Empirical evidence on corporate governance, operating performance and investors’ expectations in Europe

(1)

Empirical evidence on corporate governance, operating performance

and investors’ expectations in Europe

Janita H. Keuter S1409247

October 2008

Supervisor: Dr. T.A. Marra

(2)

Abstract

In this thesis the finding of Bauer, Guenster and Otten (2004) that poorly governed European firms have lower stock market performance is examined. Firstly, I test whether a relationship exists between corporate governance and operating performance. The results show a positive relation between corporate governance and operating performance. Secondly, I test whether this underperformance of poorly governed firms surprises the market. The results suggest that the market is surprised by the difference in operating performance between well-governed firms and poorly governed firms. In sum, these results suggest that the unawareness of shareholders of European companies about the relationship between corporate governance and operating performance is at least part of the reason of the return difference found by Bauer et al.

(3)

Table of Content

1. Introduction ...p.4 2. Literature review ...p.7 3. Data ...p.11

(4)

1. Introduction

After corporate collapses like Enron and World Com, the hype around corporate governance increased. Investors need to keep a close eye on the way that board of directors keep management in check and ensure financial disclosure, board independence and shareholder rights.

The principal-agent theory is generally considered as the starting point for any discussion on corporate governance. Agency theory concerns the relationship between a principal (shareholder) and an agent of the principal (company's managers). It involves the costs of resolving conflicts between the principals and agents and aligning interests of the two groups. Jensen and Meckling (1976) argue that better-governed firms might have more efficient operations, resulting in a higher expected future cash-flow stream. Therefore, they assume a positive relationship between corporate governance and a firm’s operating performance. The relationship between corporate governance and operating performance will be the topic of this thesis.

Corporate governance can be defined in several ways. In the broad version the term covers the relationship between all stakeholders in a company. This includes the shareholders, directors and management of a company, as defined by the corporate charter, bylaws, formal policy and rule of law (02-10-2008, www.investopedia.com). In this context corporate governance is the way in which directors handle their responsibilities towards shareholders and other company stakeholders. The narrow definition of corporate governance focuses on the relationship between shareholders and management. In this narrow version corporate governance deals with the ways in which suppliers of finance to corporations assure themselves of getting a return on their investment (Shleifer and Vishny, 1997). The focus of this thesis will be on this narrow version. Typical corporate governance arrangements include appointing non-executive directors, placing constraints on management power and ownership concentration, as well as ensuring proper disclosure of financial information and executive compensation. The better corporate governance is arranged, the more management will be stimulated to take those actions that are in the best interest of the shareholders. Eventually, this will lead to higher cash flow to the shareholders.

(5)

A second difference in business context is the identity of the shareholders. In Anglo-Saxon countries most of the shares are in hands of the agents of financial institutions. Due to regulations in Anglo-Saxon countries financial institutions are not allowed to hold shares of public companies on their own behalf. In contrast, in Continental Europe private persons and companies act directly and do not use agents to manage their affairs. Moreover, insurance companies, investment funds and pension funds have important shareholdings in the financial institutions in Continental Europe, but not in Anglo-Saxon countries.

A third difference concerns the liquidity in the financial market. In Anglo-Saxon countries the shares of many companies are publicly traded. As a consequence, many companies have little personal contact with their owners. In Continental Europe many firms are private. Consequently, the relationship between management and owners is much stronger and more personal. Often, managers and owners are the same.

A fourth difference concerns the board structure. In the Anglo-Saxon countries companies have a one-tier board. This means that the board comprises of both executive and non-executive members. In Continental Europe most companies have a two-tier board, which means that the executive and non-executive members take seat in separate boards. The main advantage of a one-tier board is that the board is well informed by the executives about the company. The main disadvantage is the lack of independence. Apparently, the opposite holds for a two-tier board.

At last, the ownership structure in Continental Europe is not as transparent as in Anglo-Saxon countries. This is due to the number of mutual shareholdings and the limited extent of information disclosure in Continental Europe, whereas in Anglo-Saxon countries regulations exist to limit the complexity of the ownership structure.

The differences in the business context are the main causes for the differences between the corporate governance models. Both models have their own governance problems. Due to the low concentration of shareholders in Anglo-Saxon countries, management is left with great decision power. These decisions may be in their own best interest, which gives rise to over-investment. Also, if voting power is dispersed, as is in the Anglo-Saxon countries, free riding may occur. In this case, the cost of control exceeds the benefits and shareholders tend not to take action. As a result, management will have dominant power. Moreover, management must report once a year to the shareholders. Management can choose to make this report look good by boosting performance in the short run, even if this is at the expense of the long run. So, great management power, over-investments, free-riding problem and short-term focus are problems that may occur in the Anglo-Saxon governance model.

(6)

company to a related, but badly performing company. So, limited financial resources and movement of cash flows are problems that may occur in the Continental European model.

In this thesis I will investigate the relationship between corporate governance and operating performance in Europe. In addition, I will test whether or not the market is surprised by this potential relationship. If the market is surprised, this will explain some of the return difference between well-governed firms and poorly well-governed firms as found by Bauer et al. (2004). The ultimate question of my thesis is: Does underperformance caused by poor corporate governance surprise the market? This thesis follows the article of Core et al. (2006). However, Core et al. focus on the United States whereas my thesis focuses on Europe.

(7)

2. Literature Review

As stated by Gompers et al. (2003), shareholder rights can have both positive and negative effects on the operating performance of a company. On the one hand, weak shareholder rights can prevent the shareholders to remove incompetent management. This might result in lower operating performance than would have been achieved in case the incompetent management would have been replaced by competent management. On the other hand, weak shareholder rights may give managers job security, which will reduce managerial myopia. Managerial myopia is the need to signal quality by boosting short-term performance at the expense of long-term value. Moreover, if managers have a certain amount of job security, managers may be protected from the lower-tail outcomes of good projects and therefore become less risk averse. So, weak shareholder rights may also lead to higher operating performance.

The relationship between shareholder rights and performance has been a topic of research for many years. Baliga et al. (1996) examined the link between duality (chairman of the board and CEO are the same individual) and corporate performance. They state that despite limited empirical evidence duality has been blamed, in many cases, for the poor performance and failure of firms to adapt to a changing environment. However, their results suggest that the market is indifferent to changes in a firm's duality status. There is little evidence of operating performance changes around changes in duality status and there is only weak evidence that duality status affects long-term performance. So, according to Baliga et al. the link between duality and corporate performance is weak.

