The size and independency of the board of directors during the crisis in the U.S.

(1)

Bachelor Thesis

The size and independency of the board of directors during the crisis in

the U.S.

Abstract

This paper focuses on whether board size and board independence affects a firm’s performance. Firm performance is measured with Tobin’s Q and ROA. An empirical analysis is performed with 166 public listed (NASDAQ) companies in the United States. OLS with fixed effects is regressed on firm performance. Overall, the results of this paper are mixed. The marginal effect of board independence during the crisis is found to be positive. However, this result is not clear due to the lack of control variables in the regression. Second, significant positive and negative effects are both found on the board size on different regressions. Finally, the percentage of females in the board, the firm’s age and the amount of assets seems to play a role in the firm performance. However, these findings could be invalid due to potential endogeneity.

Name Verhad Alaydrus

Student number 11022310

Programme Economie en Bedrijfskunde

Specialization Financiering en Organisatie Name supervisor Evgenia Zhivotova

(2)

Statement of originality

This document is written by Student Verhad Alaydrus who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Introduction

In recent years, there has been a great deal of news about corporate accounting scandals (Aebi, Sabato and Schmid, 2012). The board of directors has the legal responsibility to ensure that the firms meet its mission requirements. The board is guided by regulations that dictate the goal and objectives of the organization. A trend is now being observed wherein firms are more frequently adopting independent board members in their firm.1 Independence allows directors to be more objective and evaluate the performance and well-being of the company without any conflict of interest.

Firms are trying to improve their corporate governance by selecting the right people in the board of directors. In a complex firm, the function of the board of directors becomes more crucial than in smaller firms. Many studies2 claim that the composition of the board of directors has an impact on the valuation of the firm.

During the recent financial crisis (2007–2009), many companies experienced financial distress. Such a financial crisis brings with it a great deal of difficulty for the firms, which might result in bankruptcy. After the fall of Lehman Brothers, many banks reduced their lending (Chodorow-Reich, 2014), which produces problems for firms in terms of financing. This paper seeks to analyze whether board independence has an effect on firm performance in this crisis and from 2000–2016. Much research has been conducted in the field of board composition. This study strives to

incorporate an interaction variable in the regression model in order to study the effect of the crisis and board independence on firm performance. The following research question is answered in this paper: “Does board independence lead to a higher valuation of public firms during the financial crisis in the United States?” An empirical analysis is performed from 166 publicly listed (NASDAQ) companies in the United States. The data for the variables is retrieved for the year 2000–2016. The data is collected from the Datastream database (Thomson Reuters).

Although there are some limitations due to potential endogeneity, I conclude the following points. First, the marginal effect of board independence on firm

performance during the crisis is found to be positive.3 The marginal effect is calculated in order to estimate whether board independence in the crisis has a positive effect on firm performance. However, this result is not clear, because the regression was performed without control variables. Second, when regressing board size on firm performance, the results show that the coefficients are both positive and

1

See Figure 1. from the article of Liu, Miletkov, Wei and Yang (2015). Note that this setting is in Asia.

2

See the article from Johnson, Schnatterly and Hill (2013) about board composition (social capital, demographics and human capital).

3

In the research model stated in formula (1) in Research methodology, I take the partial derivative of Tobin’s Q with respect to Board Indepedence. The marginal effect in statistical terms is: dTobinsQ / dBINDP = β1 + β4 CRISIS, where β1 is the coefficient of the variable board independence, β4 is the coefficient of interaction term: BINDP*CRISIS and crisis is a year-dummy variable.

(5)

negative in different regressions. Finally, the percentage of females on the board, firm age and assets seem to play a role in the firm performance.

The rest of this paper is organized as follows. In section 2, I discuss relevant literature and develop my hypothesis. Section 3 describes the research methodology utilized by this paper. Section 4 reports the results, and section 5 provides the

discussion. I conclude this paper in section 6. The appendix is included at the end of this paper.

Literature review

In light of recent events, such as the financial crisis, there has been considerable concern about corporate governance. The agency problem is a notable issue that many companies are facing. This is a well-known problem that arises when a conflict of interest exists between the agent and principle (Jensen and Meckling, 1976). In the corporate governance setting, the agents are the workers, such as the managers, and the principal is the firm. The problem is that managers, such as the CEO, only want to maximize their own utility, or his/her own wealth. So, the board of directors plays an important role in reducing this agency problem. The board has to realign the interest of the agent so that he does not act for his self-interest, but for the principal. In literature, there has been a great deal of research conducted in this field due to its importance. The directors in the board are responsible for the composition and compensation of the senior management (to hire and fire personnel), and they ensure that no conflicts exist between the firm and stakeholders (Baysinger and Butler, 1985).

There is some disagreement concerning whether board independence has a positive effect on the valuation of the firm. Several studies (Rosenstein and Wyat ,1990; Weisbach, 1988) have found that board independence leads to a higher firm valuation under different circumstances, while other have found a negative

relationship (Yermack, 1996) or no relationship at all (Hermalin & Weisbach, 1991). There are quite a few reasons why board independence can have a positive effect on a firm’s valuation. Board independence can lead to more coherent decision making with respect to both parties, the firm and stakeholders. Judge and Zeithaml’s (1992) study showed that firms with more insiders on the board engage in less strategic decision making than outsiders. Further, a study by Johnson, Hoskisson and Hit (1993) showed that boards are more strategically aligned and have more incentive to restructure the firm when the board has more outsiders. This suggests that during a crisis, firms will add more independent board members to prevent a decline in performance or bankruptcy. Also, Rosenstein and Wyat (1990) supported the positive relationship between firm performance and board independence. They

(6)

mentioned that the expected gain of outside directors exceeds the expected costs of inefficient decision making and managerial entrenchment.

