Corporate Governance and Stock Returns:
Evidence from European companies
Faculty of Economics and Business
Bachelor thesis
Ioana Larisa Bagyinka
10194606
2
Abstract
This study investigates whether differences in the corporate governance level among European firms leads to the possibility of gaining abnormal returns by an investor who pursues an active investment strategy. A governance index is constructed based on several attributes known to be associated with corporate governance and the returns of governance-sorted portfolios are evaluated for 2008-2009 and 2013-2014. The main finding of this research is that poor governance is associated with higher market risk. Due to the higher risk exposure, poorly governed firms outperform better-governed firms. In contrast to some earlier papers, this study does not find significant abnormal returns after adjusting for risk. Only investors who can predict in advance which firms will improve their governance can earn abnormal returns.
Keywords: Corporate governance; Asset pricing; Equity returns; Firm performance; Market efficiency
3
1. Introduction
The most recent global financial crisis has served to weaken the health of large global enterprises. The crisis had a negative effect on investor confidence, which largely influences the movement of securities prices. Shocks in the economy, like the financial crisis, place doubt about future good firm performance in investors‟ minds. Since corporate governance plays a key role in long term performance an effective corporate governance system would have helped to mitigate the worst aspects of the crisis. According to Beiner, Drobetz, Schmid and Zimmermann (2006, p. 250) agency problems negatively affect the value of firms
through the expected cash flows accruing to investors and/or the cost of capital. They mention that a conflict of interest between a company‟s management and its stockholders make investors pessimistic about future cash flow. Moreover, Lombardo and Pagano (2000, p. 11) showed that a reduction in shareholder‟s monitoring and auditing costs decreases a company‟s cost of capital.
Dobretz, Schillhofer and Zimmermann (2004, p. 268) mention that a firm‟s valuation depends not only on its profitability or growth prospects, but also on the effectiveness of control mechanisms, which ensure that investors‟ fund are not expropriated or wasted in value-decreasing projects. The relationship between better governance and higher valuation is supported by La Porta, Lopez De Silanes, Shleifer and Vishny (2002, p. 1148), who show that better shareholder protection is empirically associated with a higher valuation of corporate assets. Similarly, Gompers, Ishii and Metrick (2003) found that U.S. firms with stronger shareholder rights had higher firm value, higher profits, higher sales growth and lower capital expenditures. According to market efficiency a firm‟s share price should adjust to reflect conflicts of interest between managers and large shareholders. Therefore, abnormal returns should be equal to zero when investing in companies that are fairly priced.
However, the empirical literature initiated by Gompers et al. (2003, p. 15) has showed that an investment strategy that bought U.S. firms with strong shareholders rights and sold U.S. firms with weak shareholders right would have earned an abnormal return of 8.5 percent per year during the sample period. Further evidence for the positive association between governance and abnormal returns was given by Dobretz et al. (2004, p. 291), who found an abnormal return of 12 percent per year for German companies. Given these results, concerns about the efficient market hypothesis can be raised and investors should consider corporate
4
governance as a factor in their investment decisions since this information is not entirely reflected by the stock price.
The primary purpose of this thesis is to test whether well-governed European firms represent a better investment that poorly governed firms. Empirical evidence for the outperformance of well-governed firms over poorly governed firms has been found in the U.S. market (Gompers et al., 2003) and in the German market (Dobretz et al., 2004). However, little research on this matter has been done on the European market. Although, Dobretz et al. (2004, p. 268) stated that continental European governance systems have already converged in several areas towards the Anglo-Saxon model, firms with international operations are still subject to a national corporate governance code and to different corporate governance practices. To analyze whether this differences among countries in corporate governance are reflected in the stock returns companies will be split into different portfolios based on their governance characteristics. Subsequently, the long-term equity returns of portfolios consisting of well-governed companies will be compared to the one of poorly governed companies‟ portfolios.
In order to separate bad form good governed firms a governance index will be constructed based on 15 attributes recognized to mitigate (exacerbate) agency conflicts, and thus being associated with good (poor) corporate governance. Most past research done on this subject used either survey approach (Dobretz et al., 2004) in constructing the governance index or a methodology based on provisions that reduce shareholders rights (Gompers et al., 2003). A contribution of this paper is the different method used in building the governance index. Cluster analysis, which allocates a set of objects to different groups based on similarities, will organize firms in different portfolios according to their governance ratings. Furthermore, to investigate whether corporate governance is fully integrated into the stock prices the returns of European companies will be analyzed for 2008-2009 and 2013-2014, two different time periods from an economic point of view.
The main findings of this analysis are that the portfolio of well-governed firms significantly underperforms the portfolio of poorly governed firms by nearly 1.18% per month for the period 2008-2009 and by nearly 0.072% for the period 2013-2014. Even though the result for the second period is not significant, it is important to note that corporate governance, which is negatively associated with excess returns, matters when making
5
Doretz et al. (2004), who found evidence for significant abnormal returns among governance-sorted portfolios, but similar to those of Aman and Nyugen (2008) and Carvahal and Nobili (2011), who also found a negative relationship between corporate governance and stock returns and no evidence for an undervaluation of stock prices.