Bauer et al. (2004) examined the impact of corporate governance on firm valuation. The results show a positive relationship between firm value variables and corporate governance. In addition, the relationship between corporate governance and operating performance is analysed, as approximated by net profit margin and return on equity. A negative relationship was found between governance standards and these earnings-based performance ratios. According to Bauer et al. a possible explanation for this finding, that is contrary to their expectations, could be that accounting numbers are biased measures of firm performance. A negative correlation between earnings and corporate governance possibly implies that badly governed firms report less conservative earnings. However, there is no evidence for this possible explanation.

(8)

Chiang (2005) examined the relationship between information transparency and operating performance (measured by return on assets (ROA), return on equity (ROE) and earnings per share (EPS)). Chiang’s results indicate that board size, board ownership, institutional ownership, financial transparency, information disclosure, and board and management structure and process have a significant positive relationship with operating performance. Hence, Chiang’s results show that good corporate governance has a positive impact on operating performance.

Lyn et al. (2005) studied the relationship between corporate governance and financial distress and found that more entrenched management increases the likelihood of experiencing financial distress. Furthermore, more entrenched management firms are less likely to emerge from bankruptcy and re-file in the following years. So, according to Lyn et al. weak shareholder rights cause lower firm performance.

Baghat and Bolton (2008) found that better governance, stock ownership of board members, and CEO-chair separation is significantly positively correlated with operating performance. However, none of the governance measures were correlated with future stock market performance. Also, given poor firm performance, the probability of disciplinary management turnover is positively correlated with stock ownership of board members, and board independence. However, better-governed firms are less likely to experience disciplinary management turnover in spite of their poor performance.

Overall, a large body of literature exists on the link between corporate governance and operating performance. Nonetheless, the results are contradictory at times. To examine whether a relationship between corporate governance and operating performance exists in Europe, I will test the following hypothesis for Europe:

H1: Corporate governance influences the operating performance of a firm.

(9)

do not include provisions that apply automatically under state law. Thus, these data are supplemented with state-level data on takeover laws as given by another IRRC publication. From this publication Gompers et al. code the presence of six types of so-called “second-generation” state takeover laws and place them in the State group (Gompers et al. 2003, p.111).

The index construction is straightforward: for every firm one point for every provision that restricts shareholder rights is added (increases managerial power). Thus, the Governance Index (“G”) is just the sum of points for the existence (or absence) of each provision. Gompers et al. pay special attention to the two extreme portfolios: the “Dictatorship Portfolio” and the “Democracy Portfolio”. The “Dictatorship Portfolio” consists of the firms with the weakest shareholder rights (G > 13), and the “Democracy Portfolio” consists of the firms with the strongest shareholder rights (G < 6).

Moreover, Gompers et al. examined the relationship between this governance index and the subsequent returns. They examined the differences in return between the Dictatorship Portfolio and the Democracy Portfolio and it turned out that the Democracy Portfolio outperformed the Dictatorship Portfolio. An investment strategy that bought firms in the lowest decile of the index (strongest rights) and sold firms in the highest decile of the index (weakest rights) would have earned abnormal returns of 8.5 percent per year during the sample period. In addition, they found that firms with stronger shareholder rights had higher firm value, higher profits, and higher sales growth, lower capital expenditures and made fewer corporate acquisitions.

Bauer et al. (2004) repeated this study for Europe based on the Deminor Corporate Governance Rating for companies included in the FTSE Eurotop 300. As done by Gompers et al. (2003) they built portfolios consisting of well-governed firms and poorly governed firms and compared their performances. They found that well-governed firms outperform the poorly governed firms by 1.6% in Europe.

As an extension of Gompers et al. (2003), Core et al. (2006) investigated why well-governed firms outperform poorly governed firms in the US. They came up with two possible causes. First, poorly governed firms have lower operating performance and this surprises the market, which causes a decline in stock prices. Second, poorly governed firms have a higher take-over protection. When the market realises this, it recognises that this reduces the chance of a take-over premium. As a consequence, the share price will drop. Core et al. examined both of these potential hypotheses. However, they do not find evidence for any of these two hypotheses.

(10)

governance practices. I will examine if corporate governance has a significant influence on the operating performance of the firm. In addition, I study whether underperformance at poorly governed firms surprises the market so that these firms have lower returns around earnings announcements. Therefore, the second hypothesis of this paper is the following:

(11)

3. Data

This section provides a description of the data. Because my research contains two hypotheses, the data description will be in two parts as well. First, I will provide a data description concerning the association between corporate governance and operating performance. Second, I will describe the data about the earnings announcement returns.

3.1 Operating performance

The part of the research with respect to the operating performance concerns several variables. These variables will be described as dependent, independent and control variables, respectively.

Dependent variable

The dependent variable in my research is the operating performance. The operating performance shows the performance of a company from the firm’s normal business operations. This does not include, for example, any profit earned from the firm’s financial investments and the effects of interest and taxes. To measure the operating performance four ratios are used in this thesis, namely return on assets (ROA), return on equity (ROE), earnings per share (EPS) and net profit margin (NPM). ROA, ROE and NPM are used, because these are also used in the basis articles of this research Gompers et al. (2003), Bauer et al. (2004) and Core et al. (2006). EPS is added, because a significant relation exists between EPS and governance indicators according to Chiang (2005). Table 1 shows the descriptive statistics from the operating performance analysis.

The ROA is defined as the earnings before interest and taxes for the year divided by the book value of total assets at the end of the year. Table 1 shows that the average ROA is 0.07. The lowest ROA in the sample can be attributed to the Dutch company Numico in 2002. The highest ROA can be attributed to the Italian company Assicurazioni Generali in 2002.

The ROE is defined as net profit for the year divided by the book value of equity at the end of the year. The average ROE equals 0.11. The lowest ROE can be attributed to Numico. Numico had this low ROE, because of a relatively low equity-to-assets value despite a huge loss in 2002. This causes the ROE to be exceptionally low.

EPS is defined as the net profit for the year divided by the number of outstanding shares at the end of the year. The average EPS is €1.46. The highest and the lowest EPS can be attributed to The Deutsche Bank in 2007 and Vivendi Universal in 2002, respectively.