Concerns have more arisen concerning corporate governance after the financial crisis. Erkens, Hung and Matos (2012) investigated the influence of

corporate governance on the financial crisis of 2007–2009. They found that firms with more board independence and higher institutional ownership experienced poor

returns during the financial crisis. To explain this, they suggested that independent boards raised more equity during the crisis, eventually leading to a transfer of wealth from existing shareholders to debt holders.

The financial crisis brought a great deal of difficulty for firms, which could even result in bankruptcy. After the fall of Lehman Brothers, many banks reduced their lending (Chodorow-Reich, 2014), thus creating further problems for firms. I believe that during a financial crisis, a firm will acquire more outside members in the board. In times when firms face difficulty, the need for outside support increases (Pfeffer and Salancik, 1978). Such outside support could provide valuable resources and

information, maintain connections with other firms and help with legal actions (Pfeiffer & Salancik, 1978). This is why board independence could help maintain firm

performance during such a crisis. To test the link between board independence in the crisis on firm performance, I use an interaction variable in the research model.

I believe that, during the crisis, firms more actively sought board independence because of its positive effects. In this study, the marginal effect4 is determined in order to analyze the effect of board independence during the crisis, but before that, I test the following hypotheses. Note that firm performance is measured by Tobin’s Q and ROA.

Hypothesis 1a: The effect of board independence on firm performance is positive. Hypothesis 1b: The effect of board independence during the crisis on firm

performance is positive.

(7)

Board size

The size of the board also has an impact on the firm’s value. Lipton and Lorsch (1992) discussed that the optimal corporate board size is approximately eight or nine members. Jensen (1993) explained that a size beyond a certain threshold (seven or eight) makes it difficult to function effectively

Yermack (1996) investigated the effects of board size on firm performance. Based on a sample of 452 large industrial corporations (between 1984 and 1991), he found that small boards of directors were more effective. He also found out that the board size has a negative effect on Tobin’s Q. The results of this finding are

consistent with other literature, such as Lipton and Lorsch (1992) and Jenssen (1993).

In a study by Fama and Jensen (1983), it was suggested that the organization of a firm is controlled by the firm’s size and complexity. Larger firms tend towards more complexity and a comprehensive hierarchical structure. Hence, the board of directors in large firms requires more directors because of the different activities.

Adams and Mehran (2011) conducted a study in the bank industry and found a positive relationship between board size and performance, measured by Tobin’s Q. They explained that subsidiary directorship may be an important element for their results. A subsidiary company (also called a daughter company) is a legal entity within a holding company. Subsidiary directorship may add value because they improve coordination and communication among different levels in the holding company, in this case, the bank.

Thus, most of the findings in the literature have reported the same results, namely, that board size negatively affects firm value. I will test the following hypothesis:

(8)

Research Methodology

This section details the research methodology utilized in this study. First, the

research model and the different variables are explained. Next, the data is described. Finally, the descriptive statistics and analysis are reviewed.

The research model used in this paper is as follows:

(1) Tobin’s Q = β0 + β1 BINDP + β2 BSINDP + β3 CRISIS + β4 CRISIS*BINDP + β5 BSIZE + β6 BMEETINGS+ β7 BFEM + β8 FAGE + β9 FPB + β10 FLEVERAGE + β11 FSIZE + Industry dummies + ε

Hypotheses 1a and 1b correspond respectively to the following statistical hypotheses:

(2) H0: β1 = 0 H1: β1 > 0 (3) H0: β4 = 0 H1: β4 > 0

marginal effect: (4) dTobinsQ / dBINDP = β1 + β4 CRISIS

and hypothesis 2

(5) H0: β5 = 0 H1: β5 < 0

To measure firm performance, I use Tobin’s Q. Tobin’s Q is defined as the market value of assets / replacement cost of assets. The ratio that I estimate is the market value of assets / bookvalue of assets. Tobin’s Q is a common measure for firm performance, and it has been utilized by a number of studies (Hermalin and

Weisbach; 1991, Jenssen; 1993). The downside of this measure is that that it might be a cause for corporate governance rather than the effect (Bennedsen, Kongsted and Nielsen). Moreover, I replicated the results using ROA (return on assets) as a measure of firm performance. This is also commonly used for the dependent variable in board studies (Yermack, 1996; Liu et. al, 2015; Eisenberg, Sundgren and Wells, 1998; Duchin, Matsusaka, Ozbas, 2010).

The first independent variable is board independence. This is the fraction of independent board members in the board size. In the tested model, I included an interaction term with the crisis to test the marginal effect. There is also a second independent variable. This variable, called “strictly board independence,” is the

(9)

percentage of strictly independent board members5 and has stronger assumptions than the first independent variable.

Next, the board size describes the number of directors that are on the board. It has already been mentioned that most literature has proven that there exists a

negative relationship between size and firm performance. It is expected to achieve the same results.

Crisis is a year-dummy variable [0,1]. It equals 1 for the years of the financial crisis, 2007–2008, and 0 for all other years. The years 2007–2008 are picked for the “crisis variable” because in this paper, I focus on the beginning of the financial crisis (fall of Lehman brothers). However, the total period of the entire dataset is 2000– 2016.

