MSc Thesis Finance

(1)

MSc Thesis Finance

Gender Diversity in the Board of Directors and Firm

Financial Performance; going beyond the surface

Author: Jorn Meuleman – S2906708 Supervisor: prof. N. Hermes

Academic Year: 2019-2020 Wordcount: 11.006

Abstract: This study examines some of the characteristics associated with female board members, and empirically aims to relate these to firm performance going beyond the general measures of gender diversity. In addition, this analysis aims to circumvent endogeneity issues by estimating fixed-effects regressions and applying the propensity scoring method. The results covering a sample of 475 firms for the years 2016-2018 show that female board members are more highly educated and have more experience compared to male board members. Moreover, there appears to be no relationship between the addition of females on boards and firm performance through both the hypothesized increased levels of education and experience associated with female board members.

(2)

2

1. Introduction

Ever since scandals such as Enron became public, attention for the governance of firms has increased in intensity. These scandals led to many governance reforms, such as the

implementation of the Sarbanes-Oxley act in 2002. In turn, this increased the amount of research devoted to the board of directors’ composition as the board of directors is a critical element of a firm’s governance (Fama and Jensen, 1983; Singh, Terjesen and Vinnicombe, 2008). One aspect which is bound to impact the composition of the board of directors, being gender diversity, has become the subject of attention as well. Many countries in Europa and the state of California in the USA have imposed regulations or soft quotas to obligate publicly listed companies to employ females in their board of directors e.g. (Fuhrmans, 2018; Staley, 2016). These are imposed in order to ensure female participation in the male dominated world of the board of directors, with approximately only 10% of board members being female in the US. It is paramount to understand how the processes within the board of directors change due to the increased participation of females ensuring that firms remain governed appropriately.

This ‘old boys’ network of top male executives is documented to, in general, be biased against women. A survey among top executives and board members documented that the male

interviewees held the belief that women lack both the qualifications and experience to be on the board of directors (Ragins, Townsend & Mattis, 1998). Also, females reported that in order to stand out, they have to work harder than their male colleagues for top management positions. One female executive was even quoted saying, “Do the best you can at every assignment no matter how trivial. Always go the extra mile” (Ragins et al., 1998, p.30). This shows the difficulty for a woman to penetrate the ‘established order’ of male executives. The females who do manage to achieve a board position have to meet high standards and hence may add a lot to the achievement of the tasks of the Board of Directors due to their higher than average capabilities.

(4)

4

essential to ensure that upper management interests are aligned with those of shareholders (Fama and Jensen, 1983).

Moreover, based on resource dependence theory, females tend to have different types as human capital such as experience and educational background as compared to males.

Diversity is thus likely to lead to a broader set of human capital (Carter, D’Souza, Simkins & Simpson, 2010; Terjesen, Sealy & Singh, 2009), which yields more resources for the firm. The latter is vital in aiding the firms' long-term survival (Hillman & Dalziel, 2003), as having more resources at its disposal is likely to have a positive effect on firm performance. Thus gender diversity within the board of directors is likely to impact firm performance positively according to theory. Analysis conducted aiming to examine this relationship empirically however finds conflicting results, therefore the exact effect of gender diversity remains unclear.

This analysis will differ from preceding research by not relating a general measure of gender diversity to firm performance. But instead examine the characteristics associated with female directors such as the relatively higher education levels compared to male directors (Burgess & Tharenou, 2002; Hillman, Cannella & Harris,2002; Singh et al., 2008) and a higher

experience level through the possession of on average more board positions simultaneously. These characteristics in turn will be related to firm performance utilizing a proxy depicting the average level of education/experience for females relative to that of men. Aiming to provide conclusive results regarding the effect of gender diversity on firm performance. The research question is as follows:

What characterizes female board members, and how do these characteristics impact the functioning of the board and ultimately firm performance.

The next section will review the literature to date in order to build a theoretical case for the relation between the functioning of the board of directors and firm performance, and how gender diversity may improve the functioning board of directors. Secondly, it will discuss the data gathered, variables constructed, and the research methods applied to examine the

(5)

5

2. Literature Review and Hypotheses

2.1. Corporate Governance and the Board of Directors

The debate remains as to what exactly constitutes corporate governance as it entails many different aspects and mechanisms and is challenging to define. There is one description of corporate governance, which was proposed based on an extensive survey of the corporate governance literature. This definition by Shleifer and Vishny (1997) is widely accepted and describes that “corporate governance deals with the ways in which suppliers of finance ensure that they get a return on their investment.” Even though the authors take the agency

perspective, the definition still is an accurate description of the function of corporate

governance mechanisms nowadays. It describes corporate governance mechanisms as highly important means of control, ensuring that the interest of management are aligned with those of parties related to the firm such as its shareholders and stakeholders allowing them to get a return on their investments (Campbell and Minguez-Vera, 2008)

One of these corporate governance mechanisms, being the board of directors, is expected to be crucial based on the separation of ownership and control (Baysinger and Butler; Fama and Jensen, 1983; Hillman and Dalziel, 2003). In practice, the board of directors performs two tasks being monitoring and providing resources (Hillman and Dalziel, 2003). The performing of these tasks affects firm performance according to two dominant theories discussed below. Firstly, according to agency theory, the board of directors is expected to impact firm

performance due to their task of monitoring the upper management. By monitoring appropriately, the board of directors ensures that a manager does not pursue his/her self-interest, which would incur agency costs, but instead takes actions that maximize firm value, A board of directors can for instance do this by setting of compensation for the upper

management and by replacing the management of a firm (Carter, Simkins and Simpson, 2003; Fama and Jensen, 1983; Hillman and Dalziel, 2003).

Secondly, according to resource-dependence theory, a board of directors is expected to add value to the organization through their provision of resources (Boyd, 1990; Hillman, Withers and Collins, 2009). Each director, when taking his/her seat at the table, brings resources to the firm where resources are “Anything that could be thought of as a strength or weakness of a given firm.” (Wernerfelt, 1984). The four primary resources a director can bring are advice and counselling, legitimacy, channels for communicating information between the external organizations and the firm, and preferential access to commitments or support from important elements outside the firm (Hillman and Dalziel, 2003). These resources impact firm

performance as these resources “help reduce dependency between the firm and external contingencies, diminish uncertainty, lower transaction costs and ultimately aid in the survival of the firm” (Hillman and Dalziel, 2003, p.386).