(12)

Table 1: Descriptive statistics operating performance analysis Independent

variable

Dependent variables Control variables

G-Score ROA ROE EPS NPM MVE

(mln*) BME GDP (th) CCG CI Average 5.43 0.07 0.11 1.46 -0.27 14,356 0.65 30.6 6.43 8.4 Variance 2.32 0.01 0.42 9.91 17.09 1,442,505 ,065,338 0.84 18.0 0.41 0.8 Minimum 1.80 -0.42 -8.27 -21.40 -57.73 7 -2.04 20.7 4.62 5.2 Maximum 8.76 0.82 2.08 13.65 1.86 344,221 11.00 41.1 7.40 9.7 Median 5.60 0.07 0.16 1.30 0.06 2,113 0.47 29.5 6.69 8.7 Mode 5.60 0.09 0.11 1.13 0.16 42,692 0.14 29.5 6.69 8.7 Kurtosis -0.64 11.04 113.67 18.19 175.36 39.78 77.49 -0.01 3.04 3.25 Skewness -0.30 1.12 -9.17 -1.39 -12.93 5.76 7.22 0.64 -2.08 -2.01 Number of observations 290 232 260 221 219 214 214 290 290 290

*This column is in millions, except for the last three (kurtosis, skewness and number of observations)

Independent variable

The independent variable in this research is corporate governance. Several corporate governance indices exist, however most of these indices are only available for free for US firms. The ones that are available for European companies are not all for free. The ones that are available for free (partly) are VEB, Deminor and Ceres.

VEB is a Dutch organisation that protects the interests of shareholders. VEB values the corporate governance of approximately 70 companies headquartered in the Netherlands. VEB bases the score on two main sub-indices, namely the rights of the shareholders and the degree that these rights can be executed. VEB values the companies on a 10-point scale. This score is free and available for the years 2004, 2005, 2006 and 2007. This research contains 232 VEB ratings.

The Deminor Group Company values the corporate governance of European firms included in the FTSE Eurotop 300. It creates five sub-indices namely corporate governance commitment, disclosure on governance, rights and duties of shareholders, absence of takeover defences and board structure and functioning. The weights that are assigned to these sub indices are 10%, 15%, 31%, 15% and 29% respectively. The companies are valued on a 10-point scale. Although 300 European companies are valued by Deminor, the score is only available for free for 50 companies in 2002. Therefore, these 50 companies are part of my research.

(13)

Because the Deminor rating and the Ceres rating have largely common elements and are both on a 10-point scale, I assume these ratings to be similar and will use them interchangeably. Although the VEB rating contains more general elements than the Deminor and Ceres rating, the main focus of the VEB rating is similar to the Deminor and Ceres rating. Therefore, all three ratings will be used interchangeably. However, a sensitivity analysis will be added to distinguish between these ratings. Table 1 shows that the scores vary between 1.80 and 8.76 and the average score equals 5.43. The highest score can be attributed to Nokia in 2002 and the lowest score to Fugro in 2002.

Control Variables

Several control variables are used in this study. Control variables to correct for firm size are added, because these are correlated with shareholder rights and profitability, see for example, Fama and French (1995). The market value of equity (MVE) is calculated as the stock price times the number of outstanding shares at year-end. The book-to-market of equity (BME) is calculated as the book value of equity divided by the market value of equity.

My research contains nine European countries, whereas Core et al. (2006) study one country, the US. The distribution of companies over the countries in my research can be found in Appendix 1. Because my research contains several European countries, some control variables are added to correct for country differences. First, the gross domestic product per capita (GDP per capita) is added, because the operating performance most likely depends on the overall economic situation of a country. Spain in 2002 has the lowest GDP per capita, whereas Switzerland in 2007 has the highest one. The average GDP per capita is €30.6 thousand.

Second, the country corporate governance index (CCG) is added to my regression analysis. Gompers et al. (2003) did not only value corporate governance of US firms, but also rated country corporate governance. To the extend that the operating performance of a company depends on the degree of corporate governance of a company, it will also depend on the overall governance situation of a country as well. Moreover, the variance of the corporate governance index between European countries is quit large. So it is reasonable to assume that country corporate governance is of influence. Consequently, this CCG index is added to my research. Table 1 shows that the average CCG score equals 6.43, the highest one can be attributed to the UK and the lowest one to France.

Lastly, the corruption index (CI) is added. The corruption index is a widely used governance indicator for countries and therefore will be added for the same reasons as is the CCG. The corruption perception index is defined by Transparency International. The lowest CPI can be attributed to Italy, whereas the highest one can be attributed to Finland.

Normality test

(14)

Table 2: One sample Kolmogorov-Smirnov test

Normal parameters Most extreme differences Kolmogorov-Smirnov

N Mean Std.

Deviation

Absolute Positive Negative Z p-value

G-Score 290 5.43 1.52 0.07 0.05 -0.07 1.26 0.08* ROA 232 -0.07 0.12 0.17 0.17 -0.16 2.63 0.00 ROE 260 0.10 0.64 0.31 0.25 -0.31 4.95 0.00 EPS 221 1.46 3.15 0.20 0.17 -0.20 2.91 0.00 NPM 219 -0.27 4.13 0.47 0.42 -0.47 6.99 0.00 MVE 214 14,355,678 37,980,325 0.35 0.30 -0.35 5.16 0.00 BME 214 0.65 0.92 0.25 0.23 -0.25 3.62 0.00 GDP 290 30.60 4.25 0.20 0.20 -0.13 3.37 0.00 CCG 290 6.43 0.64 0.46 0.33 -0.46 7.90 0.00 CI 290 8.40 0.91 0.39 0.25 -0.39 6.62 0.00

*Normally distributed at 5% significance level.

Table 3: One sample Kolmogorov-Smirnov test

N Mean Std.