I consider two board-related variables, namely, BFEM and BMEETINGS. BFEM refers to the percentage of women on the board of directors. Adams and Ferreira (2009) suggested that female directors appear to have an impact on firm performance. BMEETINGS details the number of board meetings in a fiscal year. Vafeas (1999) found a relationship between the number of board meetings and corporate governance. It seems that board meetings have an inverse effect on firm value. The results showed, namely, that share prices decline if the board meetings increase.

Next, to control for firm characteristics, I use four firm variables: FAGE, FPB, FLEVERAGE and FSIZE. FAGE is the age of the firm in years. FLEVERAGE is the ratio of total debt and total assets. FSIZE is the size of the firm, and it is measured in the natural logarithm of assets. FPB ratio is the price to book ratio measured as market value of equity divided by book value of equity. These control variables are used in various studies that investigate board composition (Duchin et al., 2010; Bennedsen et al., 2009; Francis, Hasan & Wu, 2012).

I use the SIC-1 code for grouping the different industries in the firm. In total, there are eight industries: (1) mining; (2) constructing; (3) manufacturing; (4) transportation, communications, electric, gas and sanitary service; (5) wholesale trade; (6) retail trade; (7) finance, insurance and real estate and (9) services. A summary of the variables in the model with its source is given in Appendix A.

The marginal effect of board independence is stated in formula four. The formula is obtained by taking the partial derivative of Tobin’s Q with respect to BINDP.

5_{‘Strictly’ is defined in Datastream as: Percentage of strictly independent board members (not employed by the} company; not representing or employed by a majority shareholder; not served on the board for more than ten years; not a reference shareholder with more than 5% of holdings; no cross-board membership; no recent, immediate family ties to the corporation; not accepting any compensation other than compensation for board service).

(10)

Sample

The data used in this paper is gathered from Datastream. First, all public firms in the United States were selected via the NASDAQ exchange. Only active firms were selected. Then, firms were selected with a start date of 2000–2016. Afterwards, the variables were retrieved from 2005–2016. This period was selected because there is limited variable data available before 2005. Tobin’s Q is calculated as market value of assets / bookvalue of assets (Hermalin and Weisbach; 1991, Jenssen; 1993). FAGE is calculated manually by taking the difference between the going-public date and the end of the fiscal year. FLEVERAGE is also calculated manually by dividing the total debt and total assets (Francis, Hasan, Wu, 2012). FASSETS is transformed into the natural logarithm. All the other variables that are stated in model one are unaltered. Finally, outliers were removed from the dataset using the winsor (α = 5%) function in Stata. This is done for the variables Tobin’s Q, FPB and FROA.

Descriptive statistics

Table 2 reports descriptive statistics for the entire sample of firms and board

variables. Since the data of the variables is retrieved over the period of 2000–2016, the maximum number of observables of one variable is 2,822. The missing values are due to the fact that the data is not available in Datastream. In the following paragraph, the important variables of the descriptive statistics are reviewed.

We can see that the median of the variable Tobin’s Q is approximately 1.092. A high ratio implies that the firm is overvalued. The median of board independence is 0.762 (76.2%) with a standard error of 0.131, and each company has an independent board

Table 2. Descriptive statistics

Variable Obs Mean Median Std. error Min Max Tobin’s Q FROA BINDP BSINDP CRISIS BINDP*CRISIS BSIZE BFEM BMEETINGS FAGE FLEVERAGE FSIZE FPB 1741 1772 634 534 2822 634 644 439 626 1873 1413 1859 1756 1.698 -0.009 0.746 0.527 0.118 0.083 9.952 9.526 8.759 6.744 0.267 13.839 3.377 1.092 0.037 0.762 0.5 0 0 9 0.1429 8 6 0.192 13.624 2.32 1.703 0.170 0.131 0.194 0.322 0.236 3.4866 3.487 4.280 3.936 0.317 1.943 2.872 0.096 -0.488 0.3 0.0833 0 0 4 4 2 1 0 1.386 0.54 6.143 0.191 1 0.9231 1 1 35 0.55 44 16 6.25 18.934 11.42

(11)

member (the minimum is 0.30). The board size variable has a median of 9. This number of members is the optimal board size according to Lipton and Lorsch (1992), as mentioned earlier. The median of number of females in the board (BFEM) is 0.1429, which is considerably low compared to males. The median age of firms is 6.

The correlation matrix is given in Table 3 in the appendix. None of the

independent variables (except the interaction variable) have a correlation higher than 0.60, which means that multicollinearity may not be a problem.

Analysis

Studies that investigate board composition often face serious issues with

endogeneity. The endogeneity problem arises between firm performance and the structure of the board of directors. If there is endogeneity, then we cannot decide if there is a reverse causality (performance drives governance) or if there are other unobservable factors that affect corporate governance and performance (Bennedsen, Kongsted and Nielsen, 2008). By using a fixed-effect approach, the endogeneity problem is reduced. However, this does not solve the entire problem.

Results

Table 4 (on the next page) presents the OLS estimates of the model. All regressions have their robust standard errors in parentheses. The first regression (I) only includes the variables of interest for the hypotheses with no industry effect. The coefficient of board independence, board size and constant are significant in the first regression (I). The second regression (2) does include all the variables with no industry effect. In (2), the coefficient of board size, percentage of females on the board, price-to-book ratio, firm size and constant are significant. The third regression (3) includes industry effect relative to regression (2). The coefficient of strictly board independence, board size, percentage of females on the board, firm age, price-to-book ratio, firm size and constant are significant.

