re-estimated, using P-PCGII instead of PCGI, as follows: Risks 2020, 8, 104 16 of 29 where all variables remain the same as in Equation (1), except the P-PCGII that is being used as an instrumental variable for the primary independent variable. The findings of 2SLS (robust results) and OLS estimation (primary analysis) are presented in Table 7, simultaneously, in order to compare the results.  ( . The year 2003 and auto industry were excluded from the analysis in order to avoid the dummy variable trap. The asterisks *, **, and *** denote the 10%, 5%, and 1% level of significance, respectively. The 2 stage least square (2SLS) finds a negative and significant association between PCGI and block ownership with the cost of capital. Similarly, a positive and significant nexus between director ownership and foreign ownership with the cost of capital is also consistent with the findings of the main analysis. However, minor sensitivity in the magnitude of coefficients and level of significance can be observed. For instance, director ownership is statistically significant at the 1% level, which was previously significant at the 5% level in the main analysis. Similarly, government ownership is significant at the 10% level in the 2SLS analysis, whereas it was insignificant in the principal analysis. The findings of control variables in 2SLS are mainly similar to the primary analysis by using OLS.

Conclusions
This paper has sought to empirically ascertain whether Pakistani listed firms that comply with 2002 PCCG have improved firm value and lowered the cost of capital than those with less or no compliance. Specifically, using a sample of 160 Pakistani listed firms from 2003 to 2013, this study has examined the relationship between corporate governance structure and firm cost of capital. The level of compliance with PCGI and factors influencing the level of compliance and disclosure are also examined in this study.
We found a negative and statistically significant relationship between PCGI and the cost of capital. The evidence of a statistically significant PCGI and cost of capital relation implies that, on average, better governed Pakistani listed firms tend to be associated with a lower cost of capital than their poorly governed counterparts. Firms with a high level of director ownership have a higher cost of capital. The percentage of institutional ownership and government ownership do not explain the variation in the firm-level cost of capital. The coefficient of block ownership is negative and statistically significant at 1% level of significance, suggesting that Pakistani firms with a higher level of block ownership have a lower cost of capital than those firms with a smaller percentage of block ownership. As an emerging market, good corporate governance practices are particularly relevant to minimise corporate failure and assist firms to attract capital at a lower cost, as compared to other counterparts.

Contributions and Policy Implications
This study makes numerous contributions and extensions to the extant CG literature. For instance, this study offers for the first-time direct evidence on the effectiveness of CG reforms in Pakistan. Precisely, it provides detailed findings on the level of CG compliance and disclosure with the 2002 PCCG among listed firms. Similar to a limited number of prior studies in emerging markets, the introduction of 2002 PCCG facilitates uniformity and convergence of CG practices; the findings recommend that CG practices still differ largely among Pakistani listed firms over the eleven-year period examined. Additionally, to study the value-creating role of CG mechanisms by using an alternative approach (COC) to those which were used in the previous literature (ROA, ROE, and Tobin's Q) is another contribution to the literature, as there is a lack of empirical evidence on CG compliance and COC.
Similarly, the findings obtained from investigating the nexus between the CG standards and COC have several implications, and recommendations can be drawn from these findings. For instance, the findings of the current study demonstrate that there is a negative and significant association between the PCGI and block ownership with firm-level COC. This implies that, on average, better governed Pakistani listed firms tend to be associated with lower COC than their poorly governed counterparts. To an emerging market, good CG practices are particularly important, and, as such, practices may not only assist in minimising corporate failure but may also assist firms in attracting capital at a lower cost, as compared to their counterparts. Additionally, director ownership and foreign ownership are positively and significantly associated with firm-level of COC. This implies that firms can minimise director ownership to attract external financings at a lower cost. Hence, policymakers may encourage firms to further improve their CG structures in order to attract foreign investors. Finally, using a relatively old dataset could be a limitation of this study. However, in the CG studies, such time differences are somehow acceptable because of several reasons. For instance, the data collection is a tough job because of its nature of being hand-collected. Furthermore, as CG rules do not change quickly and also do not impact the firms' decisions so quickly, findings can still be generalised. However, the latest dataset with latest techniques and with additional statistical tests can be the future avenue of this area of research. Additionally, weighted index can also be used in this regard. Acknowledgments: The first author would like to thank Kwanku Opong and Marco Guidi for their advice and encouragement as the dissertation supervisors. The paper could not have been developed and improved if they were not supporting the first author as his supervisors, this paper is developed the idea from the 1st author's dissertation and improved with the participants of all authors. The authors, however, bear full responsibility for the article. The 4th author would like to say thank you to the Ho Chi Minh City Open University to support for this research (the presentation at the International Conference on Business and the partial funding to conduct this research).

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Table A1. Summary of variables used in CG mechanisms and firm COC model.

WACC
Weighted average COC (WACC) is computed by using the after-tax cost of debt and cost of equity by using weights of total debt and total equity to the total market capitalisation of the firm.

PCGI
Pakistani Corporate Governance Index (PCGI) consists of 70 provisions from Pakistani Code of Corporate Governance, which takes a value of 1 if a particular CG provision is disclosed in annual reports of company, and 0 otherwise; scaled to a value between 0% and 100%. DOWNP Percentage of shares owned by directors to the total shares held by firm. IOWNP Percentage of shares owned by institutions to the total shares held by firm. GOWNP Percentage of shares owned by government to the total shares held by firm.

BOWNP
Percentage of shares owned by shareholders with at least 5% of total shares to the total shares held by firm. FOWNP Percentage of shares owned by foreigner to the total shares held by firm. BIG4 1 if firm is audited by one of the big-four 3 audit firms, and 0 otherwise. BSZ The total number of directors on the board of firm at the time of AGM. BGEN 1 if firm has a female board member, and 0 otherwise.

The Control Variables
LTA Natural log of total book value of assets of the firm.

ROE
Earnings before interest and tax to total equity of the firm. SALESG Sales in current year menus sales in last year to sales of last year. LEVG Total book value of debt to total book value of assets.
β Three years monthly stock returns are used to calculate Beta of firm by using a regression of stock return to market returns. Binary number 1 is assigned if it discloses the names of member who attended committees of the board in each annual report, and 0 otherwise.   Binary number 1 is assigned if it offers an explanation of the actual and potential risk of the company, and 0 otherwise. A binary number of 1 is assigned if the auditor performs duties according to IFAC, no management role, and this information is disclosed in annual reports, and 0 otherwise.