(6)

6

authors researched the optimal composition of a board of directors. Gender diversity, which will be the main focus of this paper, is one of these factors for which the influence on board composition and ultimately firm performance is discussed in the next section.

2.2. (Gender) Diversity and Firm Financial Performance

Diversity in itself can be understood in many ways, as conceptualized by Harrison and Klein (2007). The authors recognize three different types of diversity based on particular aspects. They argue that when discussing diversity as a qualitative difference between groups, the concept ‘variety’ is most applicable. Henceforth when mentioning diversity in this thesis, variety is implied.

Harrison and Klein (2007) argue that variety is associated with the increased performance of units of people who work together as it broadens the pool of behavioral and cognitive abilities at disposal. Similar arguments are proposed by Robinson and Dechant (1997) who aimed to create a business case for diversity. The authors argued that when implemented correctly, diversity might have upside potential possibly leading to more business growth. As for one, having a diversified workforce leads to a broader understanding of the marketplace and how potential customers can be reached. Secondly, diversity is argued to increase creativity and innovation as cognitive abilities and beliefs vary along with demographic variables. This increase in creativity and innovation is likely to lead to higher problem-solving quality

(Robinson and Dechant, 1997). Besides, the human capital theory argues that every individual has unique human capital in the form of skills and resources gathered; examples include for instance, education and experience. Diversity is likely to increase the different types of human capital a firm has its disposal and therefore is beneficial for the firm’s survival (Terjesen et al., 2009).

Gender diversity within the board of directors is explicitly argued to impact firm performance as well. For one, a gender diverse board is likely to have a broader set of cognitive functions and beliefs, possibly leading to more creative and innovative solutions to business-related issues. Besides, gender diversity enhances corporate leadership effectiveness (Robinson and Dechant, 1997) and leads to more efficient problem solving as a variety of different

(7)

7

implying fewer connections with members of upper management (Carter et al. 2003). The latter inference is highly relevant for the agency theory, as an independent monitor is

associated with better monitoring capabilities and henceforth better firm performance (Fama and Jenssen, 1983; Hillman and Dalziel, 2003). Lastly, aside from economic benefits, the argument for gender diversity and hence female participation is ethical. Both genders should be treated and judged equally for acceptance to positions and receive the same opportunities. Adams and Ferreira (2009) directly examined the effect on gender diversity on the

governance of boards and find that overall gender-diverse boards are stricter monitors, which is likely to be due to females being more consistent in attending board meetings and them being part of the monitoring committee within the board. Also, they find that a CEO’s pay of firms that have a gender diverse board has a more significant fraction of equity compensation, which implies that interests are more aligned between upper management and shareholders. When directly examining the effect of gender diversity on firm performance the authors find a negative effect after controlling for firm-specific factors. The authors argue that this is

because firms with good governance do not necessarily benefit from increased gender diversity, where firms with bad governance do. In general results within the field are

somewhat mixed, where some find a significant positive effect (Campbell and Minguez-Vera, 2008; Carter et al., 2003; Dsezo and Ross, 2012; Erhardt, Werber and Shrader, 2003) and others (Carter et al., 2010; Rose, 2007; Shrader, Blackburn, and Iles, 1997) find no relation between gender diversity and firm performance. In addition, as pointed out by Adams and Ferrreira (2009), endogeneity is a significant issue within the literature that could potentially influence the results. Henceforth, we turn to two meta-analyses conducted who aimed to reconcile the conflicting results.

(8)

8

differences may have on firm performance.

2.3.

What Characterizes Female Board Members?

Not much of the literature discusses the differences between the male and female board members and how ultimately these differences may be the determinants of the effect of gender diversity on firm performance. Instead, a general measure diversity is taken,

neglecting all these underlying associated characteristics associated with gender diversity, and related to firm performance. Henceforth this analysis will take a different approach by going beyond the general diversity measure and examining the differences in characteristics

between male and female board members and empirically relating these to firm performance. The common rationale applied to determine the relevance of these characteristics is the resource-dependence theory. As explained, resource-dependence theory views individuals as providers of human capital, such as knowledge through education but also social capital such as ties to providers of resources outside the firm. Each of these resources is vital in aiding the firm’s long-term survival (Hillman and Dalziel, 2003). Henceforth it is highly relevant to understand what resources female board members bring to the board and how this may potentially impact board- and ultimately firm performance (Hillman et al., 2002). This analysis will focus on two of the most documented characteristics being education and the experience of female board members.

Hillman et al. (2002) found significant differences between men and women for the board of Fortune 1000 firms. For one, they argue that for females it is more difficult to be promoted internally, as they have to jump through more ‘hoops’ due to sex-based bias. Therefore, in order to signal their capabilities females, have to find another way to differentiate themselves from their male counterparts. The authors claim that therefore, it is likely that women hold relatively more advanced degrees compared to their male colleagues and find support for this hypothesis. Similar results were found by Burgess and Tharenou (2002) and Singh et al. (2008), who found that female board members have significantly higher education levels than do male board members. Therefore, we expect that women tend to have higher education levels, hypothesis 1 follows.

(9)

9

Hypothesis 1: Female board members are more likely to hold advanced degrees as male board members

Hypothesis 2: Females on a board of directors have a positive effect on firm performance through higher education levels.

Another characteristic associated with female board members is the number of board

positions they hold, which this analysis views as a proxy of experience. Hillman et al. (2002) argue that firms usually select board members based on previous experience and directorships as these directors encompass more resources through ties with other firms and have proven themselves to be fit for the task reducing any uncertainty. Besides, search costs for new directors are reduced when selecting from the current pool of directors. Therefore, directors are likely to gain multiple directorships after getting the first position. Due to the firm’s tendency to select board members from current directors, and the given that females are a minority within the pool of current directors, chances of a female director being selected for multiple board positions are expected to be higher than that of men. Therefore, we

hypothesize, following Hillman et al. (2002), that female board members are likely to serve on more boards simultaneously than males directors will.

The hypothesized higher experience of females through more directorships is likely to affect firm performance. Due to multiple directorships, a board member has more resources at is disposal such as ties to other firms and connections with other board members. These additional resources impact firm performance according to resource-dependence theory (Hillman et al., 2002; Singh et al. 2008;). Additionally, the increased experience suggests that female board members have relatively more experience with monitoring upper management, likely reducing the agency costs incurred which is beneficial for firm performance according to agency theory. Overall female board members likely have a positive effect on firm

performance through them increasing experience within boards, hypothesis 4 follows. H3: Females are likely to serve on more boards of directors simultaneously than men. H4: Females on a board of directors have a positive effect on firm performance due to more

experience.