(12)

Board independence has a positive and significant effect on the firm performance in regression (I) (t=5.18, p < 0.01, one-sided). Thus, the marginal effect of board independence is positive since the interaction term is non-significant. Note that this result should be treated with caution because it is regressed with no control

variables.

Table 4. Regression results

Dependent variable _{Tobin’s Q}

(1) (2) (3) BINDP 2.741*** (0.529) -0.797 (0.573) -0.704 (0.571) BSINDP -0.616 (0.428) -0.472 (0.353) -0.637* (0.363) CRISIS -0.691 (1.382) -0.354 (1.021) -0.502 1.044 CRISIS*BINDP 1.100 (2.023) -0.024 (1.350) 0.293 (1.373) BSIZE -0.108*** (0.025) 0.056*** (0.018) 0.048** (0.019) BMEETINGS -0.010 (0.018) -0.001 (0.018) BFEM 2.724*** (1.022) 2.457** (1.019) FAGE 0.019 (0.018) 0.030* (0.017) FPB 0.185*** (0.036) 0.185*** (0.037) FLEVERAGE -0.307 (0.367) -0.340 (0.426) FSIZE -0.412*** (0.061) -0.375*** (0.062) Constant 0.832** 6.847*** 7.097***

Industry effects No No Yes

R^2 0.082 0.557 0.557

N 452 243 243

The dependent variable is Tobin’s Q. This table reports the regression results. The numbers are the corresponding coefficients, whereas numbers in brackets are robust standard errors. *, ** and *** denote significance at the 10%, 5% and 1% levels in a two-sides test respectively.

(13)

In all three regressions, the variable of board size is significant. In (1), the coefficient is negative (t(452)=-4.25, p < 0.01, one-sided), and in (2) (t(243)=3.07, p < 0.01, one-sided) and (3), it is positive (t(243)=2.48, p < 0.01, one-sided). The

coefficients are close to 0.

The findings of positive coefficients of board size are consistent with the findings of Adams and Mehran (2011). On the other side, findings of Yermack (1996) show that the effect of board size is negative in Tobin’s Q. The negative coefficient in (1) can be caused by the lack of the control variable: BFEM. A possible explanation for the positive coefficient is that in regression 2 and 3, the variable BFEM is

included. The correlation matrix shows that the correlation between BSIZE and BFEM is significantly negative, which means that these two variables move in the opposite direction. That means that if the board size increases, the percentage of females on the board decreases. However, BFEM and Tobin’s Q is positively

significant correlated. This last effect could be stronger than the negative correlation between BFEM and BSIZE so that the board size coefficient is positive for regression 2 and 3. Also, note that in regression (1), the constant is much smaller than in

regression (2) and (3). Note that it is difficult to assess why the coefficients are positive and negative because it also may due to endogeneity.

Next, the coefficient of the percentage of female board members is significant in regression (2). This finding is consistent with the fact that female directors have an impact on firm performance (Adams and Ferreira, 2009). Finally, the coefficient of firm size (ln of assets) is significant. This means that if the firm increases its assets, then the firm performance decreases. So, the firm’s size and complexity plays a role in the firm (Fama and Jensen, 1983).

The same regressions are executed again, but with returns of assets (ROA) as the dependent variable. Table 4 (on the next page) lists the results of this regression. In (1), the coefficient of the variables board independence, crisis, interaction term (CRISISxBINDP) and constant are significant. In (2), the coefficient of strictly board independence, firm age, price-to-book ratio, firm size and constant are significant. In (3), the coefficient of firm age, price-to-book ratio, firm size and constant are

significant.

All the variables of interest for testing the hypotheses are significant in regression (I) in Table 5. The coefficient of board independence (t(456)=-2.71, p < 0.01, one-sided) and the interaction term (t(456)=2.21, p < 0.05, one-sided) is significant. Thus, if we take the partial derivative of ROA with respect to board independence, the result is positive (-0.112 + 0.4 = 0.288). Note that (1) does not have control variables, which means that the result should be treated with caution. Board size is only significant in (1), and the coefficient is positive (t(456)=1.96, p < 0.05, one-sided) and close to zero.

(14)

Further, the firm age is in (2) and (3) negatively significant. This means that if the age of the firm increases, then the ROA decreases. A paper by Loderer and Waelchli (2010) showed that the growth in aging firms slows down. Also, the firm size is positively significant, which is the opposite result when regressing with Tobin’s Q.

Table 5. Regression results

Dependent variable ROA

(1) (2) (3) BINDP -0.112*** (0.041) 0.023 (0.058) 0.029 (0.593) BSINDP -0.011 (0.035) -0.092*** (0.050) -0.076 (0.049) CRISIS -0.283** (0.138) -0.057 (0.094) -0.099 (0.091) CRISIS*BINDP 0.400** (0.189) 0.034 (0.145) 0.077 (0.141) BSIZE 0.004* (0.002) -0.003 (0.002) -0.002 (0.002) BMEETINGS -0.001 (0.002) -0.002 (0.002) BFEM 0.125 (0.116) 0.130 (0.110) FAGE -0.006** (0.003) -0.007*** (0.003) FPB 0.006 (0.003) 0.007*** (0.003) FLEVERAGE -0.004 (0.042) -0.017 (0.049) FSIZE 0.033*** (0.007) 0.029*** (0.008) Constant 0.090*** -0.400*** -0.454***

Industry effects No No Yes

R^2 0.027 0.207 0.247

N 456 240 240

The dependent variable is Return on Assets (ROA). This table reports the regression results. The numbers are their corresponding coefficient, whereas numbers in brackets are robust standard errors. *, ** and *** denote significance at the 10%, 5% and 1% levels in a two-sides test respectively.