Deviation

LnGscore 290 1.65 0.32 0.14 0.07 -0.14 2.34 0.00 Lnroa 199 -2.65 0.91 0.11 0.08 -0.11 1.50 0.02 Lnroe 226 -1.84 0.88 0.10 0.07 -0.10 1.48 0.02 Lneps 192 0.24 1.15 0.09 0.08 -0.09 1.27 0.08* Lnnpm 191 -2.68 1.02 0.08 0.08 -0.06 1.03 0.24* Lnmve 214 14.75 2.03 0.05 0.04 -0.05 0.78 0.58* Lnbme 211 -0.78 0.87 0.06 0.06 -0.04 0.84 0.48* Lngdp 290 3.41 0.14 0.16 0.16 -0.12 2.77 0.00 Lnccg 290 1.85 0.11 0.46 0.33 -0.46 7.88 0.00 Lnci 290 2.12 0.12 0.40 0.26 -0.40 6.73 0.00

*Normally distributed at 5% significance level.

normally distributed. Because variables should fit the normality condition in an ordinary least squares regression analysis the outliers are excluded from the sample. An outlier is defined as a value that lays outside the range of the average value plus or minus two times the standard deviation. The results of this Kolmogorov-Smirnov test are shown in appendix 2 table A. The results show that the ROA is normally distributed when the outliers are excluded. Moreover, the variables are transformed to natural logarithms to make the variables follow a normal distribution. A Kolmogorov-Smirnov test on these variables leads to the results shown in table 3.

(15)

B. These results show that the exclusion of outliers of the natural logarithms sample does not lead to any more normally distributed variables. As a result, the variables G-Score, ROA (without outliers), LnEPS, LnNPM, LnMVE and LnBME are normally distributed and therefore fit the normality condition of a regression analysis. The variables ROE, CCG, CI and GDP do not fit this condition. I recognise that this is a limitation of my research.

3.2 Earnings announcement returns

Companies have to fit the sample selection criteria in order to be of use for my research. First, a governance score has to be available for the company. Second, the companies have to be public companies in order to have stock prices available for the investigated period. Finally, the date of the earnings announcement has to be available. The number of companies that fit these requirements sum up to 219. I divide the total sample into three sub-samples in order to distinguish between well-governed firms, medium-well-governed firms and poorly well-governed firms. As a result, each sub-samples contains of 73 companies. According to MacKinlay (1997) this is sufficient to do an event study. From MacKinlay it is clear that at least 50 events are necessary to minimalise the chance of type 1 or type 2 errors.

The dates of the earnings announcements are retrieved from the archive of the press releases which most companies have available on their website. The stock prices were found on DataStream. The actual returns are calculated by using the definition from Minnema (2002): Returns are defined as ln(St/St−1) where St is the value of the stock in time t. This is done because it is hypothesized that the return (St/St−1) has a lognormal distribution. Therefore, the logarithm of this random variable is normally distributed: ln(St/St−1) N(μ−(σ2/2),σ2). So, the actual returns, also known as raw returns, are calculated as ln(St/St−1). The abnormal returns are defined as the actual return minus the expected return. The computation of the expected return will be explained in further detail in the methodology section. Table 4 shows the descriptive statistics of the earnings announcements.

Table 4: Descriptive statistics earnings announcement returns in estimation window

High governance Medium governance Low governance

Abnormal returns

Raw returns Abnormal

returns

Raw returns Abnormal

(16)

4. Methodology

This section describes the methodology used in this thesis. Again, the section will be divided into two parts. First, the methodology concerning the operating performance will be described. Second, the methodology concerning the earnings announcement returns will be described.

4.1 Operating performance: regression analysis

In the first part, I will test whether corporate governance affects the operating performance of a company. Four measures for the operating performance are used and several control variables are added as already mentioned in the data section. Therefore, the following equations will be estimated:

ROAt = α + β1 CGt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (1) ROEt = α + β1 CGt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (2) EPSt = α + β1 CGt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (3) NPMt = α + β1 CGt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (4)

Equation (1) to equation (4) will be estimated using Ordinary Least Squares (OLS) regressions. The coefficients will be tested with the Student’s t test on a 0.05 significance level.

Furthermore, the analysis will focus on the two most extreme portfolios, because these two groups are the source of the return differences found by Bauer et al. (2004). The sample will be divided into three equal parts based on the corporate governance score. The best-governed firms will be referred to as Democracies, whereas the weakest governed firms will be referred to as Dictatorships. In this regression analysis the sample is restricted to these two most extreme portfolios. The following equations will be estimated:

ROAt = α + β1 Dictt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (5) ROEt = α + β1 Dictt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (6) EPSt = α + β1 Dictt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (7) NPMt = α + β1 Dictt + β2 MVE t + β3 BME t + β4 GDP t + β5 CCG t+ β6 CI t (8)

In the equation (5) to equation (8) Dict is a dummy variable that takes the value 1 if the firm is a Dictatorship and 0 is the firm is a Democracy. If the coefficient β1 is significantly negative (positive), I interpret this as evidence that weaker shareholder rights are associated with lower (higher) operating performance.

4.2 Earnings announcement returns: event study

(17)

hypothesis is assumed. This means that at any moment in time the price of a stock reflects the information that is available to the public at that time. As soon as new relevant information appears the price of this stock will change.

When using an event study, it is important to make a time frame, consisting of an estimation window, event window and post-event window. Following Core et al. (2006) I define the following windows: [-250;-21), [-20,1] and [2,11]. These windows are shown in figure 1.

Figure 1: Timeline

(Estimation window] (Event window] (Post-event window]

-251 -21 0 1 11

T

0

T

1

0 T

2

T

3

τ

1

τ τ

2

Brown and Warner (1980), Brown and Warner (1985) and MacKinlay (1997) describe the event study methodology in detail. In an event study it is investigated whether an unexpected event has a significant influence on the return by evaluating the abnormal return. The abnormal return for company i on event day t is calculated as follows:

ARit = Rit– E(Rit)

Where ARit, Rit and E(Rit) are the abnormal, the actual and the expected return for day t and company i, respectively. Returns are defined as ln(St/St−1), where St is the value of the stock at time t. To assess the expected return several models are available. Three basic models are described in Brown and Warner (1980). These models are: the mean adjusted returns model, the market adjusted returns model and the market and risk adjusted returns model. First, the mean adjusted returns model assumes that the expected return can be seen as the average return in the estimation window. The abnormal return will be calculated as follows: εi,t = Ri,t – Ki,t. Where εi,t is the abnormal return for company i and event day t, Ri,t is the actual return and Ki,t is the average return in the estimation window calculated as

follows:



  1 0 , , 1 T T t t i t i R n K

(18)

Finally, the market and risk adjusted returns model takes both the return on the market and the risk into account. The abnormal return is calculated as follows: εi,t = Ri,t – Ki,t. Where εi,t is the abnormal return for company i and event day t, Ri,t is the actual return and Ki,t is calculated as follows: Rt = Rz,t*(1-Beta)+Beta*Rm,t. Where Rt is the expected return, Rz,t is the return on a commodity with a minimum variance (e.g. gold) and Rm,t is the return on the market. The beta is the covariance between the return on the market and the return on the share divided by the variance of the return on the market.