3. Data and Methodology

3.1. Sample

(10)

10

United States as the governance system employed is market-based. Governing the firms, in this system, is attributed to the market which theoretically employs board members to govern on their behalf, adding to the significance of the task of these directors and the likelihood of impacting firm performance by performing well. Lastly, there is ample data available for most variables for US-based firms through data vendors such as BoardEx and Thomson Reuters Eikon.

The sample consists of all the S&P500 firms for the year 2016 up to and including 2018. First of all, the S&P500 is chosen since its constituents encompass many different industries and these firms represent a large amount of the total market capitalization of all US firms, thereby making the result generalizable for the economy as a whole. Secondly, the years 2016-2018 are chosen as for those years most data is available yielding the largest sample possible. This panel-data approach is in line with most papers on gender diversity (e.g., Adams and Ferreira, 2009; Campbell and Minguez-Vera, 2008; Carter et al. 2010; Lückerath-Rovers, 2013; Smith, Smith and Verner, 2006), and has its benefits over cross-sectional analysis with regards to accounting for endogeneity as will be discussed in section 3.6. Next to data

availability, the choice for three consecutive years is because board characteristics only change slowly over time, as directors tend to stay on seats for at least a year (Bhagat and Black, 2001). By including three separate years, we expect to have some changes to the composition allowing for appropriate analysis.

3.2. Dependent Variables

The dependent variable of interest is the firm financial performance of S&P500 firms and follows the assumption that the focus of companies should be the maximization of

shareholder value being the most socially efficient aim (Jensen, 2010). The debate within literature remains as to how the maximization of shareholder value can be appropriately measured. Two types of measures commonly applied are market-based measures that focus on the long-term financial performance of firms and accounting-based measures who mostly reflect short-term past financial performance. Both types of measures are widely accepted within the literature as proxies for firm financial performance and are used interchangeably (Gentry and Shen, 2010). Turning to empirical research on gender diversity many different measures are applied. Most papers either use both ROA, as an accounting-based measure and Tobin’s Q, as a market-based measure, or one of these. Therefore this analysis will include both measures mentioned above to allow for comparability among results within the field. For the approximation of Tobin’s Q, the commonly used approach put forth by Chung and Pruit (1994) is applied. The authors proved that approximating Tobin’s Q by dividing the market value of securities by the book value of total assets is an accurate representation of the original, data-wise, more complicated measure (Carter et al. 2010). In this ratio the

(11)

11

represents the value on financial markets. A Tobin’s Q higher than 1 implies that a firm possesses growth opportunities (Rose, 2007). In turn this implies that with a ratio lower than 1 a firm can use resources more efficiently to increase its market value (Campbell and

Minguez-Vera, 2008; Carter et al., 2010).

Return on Assets (ROA) can be calculated by dividing the net income with the book value of total assets and is a measure of the firm’s ability to generate revenues in excess of the actual expenses a firm incurs for the set of assets employed in its operations i.e. shareholders revenue. (Carter et al., 2010) Both measures are calculated using data extracted from the Thomson Reuters Eikon database.

3.3. Explanatory Variables

As the starting point for constructing the variable measuring education, individual education levels were subtracted for all board members within the sample from BoardEx. The

categorization assigned to each of these individual qualifications was determined by following the approach of Hillman et al. (2002), who defined three levels of education ranked from lowest to highest being Undergraduate Master and lastly, Doctoral Degrees. The same method is applied for the sample of this analysis with one exception. A few directors had an Associate Degree, being a certificate which is just below an undergraduate (bachelors) degree. For the sake of this analysis, it is set equal in terms of level. The following assignment for the sample of individual directors applies.

1. Undergraduate/Associate Degree 2. Master Degree

3. Doctoral Degree

For directors who had attained more than one qualification, the highest level of education from the above categorization attained is leading. The individual level-data on education that followed is used to test for differences in education between female and male board members corresponding to hypothesis 1

To construct a variable required to test hypothesis 2, for each observation the average

education level for males and that of females within the board is calculated. When the average education level of female board members is higher than that of males within the same board, for that specific observation a 1 is assigned and otherwise a 0. In essence, this dummy variable allows us to determine if boards were females raise the average level of education, with these boards taking a 1 on the dummy variable, are associated with increased

performance. The sign in the regression associated with the Education Dummy is expected to be positive according to hypothesis 2.

(12)

12

number of listed board positions a director currently has is extracted from BoardEx for each observation. Same as for education the individual-level data on this proxy for experience can be used to test for differences in experience between genders corresponding to hypothesis 3. Secondly, the averages are calculated for male and female board members within the board and a 1 is assigned if this average for females is higher than that of males within the same boards. The Experience Dummy can be interpreted in a similar manner as the education dummy. The sign associated with this experience dummy is expected to be positive according to hypothesis 4.

3.4. Control Variables

Control variables are added to the analysis to account for factors impacting firm financial performance, next to the explanatory variables. Firstly, board size is a widely used control variable within the literature. Some authors argue that firms profit from larger boards of directors because these are characterized by a larger pool of external linkages and expertise and benefit from enhanced decision-making, ultimately improving firm performance (Carter et al. 2010, Labelle, Francoeur and Lakhal, 2015; Pearce and Zahra, 1992). On the contrary, some authors argue that increased board size harms firm performance due to a decrease in social cohesion and more significant potential for conflict, both leading to increased agency costs (Carter et al. 2010; Labelle et al., 2015). Directly examining the effect of board size on firm performance yields opposing results, where Jackling and Johl (2009) find a significant and positive effect of board size on firm performance and O’Connel and Cramer (2010) find a negative and significant effect. Regardless the effect of board size which is somewhat

ambiguous, it is an important factor to account for as it is likely to be influential in two aspects. Firstly, it is likely to affect the degree of gender diversity within the board of directors as the more seats available the more likely it is that there are some female board members (Carter et al., 2010) and secondly as argued it is likely to affect the financial performance of a firm. The Board Size is subtracted from BoardEx, representing the total number of board members, and will be included in the analysis.

