Do board characteristics drive firm performance? An international perspective

The aim of our research is to analyze how board characteristics influence firm performance. In this paper, we specifically examine how board size, board independence, CEO duality, female directors and board compensation affect firm performance in a sample of international firms. The final panel data sample is composed of 10,314 firm-year observations belonging to 34 countries that have been grouped into six geographic zones: Africa, Asia, Europe, Latin America, North America and Oceania. Drawing on agency theory and dependence resource theory, we posit five hypotheses. The results show that some board characteristics, such as board size, board independence and having a female director, are positively associated with firm performance, whereas CEO duality, contrary to our expectations, also impacts positively on firm performance. Moreover, board compensation is not associated with firm performance. Tobin’s Q was used to measure firm performance, although an accounting measure was also employed for robustness analyses and to provide more validity to our results.


Introduction
Corporate governance refers to the procedures and processes according to which organizations are directed and controlled by their CEO, board of directors and senior management. The literature highlights the fact that the board of directors is an important and highly effective internal mechanism of corporate governance and fulfills two important functions in companies: supervision of executive management in representation of the shareholders and providing business resources and assessment. In their supervisory role, the boards use their time and resources to monitor firm performance and the behavior of the executive managers.
The theoretical basis of the supervisory function of the board is derived from agency theory, which highlights the possible conflicts of interest that may arise from the separation of ownership and control in companies (Jensen and Meckling 1976). According to Fama and Jensen (1983), agency theory views the board of directors as an essential element of the control mechanism to ensure that the problems resulting from the principal and agent relationship are controlled. This theory is the theoretical framework adopted by most research on corporate governance (Shleifer and Vishny 1997).
In addition to this theory, and in accordance with resource dependence theory, boards of directors help companies improve their performance by reducing their dependence on the external environment and contingencies (Pfeffer and Salancik 1978). This theory maintains that the board of directors is an essential link between the firm and the external resources that a firm requires to maximize its performance (Pfeffer and Salancik 1978). According to these theories, an effective corporate governance system increases confidence in the company and therefore attracts investment and talent, which in turn can result in a better firm performance.
For Hsu and Wu (2014), interest in issues regarding board characteristics has increased in recent years due to corporate fraud, corporate misconduct, negligence and massive loss of shareholders' wealth, with corporate scandals in different multinational companies being especially relevant. This has made it a priority issue for both researchers and policy makers. In this sense, previous studies have documented how certain board characteristics, such as board independence (Brown and Caylor 2006) or size (Jackling and Johl 2009;Cheng 2008), among others, influence firm performance. These research studies have mainly addressed a single country (e.g. Australia, Japan, UK), and therefore it is considered necessary to advance the knowledge of how certain board characteristics influence firm performance considering countries with different institutional environments.
Therefore, this paper aims to extend the debate outlined above and contribute to the existing literature in several ways. In the first place, we wished to analyze how certain board characteristics, such as board size, board independence, CEO duality, existence of female directors and board compensation or remuneration can influence firm performance and find out whether the results obtained are in line with previous findings according to the theories being taken into account: agency theory and resource dependence theory.

Do board characteristics drive firm performance? An…
Secondly, previous studies have focused mainly on specific time periods and on a sample of firms operating in a specific country, such as Australia, Japan and UK (Kiel and Nicholson 2003;Ullah 2017;Weir and Laing 2001;Chang et al. 2015), while a few papers, as far as we know, have drawn on a sample of several countries (e.g. Fauver et al. 2017;Claessens and Yurtoglu 2013;Kim and Lu 2013;Iliev et al. 2015). Our paper also employs an unbalanced panel dataset of 10,314 firm-year observations within 34 countries and 9 economic sectors for the period 2004-2015. By including firms in countries with different institutional environments, we increase the heterogeneity of the dependent variables and, therefore, also the robustness and generalization of the results. Moreover, the availability of data from several countries adds special interest to our analysis, since it allows the analysis to be extended to a broader point of reference than the corporate system addressed in most of the existing studies. In this regard, the firms in our sample can be split up between common and civil law countries, which have different characteristics (Fauver et al. 2017;Claessens and Yurtoglu 2013;La Porta et al. 1998). Civil law systems are more oriented to protect the government and creditors' interests, companies get financial support from financial entities, there is a low legal protection of shareholder's rights, capital markets are less liquid and ownership is highly concentrated. In contrast, common law systems are more oriented to shareholders, companies get funding from capital market because they are more liquid and ownership is more dispersed. Furthermore, while in civil law countries the conflict of interests arise between controlling shareholders and minority shareholders, in common law countries these conflicts are between managers, including CEO, and shareholders.
In the third place, we wished to check whether board characteristics affect firm performance equally, since previous studies have shown that some of them, such as board size, have a positive effect (e.g., Jackling and Johl 2009;Guest 2009;Kiel and Nicholson 2003), while other characteristics do not maintain such an obvious relationship. Independent directors, for example, can have a significant positive influence on the value creation activities of the company and, therefore, on its performance through its strategic decision-making (Gabrielsson 2007). In the case of women on the board, different studies have also found a positive relationship with firm performance (e.g. Liu et al. 2013;Kılıç and Kuzey 2016;Terjesen et al. 2016). In contrast, with other board characteristics, such obvious results have not been obtained, especially in the case of CEO duality, where several authors such as Corbetta and Salvato (2004) and Song et al. (2006) obtained the opposite results to those they had raised in their hypotheses, since the hypotheses had been formulated in negative terms, whereas CEO duality was found to have a positive effect on firm performance. For board compensation and firm performance, the relationship has not always been positive. Thus, Donaldson and Davis (1991) found that board compensation was not related to firm performance. Similar results were obtained by other authors, such as Larcker et al. (2007) and Jeppson et al. (2009). Thus, this paper aims to extend the debate outlined above and contribute to the existing literature in several ways. First of all, we analyze how certain board characteristics, such as board size, CEO, duality, female directors and board compensation or remuneration can influence firm performance and check whether the results obtained are in 1 3 line with the previous results according to the theories developed: agency theory and resource dependence theory.
In addition, a contribution is made to the supervision and control function performed by the board of directors, with agency theory and resource dependence theory being used to better understand this control function.
This paper is structured into the following sections. After the introduction, the theoretical framework and the hypotheses are presented in Sect. 2. In Sect. 3, we present the methodology used and the sample and the variables are described. In Sect. 4, the results obtained in this research are analysed and, finally, in the last section, the conclusions and the implications derived from our analysis are presented.