(15)

Discussion

The results of this paper face some limitations. The sample consists of a total of 166 firms with variables over a long time span. However, we must keep in mind that variables are missing over this long time span.

Next, there is the discussion of endogeneity. I have used the fixed effects method by keeping the industry of the firm constant. Other researchers have found this to reduce bias (Linck, Netter and Yang, 2008). Nevertheless, it is suggested that there are different estimations methods, such as instrumental variable regression / 2SLS, that could reduce the bias even further and eventually provide more reliable results.

Conclusion

This paper has investigated the role of board independence, size and different control variables on firm performance. The main question was, “Does board independence lead to a higher valuation of public firms during the financial crisis (2007–2009) in the United States?”

Although there are some limitations due to potential endogeneity, I conclude the following points. First, the marginal effect of board independence during the crisis was found to be positive. However, this result was not clear due to the lack of control variables in the regression. Second, a significant positive and negative effect was found on the board size. Finally, the percentage of females on the board, firm age and assets seem to play a role in firm performance.

Overall, the results of this paper seem to be mixed. Further research is

needed for regarding board size and independence, especially in terms of increasing the reliability of the regression results.

(16)

Bibliography

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

Adams, R., & Mehran, H. (2011, November). Corporate Performance, Board Structure, and Their Determinants in the Banking Industry. Federal Reserve Bank of New York Staff Reports, 330.

Aebi, V., Sabato, G., & Schmid, M. (2012, December). Risk management, corporate governance, and bank performance in the financial crisis. Journal of Banking & Finance, 36(12), 3213-3226.

Baysinger, B., & Butler, H. (1985). Corporate Governance and the Board of Directors: Performance Effects of Changes in Board Composition. Journal of Law,

Economics, & Organization, 1(1), 101-124.

Bennedsen, M., Kongsted, H., & Nielsen, K. (2008). The causal effect of board size in the performance of small and medium-sized firms. Journal of Banking & Finance, 32(6), 1098-1109.

Bruno, V., & Claessens, S. (2010, November). Corporate Governance and

Regulation: Can There be too Much of a Good Thing? Journal of Financial Intermediation, 19(4), 461-482.

Chodorow-Reich, G. (2014, February). The employment effects of credit market disruptions: firm-level evidence from the 2008-9 financial crisis. The Quarterly Journal of Economics, 129(1), 1-60.

Duchin, R., Matsusaka, J., & Oguzhan, O. (2010, May). When are outside directors effective? Journal of Financial Economics, 96(2), 195-214.

Eisenberg, T., Sundgren, S., & Wells, M. (1998). Larger board size and decreasing firm value in small firms. Journal of Financial Economics, 48, 35-54.

Erkens, D., Hung, M., & Matos, P. (2012). Corporate governance in the 2007–2008 financial crisis: Evidence from financial institutions worldwide. Journal of Corporate Finance, 18, 389-411.

Fama, E., & Jensen, M. (1983, June). Separation of Ownership and Control. The Journal of Law & Economics, 26(2).

Francis, B., Hasan, I., & Wu, Q. (2012, April). Do corporate boards matter during the current financial crisis? Review of Financial Economics, 21(2), 39-52.

Hambrick, D., & D'Aveni, R. (1992, November). Top Team Deterioration As Part of the Downward Spiral of Large Corporate Bankruptcies. Management Science, 38(10), 1445-1466.

Hermalin, B., & Weisbach, M. (1991). The Effects of Board Composition and Direct Incentives on Firm Performance. Financial Management, 20, 101-112.

(17)

Jensen, M. (1993). The Modern Industrial Revolution, Exit, and the Failure of Internal Control System. Journal of Finance, 48, 831-880.

Jensen, M., & Meckling, W. (1976, November). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305-360.

Johnson, R., Hoskisson, R., & Hitt, M. (1993, June). Board of director involvement in restructuring: The effects of board versus managerial controls and

characteristics. Strategic Management Journal, 14(S1), 33-50.

Johnson, S., Schantterly, K., & Hill, A. (2013, January). Board Composition Beyond Independence: Social Capital, Human Capital, and Demographics. Journal of Management, 39(1), 232-262.

Judge, W., & Zeithaml, C. (1992, November). Institutional and Strategic Choice Perspectives on Board Involvement in the Strategic Decision Process. The Academy of Management Journal, 35(4), 766-794.

Lipton, M., & Lorsch, J. (1992, November). A Modest Proposal for Improved Corporate Governance. The Business Lawyer, 48(1), 59-77.

Liu, Y., Miletkov, M., Wei, Z., & Yang, T. (2015, February). Board independence and firm performance in China. Journal of Corporate Finance, 30, 223-244.

Loderer, C.F. & Waelchli, U. Firm Age and Performance (April 30, 2010). Available at SSRN: https://ssrn.com/abstract=1342248

Pfeiffer, J., & Salancik, G. (1978). The External Control of Organizations: A Resource Dependence Perspective. New York: Haper & Row.

Rosenstein, S., & Wyatt, J. (1990). Outside directors, board independence, and shareholder wealth. Journal of Financial Economics, 26(2), 175-191. Smirlock, M., Gilligan, T., & Marshall, W. (1984). Tobin's q and the

Structure-Performance Relationship. American Economic Review, 74(5), 1051-1060. Vafeas, N. (1999, July). Board meeting frequency and firm performance. Journal of

Financial Economics, 53(1), 113-142.