(13)

13

The debt level is included in the analysis as a third control variable. On the one hand, it is argued that high debt levels pre-commit a managers to generate cashflows as debt payments have to be made, where these generated cashflows positively impact firm performance (Jackling and Johl, 2009). Nevertheless, it is argued that the debt level of a firm proxies for the risk associated with its operations (Labelle et al. 2015), and that firms who hold relatively larger levels of cash are better able to afford the regular debt payments (Dezsö and Ross, 2012). This implies that firms with higher debt levels, both have more risk associated with operations and have smaller amounts of cash at its disposal, which is likely to be negatively associated with firm performance (Campbell and Minguez-Vera, 2008). When directly examining the proposed relation, the debt level is proven to be significantly and negatively related to firm performance (Dezsö and Ross, 2012; Jackling and Johl, 2009) corroborating the importance of the inclusion of a measure of the debt level of firms in the analysis. The data required is subtracted from Thomson Reuters Eikon and the Debt Level is manually calculated by dividing the total debt with the common equity

The fourth control variable added will be a measure of gender diversity within the board. This is due to the fact that we assume that in order for the proposed relations to hold it has to take place in a gender diverse environment. For example, if there is only one female out of in total twelve board members, the individual female’s impact on firm performance through gender diversity is not likely to be large. We expect that for more female representation on the boards relative to that of males the effect becomes larger and henceforth, a measure of Gender Diversity will be included in the analysis .This measure is subtracted from BoardEx and represents the number of male board members divided by the total number of board members. Lastly, research shows that industries are significant in explaining the presence of females on boards (Carter et al. 2010). First off, it is expected that companies in industries that are characterized as labor-intensive and service-oriented have a relatively larger representation of females within the workforce and consequently their boards (Brammer, Millington and Pavelin, 2007; Farrel and Hersch, 2005). More precisely it is shown that females are more likely to be present on boards within the financial sector, which is due to the fact that the pool of women working in these sectors is much higher improving the chances of females reaching upper management positions (Carter et al. 2003; Lückerath-Rovers, 2013). Therefore, the industry is likely to explain part of the gender diversity within boards of directors and will be included in the analysis. The General Industry Classification of a firm is subtracted from Thomson Reuters Eikon where the following categorization of industries applies:

1 – Industrial 2 – Utility

3 – Transportation 4 – Bank/Savings and Loan 5 – Insurance 6 – Other Financials

(14)

14

3.5. Data Analysis

Table 1 depicts the descriptive statistics for the untransformed values of the variables. Thirty firms were removed entirely due to data unavailability with regards to both Tobin’s Q and ROA, ultimately yielding a sample of 475 firms included in the analysis. Besides, some data were missing for differing years and firms yielding an unbalanced panel dataset.

Table 1: Descriptive Statistics:

Tobin's Q and ROA are an approximation of Tobin’s Q and ROA, respectively. The Education and Experience Dummy is a variable that takes a value of 1 if the average value of the associated characteristic for females is above that of the male members within the board and 0 otherwise. Board Size depicts the number of directors on a board. Consequently, Debt Level is defined as the debt over equity, firm size is defined as the total assets of a firm in millions and lastly the gender ratio is the

ratio of male directors to the total number of directors on a board.

Tobin’s Q has an average value of 2.44 implying that firms are on average worth 2.44 times of their replacement value of assets. In addition, the maximum value for Tobin’s Q is rather high, showing the firms in the sample are quite successful. With regards to ROA, there seem to be no confounding issues with its distribution over firms, depicting an average of 6.578% income generated over the assets in place. The summary statistics for both ROA and Tobin’s Q are relatively similar to those of Adams and Ferreira (2009) which report an average value for Tobin’s Q of 2.09 and a maximum value of 77.64. The average for ROA is somewhat higher than the values reported by Adams and Ferreira (2009) however the maximum and minimum values are less of the sample are less extreme then the values reported by these authors, adding to the assumed validity of the data used in the analysis.

Examining the explanatory variables of interest it can be seen that for 58.7% (56.8%) of the boards the average education (experience) level of female board members was higher than that of male board members. The data shows that both groups, being the board were females have higher(more) education (experience) and boards were the opposite holds, are both

Variable Number of

observations

Mean Standard Deviation

Median Min Max

(15)

15

represented by at least 40% of the total observations allowing for proper analysis. Lastly, Tobin’s Q, Firm Size and Board Size are transformed to their natural logarithm due to concerns regarding normality.

Table 2: Correlation matrix

The digits represent the pairwise correlation coefficients and *, **, *** indicate significance at the 5%, 1% and 0.1% levels, respectively. The variables definitions are similar to those described

in table 1.

One of the essential assumptions of regression analysis is that the explanatory variables included in the regressions are not multicollinear, as the latter could potentially bias the results. Potential multicollinearity can be approximated using correlations between variables (Brooks, 2014). Examining the correlation matrix in table 2, a high correlation is present between Tobin's Q and ROA. This is expected as Tobin’s Q and ROA are both measures of financial performance. Regardless this is not an issue as multicollinearity only potentially exists between the explanatory variables and the variables mentioned are not employed as explanatory variables. There appears to be no concern regarding multicollinearity which is corroborated by the Variance Inflation Factor (VIF) test, for which the results can be found in Appendix A.2, reporting mean VIF values well below the strict critical value of 2.5 for all different specifications of the models. With regards to the hypothesized relations there appears to be some correlation between the education measures and ROA, however none for the experience variable nor for Tobin’s Q. Examining the control variables all, with the exception of gender ratio, are correlated with the measures of financial performance

corroborating their relevance and the resulting inclusion in the analysis. Conclusive evidence for the presence of the hypothesized relations will be discussed in detail in the Results section.

3.6. Endogeneity

Endogeneity is a common problem within the literature mainly in two forms, being omitted variable bias and reverse causality (Adams and Ferreira, 2009). For this analysis specifically,

Tobin's Q ROA Education

(16)

16

omitted variable bias implies that certain characteristics that are highly related to the proposed relation between gender diversity and firm performance are not included in the analysis. This leads to the incorrect interpretation of the coefficient depicting the relation between gender diversity and firm performance. Adams and Ferreira (2009) show that when estimating a (firm) fixed-effects regression, as opposed to a pooled OLS estimation, the sign of the coefficient depicting the relation between gender diversity and firm performance reverses becoming negative and significant. The authors suggest that the difference between

coefficients is because of the omittance of certain firm-specific variables, which influenced the coefficient of the pooled OLS specification. Henceforth, in line with Adams and Ferreira (2009), as a potential solution1_{for omitted variable bias, this paper will estimate both a pooled}

OLS and a Fixed Effects specification where the latter is to control for the possibility of omitted time-invariant firm-specific variables.