Theoretical background and hypotheses
The area of board characteristics and their link to firm performance has been of increasing interest in recent years. One of the most plausible explanations is based on agency theory, such that given the separation of ownership and control, corporate governance becomes a way of reducing agency costs. In this regard, agency theory (Jensen and Meckling 1976;Fama and Jensen 1983) upholds that the main goal of the principal (shareholder) is to maximize firm value and obtain maximum firm performance, and thus they delegate this task to the agent (manager), who is assumed to be rational, individualist, risk-averse and effort-averse, and therefore his or her objectives may be contrary to those of the principal. Since the structure of the board of directors (the principal) and its possible relation to firm performance (represented by the manager-agent) is what informs our research, it can be said that agency theory is at the foundation of the present study.
Although agency theory has been a very popular approach for previous empirical studies examining the impact of various board characteristics on firm performance, these studies have reported mixed and often contradictory results. (e.g., Hutchinson and Gul 2004;Jensen and Murphy 2004). A plausible explanation for these inconclusive results is that the agency theory-based studies have focused only on the monitoring function of the boards and ignored the resources that boards of directors provide to companies. Therefore, the application of other theories such as resource dependence theory is important.
According to resource dependence theory, boards help companies improve their performance by reducing their dependence on the external environment and contingencies (Pfeffer and Salancik 1978), thereby lowering their transaction costs, which may fundamentally assist their survival. From a resource dependence perspective, it can be similarly argued that a larger board brings greater opportunity for more links and hence access to resources. Other authors, such as Zahra and Pearce (1989), argue that when a firm designates a member to the board, it expects the directors to use their knowledge, skills and experience to provide useful counsel to the company, improve its legitimacy and reputation and support the firm's capabilities for strategic assessment and implementation (Hillman and Dalziel 2003).
Although the agency theory and dependence resource theory are accepted in the supervision, control and reputation function performed by the board of directors, 1 3

Board size
The first factor we analyze is board size. Board size refers to the total number of directors that can influence the corporate governance practices of companies and, as a consequence, their performance (Yermack 1996). Given that the strategic decisions of the firm are taken by the board, its characteristics can affect the decision-making process (Fodio and Oba 2012). Moreover, board size is an important dimension of the board structure (Noor and Fadzil 2013) and can be used as a proxy for the efficiency of the board of directors (Jia and Zhang 2013). In general, it may be thought that larger boards are more inefficient because agreements are more difficult to reach, owing to multiple interests ). However, one can also consider that an increase in the number of members will bring different personal characteristics that can help to link the organization with its external environment and ensure critical resources, including prestige and legitimacy, since with more members, there will be a greater wealth and variety of opinions and experiences and an increase in the supervisory capacity of the board (Mintzberg 1993;Pearce and Zahra 1992).
In this regard, the agency and resource dependence approaches seem to support the idea that the larger the board, the more efficient it will be. According to agency theory (Jensen and Meckling 1976), boards with a large number of directors are considered to be more effective in handling and controlling management; therefore, in terms of the agency problem, a greater number of directors involved in management activities will make the board more vigilant. The agency theory focuses precisely on the supervisory and control function of the board of directors (Gul and Leung 2004;Hutchinson and Gul 2004;Jermias 2007;Tsui et al. 2001). Furthermore, resource dependence theory suggests that large and diversified boards are more likely to contribute the knowledge capital of the business sector in a deeper way, which can later influence the quality of strategic decision-making and, in the end, have a positive impact on performance (Arosa et al. 2010). In short, board size is an essential link between the company and the external resources that it needs to maximize its performance (Pfeffer and Salancik 1978).
Therefore, drawing on agency theory, resource dependence theory and past evidence (e.g., Jackling and Johl 2009;Guest 2009;Kiel and Nicholson 2003), it seems that there is a positive relationship between board size and firm performance. This positive relation seems to derive from the fact that when the board of directors has more members, there will be a greater wealth and variety of opinions and experiences and an increase in the supervisory capacity of the board, at the same time that more bonds will be formed between the company and its surroundings, thus facilitating the obtaining of resources that have a favourable impact on the creation of value for the company and, ultimately, on firm performance (Guest 2009).
Other authors, such as Coles et al. (2008), also obtain a positive relationship for a sample of 8165 firm-year observations. Their study obtains a positive association between firm performance measured by Tobin's Q and board size in the context of large companies, which indicates that a large board size helps improve the overall performance of companies. For companies listed on the Karachi Stock Exchange 1 3 Do board characteristics drive firm performance? An… (Pakistan), Singh et al. (2018) also show a positive relationship between board size and firm performance.
Considering the above, Lehn et al. (2009) conclude that board size should be related to the size of the company and its growth opportunities and, therefore should not be set based on a specific size recommended by a standard. Therefore, the board must have a size suitable for facilitating decision-making and reducing the costs of its operation, guaranteeing diversity of criteria adapted to business needs (Pucheta-Martínez 2015). According to the results obtained in previous investigations, we propose the following hypothesis: H1 Ceteris paribus, there is a positive association between board size and firm performance.

Board independence
Independent directors are professionals without any relation to the management of the company, so they are unlikely to interfere in corporate decisions with their personal opinions (Agrawal and Knoeber 1996). Independent directors are a relevant control mechanism for companies because they make more objective decisions than managers and shareholders and provide new points of view different from the more traditional points of view, which are more focused on financial issues (Zahra and Stanton 1988). In fact, Ibrahim and Angelidis (1995) suggest that independent directors usually care about all stakeholders, given that most firms are less likely to retain useful information for their stakeholders. Furthermore, independent directors are more interested in guaranteeing the appropriate behaviour of firms and the fulfilment of the established goals. Thus, it is expected that independent directors will be more objective and independent when assessing the firms' management and behaviour than executive directors (Sonnenfeld 1981).
In this regard, agency theory argues that boards with a high proportion of independent directors are more effective in governing, handling and controlling management (Jensen and Meckling 1976;Fama and Jensen 1983;Jizi et al. 2014;Fernández-Gago et al. 2016). In the same vein, Volonte (2015) indicates that independent directors reduce agency conflicts as they act as an effective oversight mechanism for the board and are more likely to protect the interests of shareholders. Furthermore, according to resource dependence theory, independent directors can have a significant positive influence on the value creation activities of the company and, therefore, on its performance through its strategic decision-making (Gabrielsson 2007). Dalton et al. (1998) claim that independent directors provide experience and objectivity. Furthermore, they highlight some reasons which may justify the presence of independent directors on boards, such as (a) their implication in correcting unbalanced powers and agency problems, (b) their interest in improving the corporate governance quality of firms and in guaranteeing the social interest in the decision-making process and, finally, (c) their commitment to providing an effective monitoring role of managers.
Regarding the relationship between independent directors and firm performance, the results obtained in the previous literature shows a positive association. Dahya and McConnell (2007), who investigated the impact of board independence on firm performance for a sample of companies in the United Kingdom, find that companies that incorporate independent or external directors on their boards exhibit a significant improvement in their performance. In addition, for a sample of companies from the same country, Weir et al. (2002) found that board independence positively influences performance, measured by Tobin's Q. Other research conducted in other countries by different authors also obtains positive results in the relationship between independent directors and firm performance. In this sense, Coles et al. (2001) suggest with their research that a greater proportion of independent directors improves the control and strategic decision-making processes of the board through better tracking, and Ritchie (2007) Hence, drawing on agency theory and resource dependence theory, a higher presence of external or independent directors can lead to a greater increase in the internal control of the company, a greater disclosure of information as well as a more effective search for actions and decisions that protect shareholders. Accordingly, we predict a positive relationship between board independence and firm performance, proposing the following hypothesis: H2 Ceteris paribus, there is a positive association between the proportion of independent directors on boards and firm performance.