Weisbach, M. (1988). Outside directors and CEO turnover. Journal of Financial Economics, 20(1-2), 431-460.

Yermack, D. (1996, February). Higher market valuation of companies with a small board of directors. Journal of Financial Economics, 40(2), 185-211.

(18)

Appendix

Appendix A – Variable description Table 1. Variable description in research model

Variable name Description Source

Tobin’s Q Tobins’ Q calculated as: Market value of Assets / Bookvalue of assets.

Datastream. Calculated: market value of assets/ bookvalue of assets

FROA Return on assets. Datastream

BINDP Board independence calculated as the fraction of independent board members on total board size.

Datastream

BSINDP BSINDP stands for strictly board independence. Percentage of strictly independent board members. This variable is from Datastream and has stronger assumptions than BINDP.

Datastream

CRISIS Year-dummy variable. It equals to [1] if variable is in crisis year 2007-2008 and [0] for the other years.

Stata. Variable generated.

CRISIS*BINDP Interaction term with variables crisis and BINDP.

Stata and Datastream. Variable generated. BSIZE Board size. The amount of members

in a board at the end of the fiscal year.

Datastream

BMEETINGS The amount of total board meetings at the end of the fiscal year.

Datastream

BFEM Percentage of women on the board of directors.

Datastream

FAGE The firm’s age in years. Export firms from Datastream and you will get start date of the firm. Calculated using excel

FPB The price-to-book ratio of a firm. Datastream FLEVERAGE The leverage of the firm. It is the ratio

of total debt and total assets.

Datastream. Calculated: Ratio of total debt and total assets FSIZE The size of the firm measured by the

ln of assets.

Datastream. Data transformed into ln.

(19)

Appendix B – Correlation matrix

Table 3. Correlation matrix.

1 2 3 4 5 6 7 8 9 10 11 12 13 1 Tobin’s Q 1 2 FROA -0.180* (0.0) 1 3 BINDP 0.186* (0.0) -0.091* (0.037) 1 4 BSINDP -0.003 (0.947) -0.03 (0.523) 0.351 (0.0) 1 5 CRISIS -0.503* (0.036) -0.039 (0.101) -0.042 (0.293) 0.186* (0.0) 1 6 BINDP*CRISI S -0.036 (0.410) 0.031 (0.471) 0.012 (0.769) 0.206* (0.0) 0.985* (0.0) 1 7 BSIZE 0.183* (0.0) 0.044 (0.309) 0.020 (0.609) 0.108* (0.013) 0.013 (0.744) 0.022 (0.574) 1 8 BMEETINGS -0.186* (0.0) 0.077 (0.077) 0.098* (0.015) 0.071 (0.100) 0.019 (0.632) 0.014 (0.721 0.075 (0.066 1 9 BFEM 0.239* (0) 0.124* (0.016) 0.224* (0) 0.248* (0) 0.044 (0.356) 0.037 (0.447) -0.201* (0) -0.015 (0.755) 1 10 FAGE 0.049* (0.048) -0.018 (0.460) 0.298* (0) -0.277* (0) -0.286* (0) -0.315 (0) -0.044 (0.305) -0.017 (0.696) 0.033 (0.522) 1 11 FPB 0.656* (0) -0.097* (0) 0.211* (0) 0.135* (0.004) -0.055* (0.022) 0.001 (0.982) -0.084 (0.530) -0.111* (0.012) 0.161* (0.002) 0.04 (0.102) 1 12 FLEVERAGE 0.005 (0.858) -0.136* (0) 0.128* (0.006) 0.005 (0.921) 0.052 (0.052) 0.010 (0.825) -0.130* (0.005) 0.043 (0.363) -0.121 (0.029) 0.079* (0.005) 0.102* (0.0) 1 13 FSIZE -0.550* (0) 0.374 (0) -0.243* (0) 0.107* (0.019) 0.061* (0.019) 0.102* (0.009) 0.399* (0.016) 0.166* (0) -0.075 (0.139) 0.109* (0.0) -0.257* (0) -0.075* (0.006) 1 * denotes significance at the level of 5%. The numbers in brackets are their corresponding p-value.

(20)

Appendix C – Descriptive statistics output max .5455 16 11.42 6.25 18.93419 min .037 1 .54 .0000191 1.386294 sd .0870844 3.93585 2.872209 .3168335 1.942503 p50 .1429 6 2.32 .1920417 13.62405 mean .1724296 6.743727 3.376712 .2668215 13.83943 N 439 1873 1746 1413 1859 stats BFEM FAGE FPB FLEVER~E FSIZE

max 6.143049 .1909 1 .9231 1 1 35 44 min .0956544 -.4881 .3 .0833 0 0 4 2 sd 1.703278 .1696058 .1314928 .1936434 .3222468 .2355684 3.4866 4.279547 p50 1.092143 .03705 .76215 .5 0 0 9 8 mean 1.698316 -.0089758 .7461886 .5274287 .1176471 .0829975 9.526398 8.758786 N 1741 1772 634 534 2822 634 644 626 stats TobinsQ FROA BINDP BSINDP CRISIS CRISIS~P BSIZE BMEETI~S

(21)