Brown and Warner (1980) show that there is no clear relationship between the complexity of the model and the quality of the results. They state: We find that a simple methodology based on the market model performs well under a wide variety of conditions. In some situations, even simpler methods which do not explicitly adjust for marketwide factors or for risk perform no worse than the market model. Therefore, the mean adjusted returns model will be used, because it is the most straightforward one.

Rumours about performance can already be in place before the earnings are announced. In this case the stock price will already be adjusted before the official announcement occurs. To include this stock price changes that take place before the official announcement, I will use cumulative abnormal returns (CARs). Following Core et al. (2006) the following event windows will be analysed: 1,1], 3,1], [-5,1], [-10,1] and [-20,1]. The first window will capture pure announcement returns, while the last four should capture potential differences in run-up returns.

(19)

5. Results and interpretation

This section provides an overview of the results and an interpretation of these results. Again this section will be divided into two parts. The first part will concern the relationship between corporate governance and operating performance. The second part will concern the earnings announcement returns.

5.1 Operating performance

To test whether a positive relationship between corporate governance and operating performance exists, a regression analysis is conducted. The results of this regression analysis are shown in table 5. The results in table 5 show that the coefficients of the G-Score are positive with one exception. This means that better corporate governance leads to better operating performance. In three cases this positive relationship is significant at the 5% level and one more when testing at the 10% significance level. The coefficients of determination show that in most cases the goodness of fit of the equation increases if control variables for size and country differences are added. However, only the regressions without control variables are significant. Moreover, it has to be recognised that the country control variables and the dependent variable ROE do not fit the normality condition. When these variables are left out of the analysis, the relationship between corporate governance and operating performance is positive in all cases of which three of the six remaining cases are significant. In sum, these results suggest that a positive relationship exists between corporate governance and operating performance. Bauer et al. (2004) focus on the two extreme portfolios Democracies and Dictatorships. Therefore, I also examine operating performance differences between Democracies and Dictatorships. In this analysis the sample is restricted to these two most extreme portfolios. These results are shown in table 6. Table 6 shows that the coefficients of the dictatorship dummy variable are negative, with one exception. This implies that weaker shareholder rights are associated with weaker operating performance. However, only in four cases this relationship is significant. The coefficients of determination show that, in most cases, adding both size and country control variables results in a better goodness of fit. Note that the country control variables and the dependent variable ROE do not fit the normality condition. When these variables are exempted from the analysis, the relationship between corporate governance and operating performance is positive in all cases of which four of the six remaining cases are significant.

(20)

Table 5: Results regression analysis full sample

ROA ROE EPS NMP

Control Variables

None Size Size+

Country

None Size Size +

Country

Country Coefficient G-score 0.006 0.003 0.002 0.064 0.021 0.150 0.173 0.033 -0.160 0.118 0.033 0.017 T-Statistic 1.836 1.069 0.620 2.429 1.534 1.044 3.204 0.648 -3.060 2.450 0.647 0.307 P-value 0.068** 0.289 0.536 0.016* 0.127 0.298 0.002* 0.518 0.760 0.015* 0.519 0.759 R 0.123 0.382 0.396 0.150 0.299 0.367 0.226 0.452 0.518 0.175 0.376 0.384 R square 0.015 0.146 0.156 0.022 0.089 0.135 0.051 0.204 0.268 0.031 0.141 0.148 Adjusted R square 0.011 0.132 0.129 0.019 0.076 0.109 0.046 0.191 0.244 0.026 0.126 0.116 *Significant at 5% level. **Significant at 10% level.

Table 6: Results regression analysis extreme portfolios

ROA ROE EPS NMP

Country

Country Coefficient Dict. -0.013 -0.006 0.000 -0.176 -0.096 -0.060 -0.703 -0.308 -0.137 -0.542 -0.502 -0.478 T-Statistic -1.143 -0.486 -0.035 -1.495 -1.641 -0.953 -3.531 -1.450 -0.616 -2.928 -2.298 -1.939 P-value 0.255 0.628 0.972 0.137 0.103 0.342 0.001* 0.150 0.539 0.004* 0.023* 0.055** R 0.096 0.434 0.462 0.114 0.261 0.336 0.300 0.424 0.498 0.254 0.344 0.359 R square 0.009 0.188 0.214 0.013 0.068 0.113 0.090 0.180 0.248 0.065 0.118 0.129 Adjusted R square 0.002 0.168 0.174 0.007 0.048 0.074 0.083 0.160 0.211 0.057 0.094 0.079 *Significant at 5% level. **Significant at 10% level.

possible explanation. The contrast of my results to Bauer et al. might be caused by the use of a different governance measure. Although, my sample contains Deminor rated companies like used by Bauer et al. it is only a small part of my sample. A sensitivity analysis will be conducted in order to find out if corporate governance measures affect the results. So, possibly the results of the sensitivity analysis can explain some of the contradictory results.

5.2 Earnings announcement returns

(21)

Table 7: Returns of the full sample over [-1,1] window

Raw returns Abnormal Returns

Governance Score

Average std. dev. t-statistic p-value Average std. dev. t-statistic p-value Number of observations

Low -0.014 0.046 -0.298 0.766 -0.013 0.046 -0.291 0.771 73

Medium 0.001 0.044 0.017 0.987 0.003 0.044 0.057 0.954 73

High 0.000 0.048 0.003 0.997 0.002 0.048 0.039 0.969 73

Table 8: Returns for the restricted sample over various windows

Raw returns Abnormal returns

Window Demo Dict Difference p-value Demo Dict Difference p-value

car[-1,1] 0,000 -0,014 0,014 0,091* 0,002 -0,013 0,015 0,068* car[-3,1] 0,004 -0,012 0,017 0,086* 0,007 -0,012 0,019 0,058* car[-5,1] 0,010 -0,006 0,016 0,100 0,014 -0,005 0,019 0,062* car[-10,1] 0,007 -0,005 0,012 0,221 0,014 -0,004 0,018 0,131 car[-20,1] 0,007 -0,006 0,012 0,282 0,019 -0,003 0,022 0,136 *Significant at 10% level.1

well-governed firms have positive returns for both raw returns and abnormal returns. However, these are not significant as well. For both raw and abnormal returns the returns for the well-governed firms are higher than the returns of poorly governed firms. This difference is illustrated in appendix 3. In table 8 this difference is analysed.