CEO duality
CEO duality refers to the situation when the CEO also occupies the position of Chairperson (control) of the board of directors, resulting in a concentration of managerial power (Rechner and Dalton 1991;Surroca and Tribo 2008). For authors such as Jensen (1986), the concentration of power in one person increases the risk that he or she can develop strategies that favour their personal interests to the detriment of the company, since they may give rise to opportunistic actions and deplete the efficiency of the board, which would have a negative effect on business performance.
Within the frame of the agency theory (Jensen and Meckling 1976), firms should divide the roles of CEO and Chairperson to avoid a concentration of power in the hands of a single person. Considering this theory, Goyal and Park (2002) and Hashim and Devi (2009) argue that the separation of the roles of Chairperson and CEO is preferable in order to avoid a concentration of power in the hands of a single person and to provide an effective system of controls and balances regarding the activities and performance of executive directors. In this same regard, Fosberg and Nelson (1999) have argued that the separation of the functions of management and control within a company reduces agency costs and leads to better firm performance.
Authors such as Christensen et al. (2010), Dahya et al. (2009), Faleye (2007, Tuggle et al. (2010) and Singh et al. (2018) conclude that CEO duality is negatively associated with company profitability, since it prevents the board of directors from independently exercising its control function over management. This negative effect can also be due to the fact that duality facilitates the accumulation of power in the same person and gives rise to low-quality financial information, the manipulation of profits and the generation of opportunistic actions, as well as undermining the efficiency of the board of directors. Therefore, according to agency theory, CEO duality will negatively influence the outcome of the company, since according to this theory it is preferable to separate the roles of Chairperson and CEO to avoid a concentration of power in the hands of only one person.
Previous research conducted in specific countries such as New Zealand (Prevost et al. 2002), Canada (Bozec 2005Jermias 2007), Hong Kong (Tsui et al. 2001) and the United States (Jermias and Gani 2014) have obtained evidence that CEO duality has a negative impact on firm performance. These results are in line with the recommendations made on corporate governance by the Business Roundtable (2012) concerning principles of corporate governance, or the Corporate Governance Code (2010) in the UK, which consider that the role of the CEO and Chairperson should be kept separate. Considering the results of previous research studies, we propose the following hypothesis: H3 Ceteris paribus, there is a negative association between CEO duality and firm performance.

Female directors
Another variable that we will analyze is the presence of women on boards of directors, a topic that in recent years has acquired special relevance. The two theories that support the idea that women contribute to a more effective board and better firm performance are agency theory and the resource dependency theory.
Agency theory that postulates that female directors on boards are the ones that contribute a wider range of perspectives in decision-making, which increases the independence of the board while reducing agency costs and, as a consequence, the value of the company increases, all of which makes it one of the crucial theories used by academics to discuss the direct effect of female presence on the performance of the company (Hillman and Dalziel 2003). In this regard, agency theory assumes that female directors can act as a mechanism of supervision and control of board activities (Adams and Ferreira 2009). As an argument that can explain this premise, authors such as Solimene et al. (2017) argue that female directors have acquired high levels of education in recent years, such as Master's degrees and other postgraduate studies and, therefore, are considered highly professional and experienced when making important decisions on the boards of directors.
According to the resource dependency theory, female directors contribute unique and valuable resources and relationships to their boards as they are considered an essential link between the company, its environment and the external resources on which a company depends. It is a matter of capturing resources that improve the economic performance of the company thanks to the ties and contacts that female directors have with the environment (Hillman et al. 2007).
Several authors (e.g., Gul et al. 2011) have argued that women play an important role in the field of business management, since women bring to their profession values and criteria that differ from those used by men. Authors such as Seierstad (2016) consider that the role of female directors on boards is relevant for two reasons: one is due to the business model and the second is due to social justice. The first is based on the premise that having female directors on boards will contribute new skills that will in turn have an impact on improving firm performance; the second focuses on the need to establish heterogeneity on boards of directors in order to represent all the individuals on the teams. In this research, the first premise is the most appropriate one because we believe that women directors on boards can provide different points of view in decision-making processes compared to men, which can be very beneficial for business (Hoobler et al. 2016).
The arguments presented and the theoretical framework derived from agency theory (Jensen and Meckling 1976) and resource dependence theory seem to support the idea that a greater number of women on boards of directors will increase firm performance. The previous literature has also shown a positive effect of female directors on firm performance. For instance, in a sample of companies in the United States, Carter et al. (2003) find that the proportion of women on boards of directors has a positive influence on firm value measured by Tobin's Q, concluding that diversity has a positive effect on firm performance. Other research, by Erhardt and Werbel (2003), also conducted for US companies, found that a greater degree of board diversity is associated with higher firm performance, measured on this occasion by return on assets (ROA) and return on capital investments. In addition, Konrad et al. (2008) found that the impact of women on boards is greater when there is more than one woman present on that board. In these cases, women feel stronger together and, as such, can have a stronger voice in the decision-making process, implying that they will also feel more valued. Other research conducted by different authors also found a positive relationship (e.g. Liu et al. 2013;Kılıç and Kuzey 2016;Terjesen et al. 2016); hence, these studies provide strong evidence that it is better to have women on the board to be able to obtain better firm performance. Considering the previously established arguments, we propose the following hypothesis: H4 Ceteris paribus, there is a positive association between the presence of female directors on boards and firm performance.

Board compensation
In recent years, board compensation or remuneration has become one of the most debated issues in the literature on board characteristics of corporate governance, especially when it comes to finding a balance between compensation systems that align the interests of principal and agent and, at the same time, distribute the risk between both figures. One of the theories on which board compensation relies is the agency theory, in that the idea is to develop a compensation package that makes it possible to attract, retain and motivate managers (Conyon 2006). Thereby providing management with an 1 3 Do board characteristics drive firm performance? An… incentive toward better firm performance in such a way that it is possible to retain the manager, who is efficient or attract others who are even more efficient.
Over the years, the compensation of board members has mainly consisted of a fixed compensation and small additional supplements, such as all expenses paid for meetings. However, the difficulties of the traditional remuneration systems, both fixed and short term, fail to solve the agency problems by not avoiding the short-term opportunistic behavior of the directors and, therefore, not aligning the interests of directors and shareholders, which has meant that variable and long-term remuneration systems are increasingly being used to favor both shareholders and board members (Murphy 1999). In short, and following Jensen and Meckling (1976), the compensation programs should be designed to motivate board members to select and implement actions that increase the wealth and value of the company and, therefore, firm performance. In this regard, Magnan et al. (2010) argue that board compensation should be high enough in order to attract competent and experienced directors and to reward their duties. However, this remuneration should not be as high as impairing their independence, objectivity and judge. Board compensation should be determined in a transparent and objective way by considering the integration of the talent of directors and the corporate governance's philosophy of a firm, and not should base on short-term goals because it may encourage speculative behaviours contrary to the long-term targets of firms.
Diverse studies have been conducted to analyze the relationship between board compensation and firm performance. In this regard, Molonko (2004) reports that board compensation is positively associated with performance of a sample of Kenyan banks. The paper of Watson and Wilson (2005) supports the alignment role performed by board pay between managers and owners as suggested by agency theory, since their findings show that board compensation has a positive effect on firm performance. For Conyon (2006), linking board remuneration to firm performance should be considered fair for shareholders, since board members are rewarded based on their individual and firm performance. Andreas et al. (2012) provide evidence that board remuneration plays a relevant role as corporate governance mechanism, especially in companies where other corporate governance tools are weaker, by affecting positively firm performance. Consistent with this evidence, Müller et al. (2014) also show that board members well-rewarded perform better by enhancing firm performance and Handa (2018) documents that the relationship between board remuneration and bank performance is positive. Thus, based on above arguments, we posit the following hypothesis: H5 Ceteris paribus, there is a positive association between board compensation and firm performance.