Reverse causality implies that the explanatory variables may impact the dependent variables, yet the dependent variables may also impact the values which the explanatory variables takes, making the results of the analysis hard to interpret (Hermalin and Weisbach, 2003). For gender diversity, this could imply that better-performing firms may be more likely to hire female board members (Carter et al. 2010). Or that females, who are relatively scarce in the field of directors, might prefer to work for better performing firms (Farrel and Hersch, 2005). If this holds, this would imply that gender diversity may not cause improved performance but instead that females work for better-performing firms, making the results difficult to interpret. The most common method applied to combat reverse causality is by using the instrumental variable approach. Where this instrumental variable can be applied to exogenously determine the explanatory variables of interest, i.e., a variable which explains the proportion of females in the board of directors but has no relation with firm performance (Adams and Ferreira, 2009; Smith et al., 2006). Consequently, the instrument and resulting regression are estimated using either the 2SLS (Adams and Ferreira, 2009; Campbell and Minguez-Vera, 2008; Carter et al. 2003) or 3SLS model (Carter et al., 2010). These instruments however are difficult to find, due to lack of data and the fact that most variables that could be used as an instrument are mostly correlated with governance characteristics which are included in the analysis in the first place. (Adams and Ferreira, 2009; Smith et al., 2006). Therefore, this paper will turn to another method of analysis to allow for the determination of causal effects, being the

propensity scoring method (PSM), for which the method will be explained in the next section.

1_{In a simulation study conducted by Beck, Brüderl and Woywode’s (2008) the fixed effects model}

(17)

17

4. Methodology

All quantitative research applied to test the hypotheses is conducted by the use of the statistical tool STATA.

4.1. Gender-related Differences

In line with Hillman et al. (2002), for the test of hypotheses 1 and 3, chi-square analysis will be conducted to test for fundamental differences among female and male board members concerning education and experience. For each year individual board members are

categorized two ways, firstly based on their gender and secondly on the education/experience level they have. For education, the categorization applied is undergraduate/associate, graduate and doctoral degree respectively and for experience this ranges from 1 up to and including 52

current board positions. The total number of individual board members belonging to a certain category are aggregated to assess the divisions over education and experience levels for male and female board members. The aggregated numbers allow for the chi-square analysis testing if differences across groups are significant.

4.2. Firm-level Estimations

In order to test for hypotheses 2 and 4, the effect of gender diversity, through education and experience, on firm performance will be estimated. Data unavailability with regards to the education level for some directors led to a much smaller amount of observations as compared to the experience level. This is due to the fact that if the education level was missing for one or more directors of a specific board, the whole board had to be excluded as no fair averages could be calculated. Therefore, both the dummy variables will be included in separate regressions as regressing them combined would exclude many observations. The following models are estimated:

𝑹_𝒊,𝒕= 𝜷_𝟎+ 𝜷_𝟏∗ 𝑮𝑫_𝒊,𝒕+ 𝜷_𝟐∗ 𝑿_𝒊,𝒕+ 𝝐_𝒊,𝒕 (1) 𝒍𝒐𝒈(𝑸)𝒊,𝒕 = 𝜷𝟎+ 𝜷𝟏∗ 𝑮𝑫𝒊,𝒕+ 𝜷𝟐∗ 𝑿𝒊,𝒕+ 𝝐𝒊,𝒕 (2)

Where the dependent variables R and Q stand for the financial performance measures ROA and the logarithm of Tobin’s Q, where the latter is due to concerns with regards to normality. Furthermore, GD in both regressions stands for the measure of Gender Diversity which is

2_{99.57% of the total sample records shows a total number of current board positions equal to or less}

(18)

18

depicted by the Education and Experience dummy variables. Lastly, X is a vector of control variables. This vector includes the logarithm of firm size, board size and the untransformed values of debt level and the gender ratio. Besides, this vector includes the industry codes for the OLS estimation of this model, whereas it is not included in the fixed effects estimation due to collinearity. The coefficient of interest is 𝜷_𝟏 depicting the effect of gender diversity on firm performance through increasing the average level of education and experience within the board.

4.3. Propensity Scoring Method

Classical regression models, such as the one described in section 4.2, aim to determine a causal effect by regressing explanatory variables and a vector of control variables on a dependent variable, where the coefficient of the explanatory variable ceteris paribus is the effect on the dependent variable e.g. 𝜷_𝟏 representing gender diversity its effect on firm performance. These models are however unable to detect if the observations, divided by the value the observations take on the explanatory variable (for the case of this analysis a one or a zero on either dummy variable), are comparable in terms of distribution for observable

characteristics such as firm size. This implies that results could be interpreted wrongly, as the sign and magnitude of the explanatory variable may not be due to the dummy variable taking a 1 itself, but instead due to not accounting for overlap in distribution between the two different groups of observations (Li, 2013). PSM does account for this possibility by matching the observations of the two groups based on observable characteristics.

First of all, the propensity scores are calculated which can be defined as the probability of an observation belonging to the ‘treatment’ group based on several characteristics. Where treated implies that the observation takes a one on either dummy variable, and non-treated implies a zero on either dummy variable. The following probit model is estimated.

𝑻𝒊 = 𝜷𝟎+ 𝜷𝟏∗ 𝑿𝒊+ 𝝐𝒊,𝒕 (3)

Where T represents the ‘treatment’ applied represented by the dummy variables for Education and Experience respectively, and X is a vector of variables similar to the vector described in section 4.2, without the industry codes. Consequently, the propensity score for each

(19)

19

observations within each block. The outcome variables, in this case, are the dependent variables Tobin’s Q (logarithm) and ROA.

5. Results

5.1. Gender-related Differences

This section describes the results regarding the tests for individual-level differences of Education and Experience respectively.

Hypothesis 1 predicts that female board members are more likely to hold advanced degrees. Examining table 3 it can be noted that over half of all board members, hold a Masters Degree. In addition, for each consecutive year the percentage of females who hold a doctoral degree, is higher than that of males by approximately 2%. Although these differences are less extreme than those reported by Hillman et al. (2002), this does suggest that relatively more female board members hold advanced degrees. The statistics for differences across groups for each year are significant at the 10% level and corroborate that, indeed this difference is significant. Henceforth we find support for the hypothesis that female board members are on average more highly educated.