Table 8 shows that the pure announcement returns, event window [-1,1], are higher for Democracies (both raw and abnormal). This difference is significant at the 10% level, but not at the usual 5% level. However, there are two potential concerns with using short event windows. First, firms that have bad earning news most of the times announce their performance later than do firms with good news (Core et al. 2006, p.676). Therefore, shareholders already react when the earnings announcement is late. As a result, the entire negative return may not be included in the short event window of [-1,1]. Second, firms with bad news might preannounce some of this information before the actual earnings announcement. If Dictatorships have bad news on average and preannounce this before the actual earnings announcement, the actual return difference between Democracies and Dictatorships will be underestimated. For these two reasons, several longer event windows are examined as well.

The results show that the Democracies have positive raw returns and abnormal returns for all windows, whereas the Dictatorships have all negative returns. The difference varies from 1.2% to 2.2%. Analysing the raw returns it turns out that two of the five are indeed significant, whereas analysing the abnormal returns three out of the five are significant. It is remarkable that the relatively late run-up returns are not significant, whereas the pure announcement returns and the short run-up returns are. This might be caused by a higher variance in returns at the longer event-windows. A

1_{Core et al. (2006) use 9917 companies, whereas I use 290 companies. Generally, fewer companies will result in}

(22)

higher variance increases the p-value. This higher variance at longer event windows may be caused by speculations about the upcoming earnings announcement, however I have no evidence for this.

Overall, my results show that investors are surprised by the positive relationship between corporate governance and operating performance. Although Core et al. (2006) found positive differences in returns for Democracies, these were not significant at all. Therefore, in contrast to the shareholders of US firms, my research suggests that shareholders of European firms are surprised by the underperformance of poorly governed firms. The explanation of this difference in expectations between shareholders of US firms and shareholders of European firms could be caused by the institutional differences between the United States and Europe. First, due to the low concentration of shareholders in Anglo-Saxon countries management is left with great decision power. If shareholders realise that management has dominant power and that management may take actions in their own best interest, shareholders will take this into account in the share price already. Therefore, they will not be surprised when poorly governed firms announce low earnings, because they have already taken this into account. In Continental European countries ownership concentration is much higher. Consequently, the relationship between management and owners is much stronger and more personal. This may give European shareholders a feeling of influence on management and thereby on performance. However, if this influence is not real and management acts in their own best interest, shareholders will be surprised when earnings are announced. Moreover, due to a lack of ownership structure transparency in Continental Europe, companies are able to transfer cash flow from a well performing company to a related, but badly performing company. If shareholders of poorly governed European countries only realise this when earnings are announced, they will be negatively surprised, which causes the share price to drop. So, the difference of expectations between shareholders of European and US firms may be caused by these institutional differences and wrongly perceived influence of shareholders. Note, that this explanation is in contrast to the efficient market hypothesis. This hypothesis states that share prices always incorporate and reflect all relevant information (Ross et al. 2005, p352). Apparently, the way management acts, is relevant information. Nonetheless, I have no evidence for this explanation. Further research could give more insights in this.

(23)

5.3. Sensitivity analysis

A sensitivity analysis is performed in order to find out if the use of different corporate governance measures affects the results. In this sensitivity analysis the Deminor and Ceres rating are taken together, because these have more similarities with each other than with the VEB rating. Both the operating performance analysis and the earnings announcement returns are subject to a sensitivity analysis.

Appendix 4 shows the results of the operating performance analysis. The results in table A show that the relationship between the Deminor/Ceres rating and the operating performance is positive for the ROA measure of operating performance, but negative for all other measures of operating performance although only three times significant. In my sensitivity analysis this sub-sample consists of 43 Deminor rated companies and 15 Ceres rated companies, so the focus is on the Deminor rating which is used by Bauer et al. (2004), who also found a negative relation. So, although these results are in contrast to my earlier results, it is in agreement with the results of Bauer et al. who use the Deminor rating as well. Bauer et al. state that this negative relation might be due to the fact that accounting numbers are biased measures of firm performance. A negative correlation between earnings and corporate governance possibly implies that badly governed firms report less conservative earnings. However, they did not test for this possible explanation. Another explanation could be that weak shareholder rights may give managers job security, which will reduce managerial myopia. Moreover, if managers have a certain amount of job security, managers may be protected from the lower-tail outcomes of good projects and therefore become less risk averse. So, this negative relationship can be explained, it is, however contradictory to my earlier results.

The relationship between the VEB rating and operating performance is positive with one exception. This positive relationship is five times significant. This sample consists of 232 VEB rated companies. In general these results are comparable to my earlier results.

Furthermore, this analysis is conducted on the two extreme portfolios. These results are shown in table B of appendix 4. The coefficient of the dictatorship dummy is positive for most operating performance measures and three times significant. This means that weaker shareholder rights are associated with better operating performance. So, the results of the Deminor/Ceres rated companies are in contrast to the earlier results again. For the VEB rated companies, the coefficients of the dictatorship dummy are negative, with one exception. This means that weaker shareholder rights are associated with weaker operating performance. This result is comparable to my earlier results.

(24)

namely the rights of the shareholders and the degree that these rights can be executed, whereas Deminor and Ceres also have other sub-indices like absence of takeover defences and board structure and functioning. However, the ratings do not give any details on how these sub-indices are valued exactly, so it is hard to explain the difference in results by these different sub-indices. For instance, in a dual board structure the board is more independent of management and could make a more objective decision about the performance of management. Bad management will be removed, which will lead to better operating performance one might argue. On the other hand, in a dual board structure the board is less informed and therefore the board can make a less informative decision about the performance of management. Bad management might not be removed in this case, which will lead to lower operating performance. So, maybe these sub-indices cause the difference in results. However, due to a lack of detail on indices valuation, it is hard to explain the difference in results by these different sub-indices. Nonetheless, it is important to notice that the Deminor/Ceres sample consists of only 58 companies, therefore the chance of a type 1 or type 2 error is relatively large.