Sample
Our initial sample consisted of 13,179 international firm-year observations from 2004 to 2015. We then excluded from this sample financial entities and companies for which all the data necessary for the analysis was not available, removing 2865 companies in all. Thus, the final panel data sample was composed of 10,314 firmyear observations. The data (company-related, economic, financial and corporate governance) used in this research come from the Thomson Reuters database, an international database that collects all fundamental data in reporting entities from all countries in targeted global indices (FTSE 100 index, S&P 500 index, Dow Jones index, NIKKEI 225 index, Euro Stoxx index, CAC 40 index, IBEX 35, NASDAQ 100 index), among others. Our sample is unbalanced because during the period explored, not all firms are observed, given that some of them delisted because of acquisitions and mergers or because they went bankrupt, among other things. Financial entities were removed because they must comply with particular accounting rules and their supervision by financial authorities is stricter. Accordingly, it is more difficult to compare the annual financial statements of financial entities with those of the non-financial firms. The companies of the sample belong to 34 countries, shown in Table 1, where the observations by country and their frequencies are also provided. According to the figures in Table 1, the countries most represented are the United States with 33.50% of the companies in the sample, Canada with 10.20%, United Kingdom with 10% and Japan with 9.8%. In contrast, the countries with the fewest firms in the sample are Egypt, Greece and Israel with 0.1% and Jersey and Portugal with 0.2%. These countries have been classified into six big geographic areas in order to control regional effects: Africa, Asia, Europe, Latin America, North America and Oceania. By including firms in regions with different institutional environments, we increase the robustness and generalization of the results.
Corporate governance attributes vary greatly among countries. In this regard, in Table 2 we provide some institutional dimensions, which characterise the corporate governance system of countries in our sample. Specifically, we analyse, drawing on the papers of La Porta et al. (1998) and Claessens and Yurtoglu (2013), the legal origin, the legal rights' strength, the creditors' rights, the legal protection of minority shareholders, the disclosure requirements, the shareholders' rights (one share, one vote) and the efficiency of judicial system. As appreciated in Table 2, concerning the legal origin of the countries in our sample, almost 35% of them are located in common law systems, while 65% are domiciled in civil law systems. Additionally, the majority of common law countries in our sample tend to have, on average, strong formal legal rights, particularly some countries in Asia, Oceania, Europe, North America and South Africa, in comparison to civil law countries, where have, on average, a medium level of formal legal rights, mainly countries from European and Scandinavian contexts. Some countries form Latin America and African show weak formal rights. The creditors' rights tend to be stronger in common law countries than in civil law countries, although the differences between them are not very significant showing, on average, an intermediate level of creditors' rights. Regarding the legal protection of minority shareholders' rights, it can be appreciated that it is higher in common law systems than civil law systems, where it tends to be low. The disclosure requirements in common and civil law countries are consistent with the level of legal protection of minority shareholders' rights, except for some countries in both systems. The judicial efficiency in both common and civil law countries is similar, although it tends to be a little stronger in common law systems. Finally, it is observed in Table 2 that the shareholders' right-one share, one vote-prevails in civil law countries in contrast to common law, where it is hardly considered.
The companies in the sample operate in nine industry sectors, which are shown in Table 3. In this analysis, we used the TRBC economic sector classification supplied by Thomson Reuters. Consistent with the Global Industry Classification Standard (GICS), our industry classification is a market-based classification. As can be    Legal rights strength's index Strength of legal rights index ranges between 0 and 10 (0-3 = low, 3-5 = low-medium, 5-7 = mediumhigh, 7-10 = high) Creditor rights' index Creditor rights aggregate score ranges between 0 and 4 (0-1 = low, 1-2 = low-medium, 2-3 = mediumhigh, 3-4 = high) Legal protection of minority shareholders' index Legal protection of minority shareholders' index ranges between 0 and 100 (0-30 = low, 30-50 = lowmedium, 50-70 = medium-high, 70-100 = high) Disclosure requirements' index Disclosure requirements' index ranges between 0 and 100 (0-30 = low, 30-50 = low-medium, 50-70 = medium-high, 70-100 = high) Shareholders' rights-one share/one vote Dummy variable that takes the value 1 if this right exists in the law and 0, otherwise Efficiency of judicial system The degree to which the judicial system of a country can guarantee compliance with laws and recommendations ranging from 0 to 10 (0-3 = low, 3-5 = low-medium, 5-7 = medium-high, 7-10 = high) observed in Table 2, 20.70% of the companies in our sample operate in the industrial sector, 18.88% in consumer cyclical, 13.25% in basic materials and 10.76% in energy. In contrast, the telecommunications services sector represents 3.82%, the lowest proportion.

Dependent variable
In line with past research (e.g., Carter et al. 2010;Vafei et al. 2015;Kalsie and Shrivastav 2016;Hoffmann 2014;García-Ramos et al. 2017), we use Tobin's Q as proxy for calculating our dependent variable, firm performance. Most researchers who employ Tobin's Q to measure firm performance use Chung and Pruitt's (1994) approximation, which is calculated as the market value of a firm's common stock + the liquidating value of preferred stock + the book value of debt divided by the total assets (Jiang et al. 2015).

Independent variables
The first independent variable used is board size, labelled as B_SIZE and measured as the total number of board members. Board independence (B_INDEP) is the second independent variable employed, calculated as the ratio between the total number of independent members on boards and the total number of board members. The third independent variable is CEO duality (CEO_DUALITY), measured as a dummy variable that takes the value 1 when the Chairperson of the board serves simultaneously as CEO and 0, otherwise. The presence of women directors on boards is the fourth independent variable considered in this analysis, denoted by FEM_DIR and computed as the ratio between the total number of female directors on boards and the total number of directors on boards. Finally, board compensation is the fifth independent variable used, labelled as B_COMPENSATION and 1 3 Do board characteristics drive firm performance? An… measured as the log of board compensation, which is the total compensation paid to all board members.