Hypothesis 3 predicts that female board members are likely to serve on more boards

simultaneously compared to male board members. Examining table 4 it can be seen that close to 58% of male board members hold one position, whereas this percentage is only 50% for females. Also, the percentages of females holding 3 or 4 board positions simultaneously are higher than those of males that hold a similar amount of board positions. The test for

Table 3: Individual-level differences - Education

The table depicts descriptive statistics with regard to education. The statistics and associated p-value for each year can be found at the bottom of the table.

Education/Years 2016 2017 2018

Percent (N) Females Males Females Males Females Males

(20)

20

differences across groups are all highly significant at the 1% level. Henceforth we find support for Hypothesis 3 as well.

5.2. Firm-level Estimations

The previous section documented support for the hypothesis that female board members are both more highly educated, and hold more board positions simultaneously. This section will report the empirical results of the firm-level estimations relating these higher levels of education and experience to firm performance.

Hypothesis 2 predicted that females, who increase the average education level within the board, positively impact firm performance. This is measured by a dummy variable which takes the value of 1 when the average education level of females is higher than that of males within the same board of directors. The coefficient of this dummy variable can be interpreted as the effect females have on firm performance by increasing the average education level and is expected to have a positive sign. The results of the estimations can be found in table 5. For our measure of financial performance, Tobin’s Q, none of the estimations show a significant effect suggesting there is no relation between our education dummy and Tobin’s Q. Interestingly enough however the OLS specification in column 1 depicts a negative coefficient on the dummy variable. Whereas when controlling for firm-specific factors by means of a fixed-effects specification in column 2 the dummy variable has a positive sign. This suggests that the negative effect is driven by omitted firm-specific factors. Lastly, in

Table 4: Individual-level differences - Experience

The table depicts descriptive statistics with regard to the experience depicted by the total number of current board positions. The statistics and associated p-value for each year can be found at the bottom

of the table.

Experience/Years 2016 2017 2018

Percent (N) Female Male Female Male Female Male

1 50.07% (370) 57.34% (1551) 51.20% (404) 57.41% (1531) 53.24% (452) 57.33% (1502) 2 30.04% (222) 29.80% (806) 30.67% (242) 30.63% (817) 29.68% (252) 31.22% (818) 3 17.59% (130) 11.20% (303) 15.21% (120) 10.05% (268) 13.66% (116) 9.43% (247) 4 2.30% (17) 1.48% (40) 2.79% (22) 1.84% (49) 3.42% (29) 1.91% (50) 5 0.00% (0) 0.18% (5) 0.13% (1) 0.07% (2) 0.00% (0) 0.11% (3) Total 100% (739) 100% (2705) 100% (789) 100% (2667) 100% (849) 100% (2620)

Chi-square Likelihood ratio Chi-square Likelihood ratio Chi-square Likelihood ratio

Value statistic 28.096 27.610 21.493 20.375 20.760 20.300

(21)

21

column 3, it can be seen that the effect of the education dummy has a negative relation with Tobin’s Q, ceteris paribus the control variables employed in the analysis. Most importantly however none of the coefficients are significant, implying that no real inferences can be drawn with regards to Tobin’s Q.

For our second measure of financial performance, ROA, similar results are found. The OLS specification in column 4 depicts a negative coefficient and the fixed effects specification in column 5 shows a positive coefficient. Lastly, the causal effect determined using PSM depicted in column 6 is also negative. In addition, the coefficient in column 4 is significant at the 10% level suggesting that boards for which females have higher education than their male colleagues actually do worse in terms of ROA. Although the results are barely significant, based on the estimation in column 4, we conclude that hypothesis 2 is not supported

Hypothesis 4 predicts that females positively impact firm performance due to increased levels of experience within boards. This is measured by a dummy variable which takes the value of 1 when the average experience level of females is higher than that of males within the same board of directors. The coefficient of this dummy variable can be interpreted as the effect females have on firm performance by increasing the average experience level and is expected to have a positive sign. The results of the estimations can be found in table 6.

Table 5: Education and firm financial performance

Variable descriptions are similar to that of table 1. With the exception that Tobin’s Q, Board Size and Firm Size are logarithmic transformations due to concerns regarding normality. Standard errors are between brackets where in columns 1 and 4 standard errors are adjusted for group correlation at the

firm level. Consequently, in columns 2 and 5 standard errors are adjusted for potential heteroskedasticity. Asterisk indicate significance at the 1% (***), 5% (**) and 10% (*) level respectively. For columns 3 and 6 the coefficient represents the average treatment effect of higher

education of females within boards on firm performance.

.

(1) (2) (3) (4) (5) (6)

VARIABLES Tobin’s Q Tobin's Q Tobin's Q ROA ROA ROA

Education Dummy -0.022 0.025 -0.022 -1.338* 0.519 -1.255 (0.055) (0.027) (0.045) (0.759) (0.499) (0.512) Board Size 0.063 0.027 4.142* 2.528 (0.170) (0.094) (2.474) (2.596) Firm Size -0.194*** -0.349*** -1.132*** -2.616 (0.028) (0.072) (0.386) (2.076) Debt Level -0.008*** -0.001 -0.119*** -0.057 (0.003) (0.001) (0.028) (0.042) Gender Ratio -0.571* 0.050 -5.986* 7.861 (0.325) (0.214) (3.614) (5.461) Observations 562 570 563 571 R-squared 0.382 0.227 0.166 0.082

Industry Yes*** No No Yes*** No No

Firm Fixed Effects No Yes No No Yes No

Year Fixed Effects Yes Yes No Yes Yes No

(22)

22

For the measure of financial performance Tobin’s Q, all different specifications in columns 1-3 depict a negative relation with the experience dummy. With the exception of the fixed-effects estimation which is significant at the 5% level, none of the coefficients for the experience dummy are significant.

Regarding the measure of financial performance, ROA, all the coefficients of the experience dummy are negative and insignificant. All in all, these results suggest that boards, who have females with relatively more experience compared to their male colleagues actually do worse in terms of Tobin’s Q. We conclude that hypothesis 4 is not supported

There appears to be a negative relationship between both the gender diversity measures applied in this paper and firm financial performance, contradicting the hypothesized relations. We suggest that this may be mainly due to the construction of our explanatory variable for the following reason. As argued by Adams & Ferreira (2009) firms who already have good governance may not necessarily benefit from gender diversity whereas firms with bad governance do. Similar reasons may apply for this analysis as well. Since firms who are governed appropriately are likely to have board members who are both more highly educated and more experienced. For these boards of directors, it is less likely that females when taking their positions either have a higher education or experience level on average than the male directors do. This leads to the dummy variable taking the value of 0 suggesting that the firm is likely to have lesser performance. Whereas the firm is likely to perform well due to it being governed appropriately. Instead, firms who are misgoverned and are expected to do worse in terms of performance might benefit from the addition of female directors as their education or experience level is likely to be higher than that of the male directors on the board.