Appendix D – Regression output . _cons .8320897 .3801537 2.19 0.029 .0849747 1.579205 BSIZE -.1080393 .0254392 -4.25 0.000 -.1580349 -.0580436 CRISISxBINDP 1.100461 2.022568 0.54 0.587 -2.874486 5.075407 CRISIS -.6909933 1.381758 -0.50 0.617 -3.406558 2.024571 BSINDP -.6164765 .4283898 -1.44 0.151 -1.45839 .2254367 BINDP 2.740566 .528611 5.18 0.000 1.701688 3.779444 TobinsQ Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = 1.5594 R-squared = 0.0815 Prob > F = 0.0000 F(5, 446) = 9.60

Linear regression Number of obs = 452

. xi: regress TobinsQ BINDP BSINDP CRISIS CRISISxBINDP BSIZE, robust

. _cons 6.8468 1.031078 6.64 0.000 4.815281 8.878319 FSIZE -.4114951 .0614353 -6.70 0.000 -.5325404 -.2904499 FLEVERAGE -.3065258 .3666556 -0.84 0.404 -1.028942 .4158907 FPB .1847459 .035968 5.14 0.000 .1138786 .2556132 FAGE .0188929 .0181505 1.04 0.299 -.0168688 .0546545 BFEM 2.723654 1.022489 2.66 0.008 .7090579 4.738251 BMEETINGS -.0095763 .0179314 -0.53 0.594 -.0449063 .0257537 BSIZE .0555734 .0180762 3.07 0.002 .0199581 .0911887 CRISISxBINDP -.0236283 1.34963 -0.02 0.986 -2.682786 2.635529 CRISIS -.353962 1.020594 -0.35 0.729 -2.364825 1.656901 BSINDP -.4720697 .3530491 -1.34 0.182 -1.167678 .2235382 BINDP -.7966054 .5734923 -1.39 0.166 -1.92655 .3333388 TobinsQ Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .94696 R-squared = 0.5301 Prob > F = 0.0000 F(11, 231) = 13.89

> AGE FSIZE, robust

(22)

_cons 7.096676 1.115077 _IINDNUM_8 -.7977771 .7490873 _IINDNUM_7 -.8591915 .7568696 _IINDNUM_6 -.7782822 .7593457 _IINDNUM_5 -1.4591 .7457949 _IINDNUM_4 -.893511 .8257416 _IINDNUM_3 -.9875449 .7505349 _IINDNUM_2 1.240139 1.371115 FSIZE -.3751487 .0624278 FLEVERAGE -.3395975 .4255737 FPB .1849907 .0369909 FAGE .0297382 .0165732 BFEM 2.457105 1.018622 BMEETINGS -.0009216 .0184373 BSIZE .0482459 .0194197 CRISISxBINDP .2930244 1.373384 CRISIS -.5021279 1.043684 BSINDP -.6369757 .3626665 BINDP -.7038852 .5706022 TobinsQ Coef. Std. Err. Robust Root MSE = .93361 R-squared = Prob > F = F(18, 224) = 6.36 0.000 4.899294 9.294059 -1.06 0.288 -2.273937 .6783826 -1.14 0.258 -2.350687 .632304 -1.02 0.306 -2.274657 .7180927 -1.96 0.052 -2.928772 .0105715 -1.08 0.280 -2.520726 .7337045 -1.32 0.190 -2.466557 .4914674 0.90 0.367 -1.461795 3.942074 -6.01 0.000 -.4981697 -.2521278 -0.80 0.426 -1.178238 .4990426 5.00 0.000 .112096 .2578853 1.79 0.074 -.0029211 .0623975 2.41 0.017 .4497977 4.464411 -0.05 0.960 -.0372543 .035411 2.48 0.014 .0099772 .0865146 0.21 0.831 -2.413381 2.99943 -0.48 0.631 -2.558823 1.554567 -1.76 0.080 -1.35165 .0776989 -1.23 0.219 -1.82832 .4205498 t P>|t| [95% Conf. Interval] 0.5571 0.0000 11.24

Linear regression Number of obs = 243 i.INDNUM _IINDNUM_1-8 (naturally coded; _IINDNUM_1 omitted)

> AGE FSIZE i.INDNUM, robust

. xi: regress TobinsQ BINDP BSINDP CRISIS CRISISxBINDP BSIZE BMEETINGS BFEM FAGE FPB FLEVER

_cons .0896306 .0300032 2.99 0.003 .0306668 .1485944 BSIZE .0036246 .0018507 1.96 0.051 -.0000125 .0072617 CRISISxBINDP .3998703 .1885167 2.12 0.034 .0293879 .7703527 CRISIS -.2833638 .1376829 -2.06 0.040 -.5539451 -.0127826 BSINDP -.0112572 .0347292 -0.32 0.746 -.0795087 .0569942 BINDP -.1120086 .0413568 -2.71 0.007 -.193285 -.0307322 FROA Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .12791 R-squared = 0.0270 Prob > F = 0.0124 F(5, 450) = 2.95

(23)

. _cons -.4002556 .118062 -3.39 0.001 -.6328878 -.1676235 FSIZE .0325724 .0074431 4.38 0.000 .0179063 .0472385 FLEVERAGE -.0039214 .0421038 -0.09 0.926 -.0868837 .0790408 FPB .0056674 .0032133 1.76 0.079 -.0006642 .011999 FAGE -.0059317 .0025255 -2.35 0.020 -.010908 -.0009554 BFEM .12484 .1160365 1.08 0.283 -.103801 .353481 BMEETINGS -.0012134 .0017947 -0.68 0.500 -.0047496 .0023229 BSIZE -.0024701 .0016972 -1.46 0.147 -.0058143 .0008742 CRISISxBINDP .0339906 .1449494 0.23 0.815 -.2516212 .3196023 CRISIS -.0573055 .0940794 -0.61 0.543 -.2426816 .1280707 BSINDP -.0915341 .0501925 -1.82 0.070 -.1904345 .0073663 BINDP .0229603 .0575496 0.40 0.690 -.0904368 .1363573 FROA Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .12082 R-squared = 0.2073 Prob > F = 0.0016 F(11, 228) = 2.84