Control variables
Several control variables have been taken into account. Board meetings (B_MEET-ING) is the first control variable considered, measured as the number of meetings held by boards every year. In line with prior research (e.g. Zhu et al. 2016;Mishra and Kapil 2018), we also expect a positive association between board meetings and firm performance. The second control variable included in the model is firm size, SIZE, calculated as the log of total assets (Kong and Kong 2017;Tyagi et al. 2018).
We predict a positive effect of firm size on firm value, consistent with Palaniappan (2017a), who show that company size positively affects firm performance. The leverage of firms has also been considered as a control variable. The leverage is denoted by LEVERAGE and is measured as the ratio between firm debt and total assets. We expect a negative effect of leverage on firm value, since firms with more debt are more likely to perform worse. Authors such as Tsuruta (2017) and Vieira (2017) provide evidence on the negative relationship between a firm's leverage and firm performance. Another control variable used is capital intensity, labelled as CAPITAL_INTENSITY and computed as the ratio between firm long-term or fixed assets and total assets. Previous research shows mixed findings about the impact of capital intensity on firm value. Jorgenson and Vu (2005) and Bernard and Jensen (2007) report a better firm performance when capital intensity increases, while Lee and Qu (2011) demonstrate a non-linear association, concretely quadratic, between capital intensity and firm value, concluding that as the capital intensity of companies increases, firm performance decreases, but beyond a certain threshold, more capital intensity results in a higher firm value. Thus, we expect that firm performance can be positively or negatively affected by capital intensity. Ownership concentration is also another control variable considered. This variable is labeled as OWNSHIP_ CON and is measured as the average percentage of shares held by the three largest shareholders in the ten largest firms in each country, in line with Eklund and Desai (2013). As Maury and Pajuste (2005) and Boerkamp (2016) suggest, a high ownership concentration will negatively affect firm performance since dominant shareholders will use their decision power for their own benefit, in contrast with the interests of minority shareholders. Dual-class stocks has been also controlled, denoted by DUAL_CLASS and calculated as a dummy variable that codes the value 1 if companies have dual-class stocks (class A/B or registered/bearer shares) and 0, otherwise. Dual-class stocks imply that firms have two different classes of stocks, which vary voting rights, dividend payments or other characteristics. We expect a negative association between dual-class shares and firm performance, consistent with Zhang (2002). Another control variable considered is the operating performance, labelled as OPERATING_PERFORM. This variable is measured as the ratio between operating income and total assets, in line with Vafeas and Theodorou (1998), who show a positive association between operating performance and firm performance. The sector in which firms operate has also been included as a control variable (Shehata et al. 2017). We use the TRBC economic sector classification supplied by Thomson Reuters. This variable is labelled as SECTORS, and is measured as a dummy variable that takes the value 1 if the firm operates in the sector analyzed and 0, otherwise. We employ the following 9 sectors: basic materials, consumer cyclical, consumer non-cyclical, energy, healthcare, industrials, technology, telecommunication services and utilities. The regional effect has been also controlled by using six geographic zones: Africa, Asia, Europe, Latin America, North America and Oceania. This variable is defined as REGIONS and is measured as a dummy variable that will be code 1 if the country of the sample belongs to the region explored and 0, otherwise. Finally, year effects (YEAR t ) are also controlled by considering a set of dummy variables. In Table 4, we show a description of all variables used in this research.

Model
Below, we provide the model estimated to check our hypotheses: where the individual effect Ʊ i represents each company, but it does vary over time, while θ it differs over time among companies. Our panel data model was estimated with pooled OLS regressions with corrected (clustered) standard errors in order to control for firm-fixed effects (e.g., Rogers 1993;Williams 2000).

Descriptive statistics and correlations
In Table 5, we offer the descriptive statistics of all the variables used in this analysis. As can be observed, the Q_TOBIN ratio of international firms, on average, is 0.57 (median: 0.58; minimum: 0.07 and maximum: 2.29), in line with Kiel and Nicholson (2003), who obtained a Tobin's Q value of 0.13. Board size (B_SIZE) is, on average, 10.61 board members, board independence (B_INDEP), on average, accounts for 63.67% and female directors (FEM_DIR) on boards represent, on average, 11.97%. CEO_DUALITY, that is, when the Chairperson of the board serves simultaneously as CEO of the firm, on average, accounts for 32.40% and board compensation (B_COMPENSATION) is 11.69 (measured as the log of board compensation). Companies' boards (B_MEETING) meet, on average, 8.50 times every year, Do board characteristics drive firm performance? An…  Mean and standard deviation. Q_TOBIN is the market capitalization of common stock + book value liabilities divided by the book value of total assets; B_SIZE is the total number of directors on boards; B_INDEP is the proportion of independent directors on boards = total number of independent on boards/total number of directors on boards; CEO_DUALITY is a dummy variable that takes the value 1 if the same person serves simultaneously as CEO and President of the board and 0, otherwise; FEM_DIR is the proportion of female directors on boards = total number of female directors on boards/total number of directors no boards; B_COMPENSATION is the log of board compensation; B_MEETING is the number of meetings held by boards every year; SIZE is the log of total assets; LEVERAGE is debt over total assets; CAPITAL_INTENSITY is the ratio of long-term or fixed assets over total assets; OWNSHIP_CON is the average percentage of shares held by the three largest shareholders in the ten largest firms in each country; DUAL_CLASS is a dummy variable that takes the value 1 if firms have dual-class stocks (class A/B or registered/bearer shares) and 0, otherwise; OPERAT-ING_PERFORM is the ratio between operating income and total In Table 6, we show the correlation matrix in order to check whether there are multicollinearity concerns. According to the figures provided in Table 6, none of the correlation coefficients are greater than 0.80. As a result, we can confirm that multicollinearity is not a problem in our analysis.