The results of the earnings announcement analysis are shown in appendix 5. Table A provides a distinction between the Deminor and Ceres rating on the upper side and the VEB rating on the downside over the [-1,1] window. The Deminor/Ceres analysis shows that the cumulative raw and abnormal returns of the poorly and well-governed firms are both positive, whereas these are negative for the medium governed firms. However, these are not significant. It is noteworthy to realise that only a small part of my research sample contains a governance score rated by Deminor or Ceres. So, when dividing this sub-sample into different governance levels only a few companies remain per governance level. From MacKinlay (1997) it is clear that at least 50 events are necessary in order to do a decent event study. So, the foundation of conclusions for this sub-sample is weak.

The VEB analysis shows negative cumulative raw and abnormal returns for the poorly and well-governed firms, but positive returns for the medium well-governed firm. Nonetheless, these results are far from significant. For the Deminor/Ceres sub-sample the returns of well-governed firms are lower than the returns of poorly governed firms, whereas the opposite holds for the VEB sub-sample. In appendix 6 the differences between Democracies and Dictatorships are illustrated. In table B of appendix 5 these differences are analysed.

The results of the Deminor/Ceres sub-sample show that in each event window Democracies have lower cumulative raw and abnormal returns than Dictatorships. Although these differences are not significant, this is in contrast to expectations. However, drawing conclusions from portfolios that contain only thirteen events is unreasonable.

(25)

explanation. Nonetheless, from the results of this sub-sample it can be concluded that the market is surprised by the effect of corporate governance on operating performance. Therefore, it is plausible to assume, that the unawareness about the relationship between corporate governance and operating performance, (partly) causes the return difference found by Bauer et al. (2004). Thus, for the earnings announcement analysis, it is hard to state that the results are affected by the use of different governance ratings, because the Deminor/Ceres sub-sample is too small to draw any real conclusions from. Leaving the Deminor/Ceres ratings out, the results are comparable to the earlier results.

(26)

6. Conclusion and Limitations

The results of my study suggest that poorly governed firms have lower operating performance. Management in these firms is probably not taking the actions that are in the best interests of the shareholders and therefore these firms have lower operating performance. However, this does not explain the lower stock returns of these firms as found by Bauer et al. (2004). Stock prices are based on the expectations of shareholders. Therefore, if shareholders expect that poorly governed firms have lower operating performance, this would not cause any difference in stock returns with well-governed firms. However, the results of the earnings announcement analysis suggest that shareholders are surprised by the lower operating performance. These results are contrary to Core et al. (2006) who found that the shareholders of US firms where not surprised by the lower operating performance of poorly governed firms. This contrast could be explained by institutional differences between Europe and the United States, however I have no evidence for this.

Nonetheless, the unexpected underperformance of poorly governed European firms is at least part of the reason why these firms have lower stock returns as found by Bauer et al. Another reason could be that the market realises that poorly governed firms have higher take-over protection and therefore have a reduced chance of receiving a take-over premium. This could be part of the reason of the return difference as well, however I did not examine this in my thesis. Overall, my results suggest that underperformance caused by poor corporate governance does surprise the market.

Nonetheless, this research has some limitations. First of all, this research does not contain the same firms as used by Bauer et al. Core et al. do examine the same firms used by Gompers et al. and therefore could go deeper into the exact return difference. Due to limited free availability of the governance scores of the firms used by Bauer et al., the firms used in my thesis are not equal to the firms of Bauer et al.

Second, the limited free availability of governance scores for European companies causes that different scores are used interchangeably. Although, these score largely contain the same elements, they are not exactly equal. Therefore, it could cause a company to receive a higher score from one rating agency, while receiving a lower one from another rating agency. The sensitivity analysis showed that the use of different governance scores does affect the results. Apparently, the use of different scores causes inconsistencies in my research.

(27)

also of the US. So, according to Rose governance practice in the Netherlands is comparable to other European jurisdictions. Nonetheless, the use of mainly Dutch companies reduces the chance of a random event, which is necessary to generalise the results to Europe.

Fourth, not all variables used in this research are normally distributed. Consequently, for some variables this important requirement for doing a regression analysis is not met. This should be taken into account when evaluating the results including these non-normal variables, because this implies that a perceived linear relationship is not founded.

Fifth, four operating performance measures are used, namely ROA, ROE, EPS and NPM. However, many more exist, like EVA and Economic profit. However, because the former ones are more straightforward to calculate, these are used. However, the latter ones contain more information and less noise and therefore are better operating measures (Garrison et al. 2003, p. 614). So, for further research it might be wise to extend this research with these operating measures.

Sixth, in this research the earnings announcement returns are analysed at the full-year earnings announcement. However, many firms announce quarter-year earnings. At these quarter-year announcements shareholders may already become aware of the underperformance of poorly governed firms and react on this at the quarter-year announcement. Consequently, at the full-year earnings announcements the returns may already be muted. Because this research only takes into account the returns around the full-year earnings announcements, the return difference is most likely not the entire difference between well-governed firms and poorly governed firms.

(28)

7. References

1. Arndt, C. and C. Oman, Uses and Abuses of Governance indicators, OECD Development Centre, Paris, 2006

2. Baghat, S. and B. Bolton, 2008, Corporate governance and firm performance, Journal of Corporate Finance, Vol. 14, pp. 257-273

3. Baliga, B.R., R.C. Moyer and S. Ramesh, 1996, CEO Duality and Firm Performance: What's the Fuss? Strategic Management Journal, Vol. 17, No. 1, pp. 41-53

4. Bauer, R., N. Guenster and R. Otten, 2004, Empirical evidence on corporate governance in Europe: The effect on stock returns, firm value and performance, Journal of Asset Management, Vol. 5, No. 2, pp. 91–104

5. Bebchuk, L., A. Cohen, and A. Ferrell, What Matters in Corporate Governance?, 2004, Discussion Paper

6. Brown, L.D., and M.L. Caylor, 2006, Corporate Governance and Firm Performance, Journal of Accounting & Public Policy, Vol. 25, No. 4, pp. 409-434

7. Brown, S.J. and J.B. Warner, 1980, Measuring security price performance, Journal of Financial Economics, Vol. 8, pp. 205-258

8. Brown, S.J. and J.B. Warner, 1985, Using daily stock returns: The case of Event Studies, Journal of Financial Economics, Vol. 4, pp. 3-31

9. Chiang H., 2005, An Empirical Study of Corporate Governance and Corporate Performance, Journal of American Academy of Business, pp. 95-101

10. Core, J.E., R.G. Wayne and T.O. Rusticus, 2006, Does weak governance cause weak stock returns? An examination of firm operating performance and investors’ expectations, Journal of Finance, Vol. 61, No 2, pp. 655-687