Consequently, for these firms the dummy variables take the value of 1.

(23)

23

Table 6: Experience and firm financial performance

Variable descriptions are similar to that of table 1. With the exception that Tobin’s Q, Board Size and Firm Size are logarithmic transformations due to concerns regarding normality. Standard errors are between brackets where in columns 1 and 4 standard errors are adjusted for group correlation at the

firm level. Consequently, in columns 2 and 5 standard errors are adjusted for potential heteroskedasticity. Asterisk indicate significance at the 1% (***), 5% (**) and 10% (*) level respectively. For columns 3 and 6 the coefficient represents the average treatment effect of the higher

experience of females within boards on firm performance.

Concerning the control variables included a few things can be noted. Board size is significant for only two out of the eight specifications suggesting that board size is not a useful control variable. The debt level is significant in all four OLS specifications and has a negative sign, in line with the results of Dsezo and Ross (2012) and Jackling and Johl (2009). Firm size

appears to be negatively related to firm performance, which is in line with findings of (Campbell and Minguez-Vera, 2009), and is significant for 6 out of 8 specifications.

Gender ratio has a significant and negative coefficient for all OLS specifications which is in line with the expectation. As theory suggests gender diversity has a positive effect on firm performance, and the gender ratio for this analysis is measured by the number of male board members over total board members which implies that higher values depict less gender diversity. Also, when accounting for firm-specific effects through the fixed effects estimations, the sign of the relation for gender diversity reverses which is in line with the results found by Adams and Ferreira (2009).

Industry codes included in the OLS specifications are all significant at the 1% level and increase the goodness of fit of the model, adding to the validity of the inclusion as a control variable.

(1) (2) (3) (4) (5) (6)

VARIABLES Tobin’s Q Tobin's Q Tobin's Q ROA ROA ROA

Experience Dummy -0.022 -0.033** -0.016 -0.108 -0.108 -0.038 (0.034) (0.014) (0.028) (-0.249) (-0.269) (0.415) Board Size 0.067 0.101* 2.591 0.031 (0.110) (0.056) (1.617) (0.018) Firm Size -0.231*** -0.389*** -1.962*** -4.046* (0.021) (0.064) (-7.389) (-1.786) Debt Level -0.004* -0.001 -0.062** -0.023 (0.002) (0.001) (-2.133) (-1.166) Gender Ratio -0.502** 0.035 -6.059** 3.307 (0.210) (0.109) (-2.311) (0.914) Observations 1,375 1,392 1,377 1,394 R-squared 0.403 0.267 0.177 0.059

Industry Yes*** No No Yes*** No No

Firm Fixed Effects No Yes No No Yes No

Year Fixed Effects Yes Yes No Yes Yes No

(24)

24

6. Conclusion

This analysis investigated the characteristics associated with female board members, and how these can be empirically related to firm performance. Firstly, the theory suggests that females have to meet higher standards in order to stand out for their first upper management position and are therefore more likely to have a higher level of education compared to males. In

addition, due to the relative scarceness of female board members, and the tendency of firms to select new directors from the pool of current directors, female board members have a higher chance at attaining multiple directorships after being selected for their first board position. This analysis has examined if female board members are more highly educated and hold more board positions simultaneously, by estimating statistics allowing the determination of

significant differences across gender. More importantly, this analysis differentiates from previous research by going beyond the general diversity measure. It does so by constructing a measure which measures if the average level of education and experience within the board increased due to the presence of (a) female director(s), and how this is related to firm

performance Additionally, it addresses endogeneity concerns common within the literature by applying fixed-effects estimations and using the propensity scoring method.

Utilizing a panel dataset comprising 475 US-based companies covering the period 2016 – 2018, the results show that indeed, female board members are on average more highly educated and hold more board positions simultaneously than male directors do. Translating these findings and relating them to firm performance however yields opposing and mostly insignificant results. We expect that this is due to endogeneity in the form of reverse causality. Better performing firms are expected to be governed appropriately and have high average levels of experience and education on their board already, implying that these benefit less from the increase in education/experience through the addition of (a) female director(s). This adds to the relevance of addressing endogeneity which is a major problem in the literature as pointed out by Adams & Ferreira (2009)

Also, due to data unavailability with regards to our gender diversity measure of education we were unable to construct a multivariate analysis, as this would exclude a large number

observations (over 800) for our proxy of experience. As Pletzer et al. (2015) pointed out the relation between gender diversity and firm performance may be to difficult to examine on a univariate level making this a significant limitation of the analysis conducted. Therefore further attempts should be made to examine the relation between gender diversity and firm performance on a multivariate level, going beyond the general diversity measure.

(25)

25

References:

Adams, R., and Ferreira, D. (2009). Women in the boardroom and their impact on governance and performance. Journal of Financial Economics, 94(2), 291-309.

Baysinger, B., and Butler, H. (1985). Corporate governance and the board of directors: Performance effects of changes in board composition. Journal of Law, Economics, and Organization, 1(1), 101-124.

Beck, N., Brüderl, J., and Woywode, M. (2008). Momentum or deceleration? Theoretical and methodological reflections on the analysis of organizational change. Academy of

Management Journal, 51(3), 413-435.

Boyd, B. (1990). Corporate linkages and organizational environment: A test of the resource dependence model. Strategic Management Journal, 11(6), 419-430.

Brammer, S., Millington, A., and Pavelin, S. (2007). Gender and ethnic diversity among UK corporate boards. Corporate Governance: An International Review, 15(2), 393-403.

Brooks, C. (2014) Introductory Econometrics for Finance. Cambridge: Cambridge University Press

Burgess, Z., and Tharenou, P. (2002). Women board directors: Characteristics of the few. Journal of Business Ethics, 37(1), 39-49.

Campbell, K., and Mínguez-Vera, A. (2008). Gender diversity in the boardroom and firm financial performance. Journal of Business Ethics, 83(3), 435-451.