Multivariate analysis
In Table 7, we provide the findings of all models estimated. In Model 1, we analyse how board size affects firm performance, in Models 2 and 3, we examine the effect of board independence and CEO duality, respectively, on firm value and, finally, in Models 4 and 5, we explore the impact of women directors on boards and board compensation on firm performance, respectively. In Model 6, we show the finding for all independent variables estimated in one model.
In Model 1, the variable board size (B_SIZE) has the expected sign and is statistically significant. Therefore, the first hypothesis is accepted, suggesting that board size in our sample of international firms has a positive effect on firm performance. In this regard, larger boards are more likely to enhance firm value. Our findings are consistent with Bhatt and Bhattacharya (2015) and Kalsie and Shrivastav (2016), assets; BASIC MATERIALS is a dummy variable: 1 = Basic Materials; 0 = Otherwise; CONSUMER CYCLICAL is a dummy variable: 1 = Consumer Cyclical; 0 = Otherwise; CONSUMER NON-CYCLICAL is a dummy variable: 1 = Consumer Non-Cyclical; 0 = Otherwise; HEALTHCARE is a dummy variable: 1 = Healthcare; 0 = Otherwise; INDUSTRIALS is a dummy variable: 1 = Industrial; 0 = Otherwise; TECHNOLOGY is a dummy variable: 1 = Technology; 0 = Otherwise; TELECOMMUNICATION SERVICES is a dummy variable: 1 = Telecommunication Services; 0 = Otherwise and UTILITIES is a dummy variable: 1 = Utilities; 0 = Otherwise. AFRICA is a dummy variable: 1 = If the country is in Africa; 0 = Otherwise; ASIA is a dummy variable: 1 = If the country is in Asia; 0 = Otherwise; EUROPE is a dummy variable: 1 = If the country is in Europe; 0 = Otherwise; LATINAMERICA is a dummy variable: 1 = If the country is in Latin America; 0 = Otherwise; NORTH-AMERICA is a dummy variable: 1 = If the country is in North America; 0 = Otherwise; OCEANIA is a dummy variable: 1 = If the country is in Oceania; 0 = Otherwise  1 3 Do board characteristics drive firm performance? An…    Do board characteristics drive firm performance? An… Table 6 (continued) Q_TOBIN is the market capitalization of common stock + book value liabilities divided by the book value of total assets; B_SIZE is the total number of directors on boards; B_INDEP is the proportion of independent directors on boards = total number of independent on boards/total number of directors on boards; CEO_DUALITY is a dummy variable that takes the value 1 if the same person serves simultaneously as CEO and President of the board and 0, otherwise; FEM_DIR is the proportion of female directors on boards = total number of female directors on boards/total number of directors no boards; B_COMPENSATION is the log of board compensation; B_MEET-ING is the number of meetings held by boards every year; SIZE is the log of total assets; LEVERAGE is debt over total assets; CAPITAL_INTENSITY is the ratio of long-term or fixed assets over total assets; OWNSHIP_CON is the average percentage of shares held by the three largest shareholders in the ten largest firms in each country; DUAL_CLASS is a dummy variable that takes the value 1 if firms have dual-class stocks (class A/B or registered/bearer shares) and 0, otherwise; OPERATING_ PERFORM is the ratio between operating income and total assets; BASIC MATERIALS is a dummy variable: 1 = Basic Materials; 0 = Otherwise; CONSUMER CYCLI-CAL is a dummy variable: 1 = Consumer Cyclical; 0 = Otherwise; CONSUMER NON-CYCLICAL is a dummy variable: 1 = Consumer Non-Cyclical; 0 = Otherwise; HEALTHCARE is a dummy variable: 1 = Healthcare; 0 = Otherwise; INDUSTRIALS is a dummy variable: 1 = Industrial; 0 = Otherwise; TECHNOLOGY is a dummy variable: 1 = Technology; 0 = Otherwise; TELECOMMUNICATION SERVICES is a dummy variable: 1 = Telecommunication Services; 0 = Otherwise and UTILITIES is a dummy variable: 1 = Utilities; 0 = Otherwise. AFRICA is a dummy variable: 1 = If the country is in Africa; 0 = Otherwise; ASIA is a dummy variable: 1 = If the country is in Asia; 0 = Otherwise; EUROPE is a dummy variable: 1 = If the country is in Europe; 0 = Otherwise; LATINAMERICA is a dummy variable: 1 = If the country is in Latin America; 0 = Otherwise; NORTHAMERICA is a dummy variable: 1 = If the country is in North America; 0 = Otherwise; OCEANIA is a dummy variable: 1 = If the country is in Oceania; 0 = Otherwise *P value < 0.1; **P value < 0.05; ***P value < 0.01     Q_TOBIN is the market capitalization of common stock + book value liabilities divided by the book value of total assets; B_SIZE is the total number of directors on boards; B_INDEP is the proportion of independent directors on boards = total number of independent on boards/total number of directors on boards; CEO_DUALITY is a dummy variable that takes the value 1 if the same person serves simultaneously as CEO and President of the board and 0, otherwise; FEM_DIR is the proportion of female directors on boards = total number of female directors on boards/total number of directors no boards; B_COMPENSATION is the log of board compensation; B_MEET-ING is the number of meetings held by boards every year; SIZE is the log of total assets; LEVERAGE is debt over total assets; CAPITAL_INTENSITY is the ratio of long-term or fixed assets over total assets; OWNSHIP_CON is the average percentage of shares held by the three largest shareholders in the ten largest firms in each country; DUAL_CLASS is a dummy variable that takes the value 1 if firms have dual-class stocks (class A/B or registered/bearer shares) and 0, otherwise; OPERATING_ PERFORM is the ratio between operating income and total assets; BASIC MATERIALS is a dummy variable: 1 = Basic Materials; 0 = Otherwise; CONSUMER CYCLI-CAL is a dummy variable: 1 = Consumer Cyclical; 0 = Otherwise; CONSUMER NON-CYCLICAL is a dummy variable: 1 = Consumer Non-Cyclical; 0 = Otherwise; HEALTHCARE is a dummy variable: 1 = Healthcare; 0 = Otherwise; INDUSTRIALS is a dummy variable: 1 = Industrial; 0 = Otherwise; TECHNOLOGY is a dummy variable: 1 = Technology; 0 = Otherwise; TELECOMMUNICATION SERVICES is a dummy variable: 1 = Telecommunication Services; 0 = Otherwise and UTILITIES is a dummy variable: 1 = Utilities; 0 = Otherwise. AFRICA is a dummy variable: 1 = If the country is in Africa; 0 = Otherwise; ASIA is a dummy variable: 1 = If the country is in Asia; 0 = Otherwise; EUROPE is a dummy variable: 1 = If the country is in Europe; 0 = Otherwise; LATINAMERICA is a dummy variable: 1 = If the country is in Latin America; 0 = Otherwise; NORTHAMERICA is a dummy variable: 1 = If the country is in North America; 0 = Otherwise; OCEANIA is a dummy variable: 1 = If the country is in Oceania; 0 = Otherwise *P value < 0.1; **P value < 0.05; ***P value < 0.01 who also demonstrate a positive association between board size and firm value. This evidence shows that board size in fact matters for companies since larger boards will positively affect firm value and, therefore, it will imply a higher wealth for shareholders. Boards with a high number of directors will provide human and social capital to the board, given that they might bring diverse past expertise, skills and professional backgrounds, as well as the knowledge of similar industries. This will be beneficial for the decision-making process. Additionally, it will be more complicated for CEOs or managers to pressure directors on a larger board. Thus, board size plays a significant role in supervising the management team and giving advice and counsel, in line with Klein (1998) or Singh et al. (2018), and this will have a positive impact on firm performance. Model 2 explores how board independence affects firm value. The variable proportion of independent directors on boards provides a positive sign, as expected, and is significant. As a result, the second hypothesis is accepted. This finding suggests that board independence is positively associated with firm performance. Authors such as Mak and Kusnadi (2005), Palaniappan (2017b) and Palmberg (2015), among others, also find that the presence of independent directors on boards influences firm value. Our evidence shows that independent directors on boards may perform a relevant monitoring role by mitigating the power that managers and CEOs might have on the whole board because they are outsiders, external or non-executive directors. Independent directors usually defend the rights of shareholders, particularly minorities, given that they tend to be independent from managers and can better supervise them. In this regard, and according to Haniffa and Cooke (2002), independent directors "are seen as a checks and balances" tool in improving board efficiency. Independent directors will be more likely to align shareholders and managers' interests and needs. Fama and Jensen (1983) argue the importance of the role performed by independent directors, because they these directors are available to provide valuable experience and professional background as well as ties or connections with external organizations, positively affecting corporate value. Thus, board independence will improve firm performance and safeguard shareholders' wealth, interests and demands (Gabrielsson 2007).
In Model 3, we analyse the relationship between CEO duality and firm value. The variable CEO duality (CEO_DUALITY) shows a positive sign, contrary to our expectations, and is statistically significant. Hence, according to this finding, the third hypothesis has to be rejected. This evidence suggests that when the roles of Chairperson of the board and CEO of the company are combined, firm value is enhanced, in contrast with the findings of most past research (e.g., Jermias 2007; Jermias and Gani 2014), which reports that CEO duality results in poorer firm performance. Miller and Yang (2015), however, also show that CEO duality has a positive effect on corporate value. The arguments defining this positive relationship are centered on the fact that when the CEO and the Chairperson are the same person, power and control will be unified, thus avoiding the confusion of having two public spokespersons in the company, while the transmission of information will be improved and the coordination costs reduced, thus improving firm performance (Corbetta and Salvato 2004;Song et al. 2006). CEO duality implies a clear and higher authority in firms, which may drive companies to the same strategic direction addressing conflicts and lack of coordination between senior managers and stakeholders, mainly shareholders. This will result in a more efficient decision-making process and could lead to greater firm performance. The concentration of power in the CEO might be beneficial for firms to survive and be successful, in that the leadership will be more effective and have more legitimacy (e.g., Dalton et al. 1998). The positive impact of CEO duality on firm performance may also be explained by the fact that CEOs have specific knowledge of the firm, accumulating a lot of the firm's internal knowledge and the knowledge of the uncertain environment and context in which firms operate, in addition to being aware of the competitive advantages and disadvantages of the firm and its competitors, since they daily manage firms. This could lead to better firm performance.
Model 4 examines the relationship between board gender diversity and firm performance. The variable proportion of female directors on boards (FEM_DIR) offers a positive sign, as predicted, and is significant and, therefore, the fourth hypothesis has to be accepted. This result shows that the presence of women directors on boards is positively associated with firm value (e.g. Liu et al. 2013;Kılıç and Kuzey 2016;Terjesen et al. 2016), hence these studies provide solid evidence that having women on the board can lead to better firm performance.
Our finding suggests that the incorporation of a minority group on boards, such as female directors, may help represent and protect shareholders' demands and interests (Webb 2004) by showing objective and critical judgments in the decision-making process, which might result in a higher firm performance. Better firm performance can also be due to the greater board independence that female directors on boards provide since directors with more conventional backgrounds do not usually raise the questions that may come from directors of a different ethnicity, cultural background or gender, as Carter et al. (2003) argue. Likewise, females on boards may contribute a diversity of opinions and views to board discussions, leading to more responsive policies, increasing more involvement in the decision-making process and encouraging more democratic decisions (Nielsen and Huse 2010;Webb 2004). All these qualities of female behaviour may also align the interests of all shareholders and, as a consequence, it can be assumed that female directors may help to avoid conflicts and make decisions that benefit shareholders, which might enhance corporate value.
Finally, in Model 5, we analyze the impact of board compensation on firm value. The variable board compensation (B_COMPENSATION) shows a positive sign, in line with our predictions, but it is insignificant. According to this result, the fifth hypothesis has to be rejected, suggesting that well-paid board members do not have any effect on corporate value. In other words, board members with higher compensation will not have more incentives to monitor managers, which suggests that wellrewarded board members will not result in more supervising efforts and will not be more likely to converge their interests with those of shareholders. Our evidence does not support the agent-principal or agency approach, which argues the positive effect of board remuneration on company value.
Model 6, where all independent variables are jointly estimated, reports the same results as all the independent variables are individually estimated. Thus, we reach the same conclusion both when independent variables are individually estimated, and when they are jointly estimated.
Regarding the control variables, board meetings, size and capital intensity show the expected sign and are statistically significant. Thus, large firms whose boards tend to meet several times each year are positively correlated with firm value, while increases in capital intensity reduce firm performance. Furthermore, firms operating in countries characterized by high ownership concentration underperform. Contrary to our predictions, firms with high levels of operating performance also tend to reduce firm performance. Additionally, all the sectors analyzed exhibit a negative sign and are statistically significant, showing that firms operating in all the sectors are more likely to underperform. The previous evidence regarding the repercussion of sector of activity on firm performance is not conclusive. In this regard, Jermias and Gani (2014), for a sample of US companies that they obtained from Compustat S&P 500, conclude that the activity sectors do not have an effect when analyzing the impact of board characteristics on firm performance. For Spanish companies, within the context of the influence of corporate governance on firm performance as well, Pucheta-Martínez (2015) obtains a negative but not significant relationship for some sectors, and a negative and significant relationship for others, such as the oil and energy sectors. Concerning the geographic zone, our findings show that firms operating in Europe and Latin America perform better, and in Oceania worse, in contrast to companies operating in North America. Moreover, firms operating in Asia with large boards and well-paid directors tend to perform worse than companies operating in North America. Finally, the remainder of the control variables are not significant.
Additionally, endogeneity concerns have been also addressed. Although it is expected that board attributes (board size, board independence, CEO duality, board gender diversity and board compensation) affect firm value, it is also possible that over-performing companies will make up boards with certain characteristics regarding the board attributes explored in this analysis. To check for endogeneity, we estimated our models by lagging the independent variables, consistent with Ozkan (2007), who supports the idea that endogeneity problems might be reduced by lagging explanatory variables (independent variables). Our results, provide in Table 8, are in line with the findings shown in our base models when the independent variables are individually estimated. However, when all independent variables are jointly estimated, the board compensation variable exhibits a negative sign, contrary to our predictions, and is statistically significant, in comparison to the finding presented in Table 7, where it was insignificant.