(29)

12. Garrison, R.H., E.W. Noreen and W. Seal, Management Accounting, McGraw-Hill, Berkshire. 2003

13. Gompers, P.A., J.L. Ishii and A. Metrick, 2003, Corporate governance and equity prices, Quarterly Journal of Economics, pp. 107-155

14. Huizingh, E., Inleiding SPSS 11 voor Windows, Academic Service, Schoonhoven, 2002

15. Investopedia 2008

http://www.investopedia.com/terms/c/corporategovernance.asp

16. Jensen, M.C., and W.H. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs and ownership structure, Journal of Financial Economics, Vol. 3, pp. 305-360

17. Lyn, E.O., M. Petrova and A.C. Spieler, Does corporate governance impact the probability and resolution of financial distress?, 2005, Working Paper

18. MacKinlay, A.C., 1997, Event Studies in Economics and Finance, Journal of Economic Literature, Vol. 19, pp. 13-39

19. NewBold, P., W.L. Carlson and B.M. Thorne, Statistics for Business and Economics, Prentice Hall, New Jersey, 2003

20. Ooghe, H. and V. De Vuyst, 2001, The Anglo-Saxon versus the Continental European governance model: Empirical evidence of board composition in Belgium, Vlerick Leuven Gent Management School Working Paper Series 2001-6

21. Rose, N., 2003, Boardroom survival in Europe, Corporate Finance, Vol. 8, pp. 14-21

22. Ross, S.A., R.W. Westerfield and J. Jaffe, Corporate Finance, McGraw-Hill, New York, 2005

(30)

(31)

Appendix 1

Distribution of companies over countries

Table A: Number of companies per country

Country Number of companies

(32)

Appendix 2

Kolmogorov-Smirnov test

Table A: One sample Kolmogorov-Smirnov test without outliers

N Mean Std.

Deviation

G-Score 284 5.482 1.459 0.069 0.048 -0.069 1.166 0.132* ROA 220 0.068 0.069 0.091 0.091 -0.090 1.347 0.053* ROE 257 0.143 0.291 0.200 0.159 -0.200 3.199 0.000 EPS 209 1.361 1.571 0.085 0.085 -0.078 1.223 0.100 NPM 217 0.0809 0.211 0.248 0.248 -0.242 3.650 0.000 MVE 208 8,980,999 15,239,153 0.278 0.263 -0.278 4.009 0.000 BME 207 0.550 0.402 0.107 0.106 -0.107 1.544 0.017 GDP 283 30.705 3.992 0.204 0.204 -0.134 3.426 0.000 CCG 257 6.641 0.216 0.496 0.395 -0.496 7.949 0.000 CI 265 8.626 0.516 0.355 0.231 -0.355 5.785 0.000 *Significant at 5% level.

Table B: One sample Kolmogorov-Smirnov test without outliers

N Mean Std.

Deviation

(33)

Appendix 3

Figures cumulative returns (%) extreme portfolios

Figure A: Cumulative raw returns

-1,5

-1

-0,5

0 0,5

1 1,5

[-1,1]

[-3,1]

[-5,1] [-10,1] [-20,1]

Democracies

Dictatorships

Figure B: Cumulative abnormal returns

(34)

Appendix 4

Sensitivity analysis: Operating Performance

Table A: Results regression analysis full sample sensitivity analysis

ROA ROE EPS NMP

Country

Country Coefficient G-Score 0.015 0.012 0.001 -0.105 -0.026 -0.054 -0.317 -0.303 -0.131 -0.284 -0.291 -0.621 T-Statistic 0.878 0.938 0.081 -1.518 -0.356 -0.564 -2.113 -1.933 -0.599 -1.097 -1.337 -3.478 P-value 0.387 0.360 0.937 0.137 0.725 0.578 0.045* 0.066** 0.556 0.283 0.198 0.003* R 0.158 0.738 0.786 0.228 0.269 0.335 0.389 0.490 0.572 0.210 0.597 0.845 R square 0.025 0.545 0.618 0.052 0.072 0.112 0.152 0.240 0.327 0.044 0.357 0.713 Deminor/ Ceres Adjusted R square -0.007 0.477 0.483 0.029 -0.031 -0.110 0.118 0.137 0.115 0.007 0.250 0.599 Coefficient G-Score 0.005 0.003 0.001 0.090 0.029 0.012 0.161 0.033 -0.013 0.120 0.050 0.056 T-Statistic 1.707 0.893 0.190 2.908 2.219 0.803 2.765 0.611 -0.222 2.377 0.942 0.955 VEB P-value 0.090** 0.373 0.850 0.004* 0.028* 0.423 0.006* 0.542 0.825 0.019* 0.348 0.341 R 0.124 0.314 0.345 0.195 0.349 0.397 0.212 0.424 0.489 0.184 0.375 0.430 R square 0.015 0.098 0.119 0.038 0.122 0.157 0.045 0.180 0.239 0.034 0.141 0.185 Adjusted R square 0.010 0.082 0.086 0.034 0.107 0.128 0.039 0.164 0.209 0.028 0.123 0.150 *Significant at 5% level. **Significant at 10% level.

Table B: Results regression analysis extreme portfolios sensitivity analysis

ROA ROE EPS NMP

Country

(35)

Appendix 5

Sensitivity analysis: Earnings announcements

Table A: Returns of the full sample over [-1,1] window

Raw returns Abnormal Returns

Governance Score

Average std. dev. t-statistic p-value Average std. dev. t-statistic p-value Number of observations Low 0.026 0.144 0.178 0.859 0.031 0.144 0.214 0.831 13 Medium -0.011 0.161 -0.070 0.945 -0.005 0.164 -0.029 0.977 14 Deminor/ Ceres _High _0.002 _0.133 _0.017 _0.986 _0.008 _0.133 _0.056 _0.955 ₁₃ Low -0.034 0.051 -0.662 0.509 -0.032 0.051 -0.632 0.528 60 Medium 0.002 0.044 0.050 0.960 0.003 0.044 0.073 0.942 59 VEB High -0.013 0.041 -0.313 0.754 -0.015 0.041 -0.357 0.722 60

Table B: Returns for the restricted sample over various windows

Raw returns Abnormal returns

Window Demo Dict Difference p-value Demo Dict Difference p-value

Empirical evidence on corporate governance, operating performance and investors’ expectations in Europe