Carter, D., D'Souza, F., Simkins, B., and Simpson, W. (2010). The gender and ethnic diversity of US boards and board committees and firm financial performance. Corporate Governance: An International Review, 18(5), 396-414.

Carter, D., Simkins, B., and Simpson, W. (2003). Corporate governance, board diversity, and firm value. Financial Review, 38(1), 33-53.

Chung, K., and Pruitt, S. (1994). A simple approximation of Tobin's q. Financial Management, 23(3), 70-74.

Dezsö, C., and Ross, D. (2012). Does female representation in top management improve firm performance? A panel data investigation. Strategic Management Journal, 33(9), 1072-1089. Erhardt, N., Werbel, J., and Shrader, C. (2003). Board of director diversity and firm financial performance. Corporate Governance: an International Review, 11(2), 102-111.

Fama, E., and Jensen, M. (1983). Separation of ownership and control. The Journal of Law and Economics, 26(2), 301-325.

(26)

26

Fuhrmans, V. (2018, 30 September). California becomes first state to mandate female board directors. The Wall Street Journal. Retrieved from: https://www.wsj.com/articles/california-becomes-first-state-to-mandate-female-board-directors-1538341932

Gentry, R., and Shen, W. (2010). The relationship between accounting and market measures of firm financial performance: How strong is it? Journal of Managerial Issues, 22(4), 514-530. Harrison, D., & Klein, K. (2007). What's the difference? Diversity constructs as separation, variety, or disparity in organizations. Academy of Management Review, 32(4), 1199-1228. Hermalin, B., Weisbach, M. (2003). Boards of directors as an endogenously-determined institution: a survey of the economic literature. Economic Policy Review 9, 7–26.

Hillman, A., and Dalziel, T. (2003). Boards of directors and firm performance: Integrating agency and resource dependence perspectives. Academy of Management review, 28(3), 383-396.

Hillman, A., Cannella Jr, A., and Harris, I. (2002). Women and racial minorities in the boardroom: how do directors differ?. Journal of Management, 28(6), 747-763.

Hillman, A., Withers, M., and Collins, B. (2009). Resource dependence theory: A review. Journal of Management, 35(6), 1404-1427.

Jackling, B., and Johl, S. (2009). Board structure and firm performance: Evidence from India's top companies. Corporate Governance: An International Review, 17(4), 492-509. Jensen, M. (2010). Value maximization, stakeholder theory, and the corporate objective function. Journal of Applied Corporate Finance, 22(1), 32-42.

Labelle, R., Francoeur, C., and Lakhal, F. (2015). To regulate or not to regulate? Early evidence on the means used around the world to promote gender diversity in the boardroom. Gender, Work and Organization, 22(4), 339-363.

Li, M. (2013). Using the propensity score method to estimate causal effects: A review and practical guide. Organizational Research Methods, 16(2), 188-226.

Lückerath-Rovers, M. (2013). Women on boards and firm performance. Journal of Management and Governance, 17(2), 491-509.

O’Connell, V., and Cramer, N. (2010). The relationship between firm performance and board characteristics in Ireland. European Management Journal, 28(5), 387-399.

Pearce, J., and Zahra, S. (1992). Board composition from a strategic contingency perspective. Journal of Management Studies, 29(4), 411-438.

(27)

27

Post, C., and Byron, K. (2015). Women on boards and firm financial performance: A meta-analysis. Academy of Management Journal, 58(5), 1546-1571.

Ragins, B., Townsend, B., and Mattis, M. (1998). Gender gap in the executive suite: CEOs and female executives report on breaking the glass ceiling. Academy of Management Perspectives, 12(1), 28-42.

Richard, O. (2000). Racial diversity, business strategy, and firm performance: A resource-based view. Academy of Management Journal, 43(2), 164-177.

Roberts, J., McNulty, T., and Stiles, P. (2005). Beyond agency conceptions of the work of the non‐executive director: Creating accountability in the boardroom. British Journal of

Management, 16(s1), S5-S26.

Robinson, G., and Dechant, K. (1997). Building a business case for diversity. Academy of Management Perspectives, 11(3), 21-31.

Rose, C. (2007). Does female board representation influence firm performance? The Danish evidence. Corporate Governance: An International Review, 15(2), 404-413.

Shleifer, A., and Vishny, R. (1997). A survey of corporate governance. The Journal of Finance, 52(2), 737-783.

Shrader, C., Blackburn, V., and Iles, P. (1997). Women In Management And Firm Financial Performance: An Exploratory Study. Journal of Managerial Issues, 9(3), 355–372.

Singh, V., Terjesen, S., and Vinnicombe, S. (2008). Newly appointed directors in the

boardroom:: How do women and men differ?. European Management Journal, 26(1), 48-58. Staley, O. (2016, 3 May). You know those quotas for female board members in Europe? They’re working. Quartz. Retrieved from: https://qz.com/674276/you-know-those-quotas-for-female-board-members-in-europe-theyre-working/

Smith, N., Smith, V., and Verner, M. (2006). Do women in top management affect firm performance? A panel study of 2,500 Danish firms. International Journal of Productivity and Performance Management, 55(7), 569-593.

Terjesen, S., Sealy, R., and Singh, V. (2009). Women directors on corporate boards: A review and research agenda. Corporate governance: an International Review, 17(3), 320-337.

(28)

28

Appendix:

Appendix A.I: Descriptive statistics Industries

Note: The division of firms over industries depicted above is for the year 2018, as for this year the most individual firms were available, providing the most accurate picture. For the other years the separate percentages for each code do not differ with more than 0.20 percent, thus the statistics above

are generalizable for other years as well.

Appendix A.II: VIF test for Multicollinearity

Note: VIF tests, for which the results are in this table, were conducted on the OLS specifications of the model. The dependent variables of each specification are in the top row. All mean VIF values are

below 2.5 indication no concern for multicollinearity.

Industry Classification Code Company count Percentage Cumulative Percentage

Industrial 1 334 71.52 71.52

Utility 2 37 7.92 79.44

Transportation 3 15 3.21 82.66

Bank/Savings and Loan 4 22 4.71 87.37

Insurance 5 20 4.28 91.65

Other Financials 6 39 8.35 100.00

Total 467 100.00

(1) Tobin’s Q (2) Tobin’s Q (3) ROA (4) ROA

Variable VIF VIF VIF VIF

MSc Thesis Finance