Robustness analysis
We have conducted a robustness check in order to assess if our findings change in the presence of an alternative measure for firm performance. Concretely, we use as proxy variable for firm performance the ROE ratio (return on equity), measured as the ratio of net profit over total equity, following Matolcsy andWright (2011) andLi et al. (2015). In Table 9, we offer the results of all our models, namely, estimating all     Q_TOBIN is the market capitalization of common stock + book value liabilities divided by the book value of total assets; B_SIZE is the total number of directors on boards; B_INDEP is the proportion of independent directors on boards = total number of independent on boards/total number of directors on boards; CEO_DUALITY is a dummy variable that takes the value 1 if the same person serves simultaneously as CEO and President of the board and 0, otherwise; FEM_DIR is the proportion of female directors on boards = total number of female directors on boards/total number of directors no boards; B_COMPENSATION is the log of board compensation; B_MEET-ING is the number of meetings held by boards every year; SIZE is the log of total assets; LEVERAGE is debt over total assets; CAPITAL_INTENSITY is the ratio of long-term or fixed assets over total assets; OWNSHIP_CON is the average percentage of shares held by the three largest shareholders in the ten largest firms in each country; DUAL_CLASS is a dummy variable that takes the value 1 if firms have dual-class stocks (class A/B or registered/bearer shares) and 0, otherwise; OPERATING_ PERFORM is the ratio between operating income and total assets; BASIC MATERIALS is a dummy variable: 1 = Basic Materials; 0 = Otherwise; CONSUMER CYCLI-CAL is a dummy variable: 1 = Consumer Cyclical; 0 = Otherwise; CONSUMER NON-CYCLICAL is a dummy variable: 1 = Consumer Non-Cyclical; 0 = Otherwise; HEALTHCARE is a dummy variable: 1 = Healthcare; 0 = Otherwise; INDUSTRIALS is a dummy variable: 1 = Industrial; 0 = Otherwise; TECHNOLOGY is a dummy variable: 1 = Technology; 0 = Otherwise; TELECOMMUNICATION SERVICES is a dummy variable: 1 = Telecommunication Services; 0 = Otherwise and UTILITIES is a dummy variable: 1 = Utilities; 0 = Otherwise. AFRICA is a dummy variable: 1 = If the country is in Africa; 0 = Otherwise; ASIA is a dummy variable: 1 = If the country is in Asia; 0 = Otherwise; EUROPE is a dummy variable: 1 = If the country is in Europe; 0 = Otherwise; LATINAMERICA is a dummy variable: 1 = If the country is in Latin America; 0 = Otherwise; NORTHAMERICA is a dummy variable: 1 = If the country is in North America; 0 = Otherwise; OCEANIA is a dummy variable: 1 = If the country is in Oceania; 0 = Otherwise *P value < 0.1; **P value < 0.05; ***P value < 0.01 Do board characteristics drive firm performance? An…   ROE ratio (return on equity), measured as the ratio of net profit over total equity; B_SIZE is the total number of directors on boards; B_INDEP is the proportion of independent directors on boards = total number of independent on boards/total number of directors on boards; CEO_DUALITY is a dummy variable that takes the value 1 if the same person serves simultaneously as CEO and President of the board and 0, otherwise; FEM_DIR is the proportion of female directors on boards = total number of female directors on boards/total number of directors no boards; B_COMPENSATION is the log of board compensation; B_MEETING is the number of meetings held by boards every year; SIZE is the log of total assets; LEVERAGE is debt over total assets; CAPITAL_INTENSITY is the ratio of long-term or fixed assets over total assets; OWNSHIP_CON is the average percentage of shares held by the three largest shareholders in the ten largest firms in each country; DUAL_CLASS is a dummy variable that takes the value 1 if firms have dual-class stocks (class A/B or registered/bearer shares) and 0, otherwise; OPERATING_PERFORM is the ratio between operating income and total assets; BASIC MATERIALS is a dummy variable: 1 = Basic Materials; 0 = Otherwise; CONSUMER CYCLICAL is a dummy variable: 1 = Consumer Cyclical; 0 = Otherwise; CONSUMER NON-CYCLICAL is a dummy variable: 1 = Consumer Non-Cyclical; 0 = Otherwise; HEALTHCARE is a dummy variable: 1 = Healthcare; 0 = Otherwise; INDUSTRIALS is a dummy variable: 1 = Industrial; 0 = Otherwise; TECHNOLOGY is a dummy variable: 1 = Technology; 0 = Otherwise; TELECOMMUNICATION SERVICES is a dummy variable: 1 = Telecommunication Services; 0 = Otherwise and UTILITIES is a dummy variable: 1 = Utilities; 0 = Otherwise. AFRICA is a dummy variable: 1 = If the country is in Africa; 0 = Otherwise; ASIA is a dummy variable: 1 = If the country is in Asia; 0 = Otherwise; EUROPE is a dummy variable: 1 = If the country is in Europe; 0 = Otherwise; LATINAMERICA is a dummy variable: 1 = If the country is in Latin America; 0 = Otherwise; NORTH-AMERICA is a dummy variable: 1 = If the country is in North America; 0 = Otherwise; OCEANIA is a dummy variable: 1 = If the country is in Oceania; 0 = Otherwise *P value < 0.1; **P value < 0.05; ***P value < 0.01 independent variables individually and all together in one model. As can be appreciated, the findings corroborate the result of our baseline models. In addition, the board compensation variable shows a positive sign, as expected, and is significant, both when all independent variables are individually and jointly estimated, while in our baseline model, it was insignificant. Thus, ROE can be considered a robust measure for firm performance since it not only supports our core findings, but also the sign and significance of the board compensation variable.