> robust

. xi:regress FROA BINDP BSINDP CRISIS CRISISxBINDP BSIZE BMEETINGS BFEM FAGE FPB FLEVERAGE FSIZE,

_cons -.4538905 .1287231 -3.53 0.001 -.7075724 -.2002086 _IINDNUM_8 .0767634 .0719714 1.07 0.287 -.0650746 .2186015 _IINDNUM_7 .1020639 .0726935 1.40 0.162 -.0411974 .2453251 _IINDNUM_6 .1389214 .0712966 1.95 0.053 -.0015869 .2794297 _IINDNUM_5 .1959238 .0726893 2.70 0.008 .0526708 .3391767 _IINDNUM_4 .0906199 .0912176 0.99 0.322 -.0891477 .2703875 _IINDNUM_3 .1075966 .0710865 1.51 0.132 -.0324975 .2476908 _IINDNUM_2 -.0885706 .1739535 -0.51 0.611 -.4313906 .2542494 FSIZE .0294972 .0078319 3.77 0.000 .0140624 .0449319 FLEVERAGE -.0170507 .048983 -0.35 0.728 -.1135842 .0794828 FPB .0067272 .0034809 1.93 0.055 -.0001329 .0135872 FAGE -.0065879 .0022622 -2.91 0.004 -.0110462 -.0021296 BFEM .1300778 .1098733 1.18 0.238 -.0864558 .3466113 BMEETINGS -.0017949 .0018318 -0.98 0.328 -.005405 .0018152 BSIZE -.0021269 .0019037 -1.12 0.265 -.0058787 .0016248 CRISISxBINDP .0766316 .1406347 0.54 0.586 -.2005251 .3537883 CRISIS -.0994452 .091001 -1.09 0.276 -.2787859 .0798955 BSINDP -.0757414 .0486656 -1.56 0.121 -.1716494 .0201666 BINDP .0291718 .0593217 0.49 0.623 -.0877368 .1460805 FROA Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .11961 R-squared = 0.2469 Prob > F = 0.0016 F(18, 221) = 2.39

Linear regression Number of obs = 240 i.INDNUM _IINDNUM_1-8 (naturally coded; _IINDNUM_1 omitted)

> SIZE i.INDNUM, robust

(24)

Appendix E – Correlation matrix output 0.0001 0.1391 0.0000 0.0000 0.0060 FSIZE 0.1656* -0.0749 0.1089* -0.2569* -0.0750* 1.0000 0.3630 0.0291 0.0048 0.0002 FLEVERAGE 0.0430 -0.1214* 0.0791* 0.1022* 1.0000 0.0122 0.0020 0.1022 FPB -0.1097* 0.1614* 0.0400 1.0000 0.6961 0.5218 FAGE -0.0171 0.0334 1.0000 0.7549 BFEM -0.0153 1.0000 BMEETINGS 1.0000 BMEETI~S BFEM FAGE FPB FLEVER~E FSIZE

0.0000 0.0000 0.0000 0.0192 0.0087 0.0160 0.0000 FSIZE -0.5503* 0.3737* -0.2430* 0.1073* -0.0608* 0.1020* 0.3987* 0.8579 0.0000 0.0061 0.9213 0.0523 0.8245 0.0050 FLEVERAGE 0.0050 -0.1362* 0.1281* 0.0050 -0.0516 0.0104 -0.1298* 0.0000 0.0001 0.0000 0.0042 0.0222 0.9821 0.0530 FPB 0.6564* -0.0974* 0.2106* 0.1346* -0.0547* 0.0010 -0.0837 0.0476 0.4604 0.0000 0.0000 0.0000 0.0000 0.3046 FAGE 0.0485* -0.0184 0.2977* -0.2768* -0.2858* -0.3151* -0.0444 0.0000 0.0162 0.0000 0.0000 0.3564 0.4472 0.0000 BFEM 0.2390* 0.1241* 0.2235* 0.2483* 0.0441 0.0369 -0.2010* 0.0000 0.0774 0.0149 0.1003 0.6322 0.7206 0.0659 BMEETINGS -0.1866* -0.0773 0.0981* 0.0713 0.0192 0.0144 0.0738 0.0000 0.3090 0.6091 0.0129 0.7444 0.5741 BSIZE -0.1830* 0.0439 0.0204 0.1076* 0.0129 0.0224 1.0000 0.4103 0.4712 0.7688 0.0000 0.0000 CRISISxBINDP -0.0359 0.0313 0.0117 0.2064* 0.9851* 1.0000 0.0359 0.1014 0.2931 0.0000 CRISIS -0.0503* -0.0389 -0.0418 0.1855* 1.0000 0.9472 0.5226 0.0000 BSINDP -0.0031 -0.0300 0.3512* 1.0000 0.0000 0.0365 BINDP 0.1863* -0.0908* 1.0000 0.0000 FROA -0.1800* 1.0000 TobinsQ 1.0000

TobinsQ FROA BINDP BSINDP CRISIS CRISIS~P BSIZE . pwcorr TobinsQ-FSIZE, star(0.05) sig

The size and independency of the board of directors during the crisis in the U.S.

Bachelor Thesis