Concluding remarks
In this research, we aim at exploring the relationship between certain corporate governance mechanisms and firm performance. Particularly, we analyze how board size, board independence, CEO duality, board gender diversity and board compensation affect firm performance using a sample of international firms. Drawing on agency and resource dependence theories, we hypothesized that all these board attributes are positively correlated with firm value, except CEO duality, for which we predicted a negative effect. Our findings report, as expected, that large boards with a high presence of independent and female directors tend to increase firm value, supporting agency and resource dependence approaches, and in line with the monitoring or supervising hypothesis. On the other hand, the results show that CEO duality, contrary to our predictions, has a positive impact on firm performance, and board compensation does not affect firm performance, also in contrast to our expectations. Thus, our evidence supports all the hypotheses posited, except those relating to CEO duality and board compensation. Certain board attributes, such as board size, board independence and board gender diversity are more likely to improve shareholder wealth and thus, shareholders might take them into account, especially potential investors. CEO duality implies a concentration of power in the CEO and, as a consequence, it was hypothesized that CEO duality might have a negative effect on firm performance since CEOs will be more interested in satisfying their needs and interests at the expense of shareholders' interests. However, our results are not consistent with this idea and CEOs can make an effort to enhance corporate value, given that this may be beneficial for their success, legitimation and reputation, among other things. This research has several implications. Firstly, the findings of this investigation focused on a sample of international firms suggest that particular board attributes such as board size, independent directors, CEO duality, female directors and board remuneration are relevant corporate governance mechanisms for improving firm performance. Thus, this evidence may be useful for managers who are willing to enhance their reputation and to gain legitimation by shareholders, stakeholders and society. Practitioners with a shareholders' orientation should make up boards with these effective corporate governance mechanisms because they will be more likely to over perform firms. Secondly, our results also support the thesis that CEO duality does not always have a negative effect on business decisions, particularly firm performance. Therefore, this evidence should lead policymakers to consider the possibility of not reinforcing the separation of the figure of CEO and Chairperson of the board. Regulatory bodies have established recommendations or laws supporting the thesis of the positive benefits of different individuals holding the positions of CEO and Chairperson of the board, but our findings are contrary to this idea, since CEO duality improves firm performance. Perhaps, it would be more prudent to assess the impact of CEO duality for each business decision in particular, and according to this evidence, regulate in a more precise way the dual role of Chairperson of the board and CEO. Thirdly, current shareholders and potential investors may benefit from our findings because they could be useful for making decisions about investments and about the most appropriate board structure for increasing corporate value. Most shareholders and potential investors are interested in increasing the probability of a return on their investments, in the short or long term, depending on their orientations and, therefore, they will be willing to form boards, which encourage better firm value. Finally, the findings of this study also suggest that more research is needed. Researchers may extend our study by examining the association between other corporate governance mechanisms and firm performance, which will be useful in shedding new light on what board composition or board structure is most suitable to make a well-performing company.
This research also is subject to several limitations. Firstly, it is possible that we have disregarded some factors that potentially affect firm performance, according to prior theoretical and empirical evidence. Secondly, we have explored all countries for which it has been possible to collect all data necessary for conducting this research. Future research extending the sample scope to more countries should be carried out. Finally, we use as dependent variable a market performance measure (Tobin's Q), which limits this study to the scope of listed companies. This analysis may be extended to non-listed firms, particularly small and medium-sized enterprises, using other performance measures different from market ones.
Future lines of research can be derived from this investigation. Firstly, it would be interesting to analyze the association between board attributes and firm value, differentiating between developing and emerging countries. Secondly, we would also encourage scholars to explore which corporate governance mechanisms are most significant for enhancing corporate value in the banking industry, since our sample excludes this sector.