The rest of the paper proceeds as follows. Section 2 summarizes the theory underneath the relationship between governance, firm value and stock returns, as well as the empirical findings in the literature. Section 3 provides a description of the hand-collected variables used in the index construction and a description of the database used throughout the analysis. In section 4, cluster analysis will be applied to form equity portfolios based on governance characteristics and the returns adjusted for risk of governance-sorted portfolios will be presented. Section 5 contains a brief statement of the main findings, while in the last section conclusions are drawn.
2. Literature review
If corporate governance matters for firm performance and this relationship is fully
incorporated by the market, then stock prices should reflect any relevant changes in the firm‟s governance. Since investors will anticipate the corporate governance effect on performance, long-term returns will reflect their expectations.
2.1 Corporate governance and firm performance
A typical view in finance on corporate governance was given by Shleifer and Vishny (1997, p. 737) as they defined it as a mechanism that „deals with the ways suppliers of finance to companies assure themselves of getting a return on their investment‟. Although this definition of corporate governance is hardly correct and diligent, it does nonetheless capture the essence of the common concern in the finance literature with agency costs.
The principal-agent problem arises when managers are motivated to act in their own best interest rather than those of the minority shareholders. Beiner et al. (2006, p. 250) stressed out that agency problems, from a theoretical point of view, can affect the value of firms through the expected cash flows accruing to investors and/or the cost of capital. Investors recognize that with better governance more of the firm‟s profits would come back to them as interest or dividends as opposed to being wasted or expropriated by the company‟s
6
shareholders‟ monitoring and auditing costs and consequently this leads to a decrease in the company‟s cost of capital (Lombardo and Pagano, 2002, p. 11). Therefore, better shareholder protection is associated with higher valuation of corporate assets1. This conclusion is
supported by Klapper and Love (2004, p. 723), who proved that companies with higher-quality corporate governance are valued higher. Moreover, they found a positive correlation between good governance and operating performance.
One of the most well-known empirical study on the relationship between firm value and corporate governance is the one by La Porta et al. (2002, p. 1166), who analyze the
consequences of corporate ownership for corporate valuations in different legal regimes. They find evidence that poor shareholder protection is penalized with lower valuation, especially in countries with less developed standards for investor protection (2002, p. 1167).
The positive relation between firm performance and corporate governance can be found as well in the study of Gompers et al. (2003). By constructing a governance index based on takeover defenses for a sample of U.S. firms, they find that increased agency costs at poor-governance firms directly affect firm performance. Moreover, they examine the empirical relationship of the governance index and two possible sources of agency costs, capital expenditure and acquisition behavior. Their results show that firms with poor governance engage in unexpectedly large amount of inefficient investments and higher expenditures, resulting in a worse operating performance2. Furthermore, Dobretz et al. (2004) find proof for a positive relationship between governance and firm performance by constructing a German corporate governance rating. Contrary to other studies, which look at the regulatory
environment or ownership structure among different countries, they focus on the relationship between a large set of governance proxies and firm performance within a country.
2.2 Corporate governance and stock returns
In an efficient market, the information that good governance leads to higher firm performance should be reflected by equity prices. Therefore, investors would do well by buying stocks of well-governed firms while avoiding stocks of poorly governed firms (Kouwenberg, Salomons and Thontirawong, 2014, p. 965). However, Aman and Nyugen (2008, p. 648) argued that the fact that better-governed firms achieve higher valuations does not necessarily imply that governance quality if fully incorporated in stock prices. Most of
1
See LaPorta et al. (2002, p. 1148)
7
the studies conducted on the return and risk of governance-sorted portfolios test this hypothesis in a similar way, by constructing a governance index.
Gompers et al. (2003) construct an index based on twenty-four governance provisions for a sample of U.S. companies3. They group the firms in multiple portfolios and note two extremes, a portfolio with the weakest shareholders rights (“Dictatorship Portfolio”) and a portfolio with the strongest shareholders rights (“Democracy Portfolio”). An abnormal return of 8.5% per year in the period 1990-1999 is found on a zero-investment strategy that buys the Democracy Portfolios and sells short the Dictatorship Portfolios. Gompers et al. (2003, p. 14) explained that this anomaly can be caused by the fact that poor governance might
accidentally be correlated with other factors that earned an abnormal return in the period 1990-1999 or that investors systematically underestimate the agency costs caused by poor governance. Furthermore, Bebchuk, Cohen and Ferrell (2009, p. 788) hypothesized that only six out of the twenty-four provisions used by Gompers et al. (2003) play a significant role in the correlation with firm value. They found that the abnormal return disappears in the years after 1999 for the U.S. market and add as a possible explanation the increased attention to corporate governance by the media, investors and academics since the accounting scandals in early 2000s4 (2004, p. 813). This argument implies that an economic downturn should raise the importance of corporate governance in the equity valuation, leading to the following hypothesis:
Hypothesis 1: Abnormal governance returns should be higher in “bad” years than in
“good” years
I expect that the awareness of the corporate governance effect on firm performance, and thus returns, should increase among investors after an economic downturn since they will start making more thoughtful investment choices. However, in the case this hypothesis is rejected it means that investors generally do not underestimate the effect of agency costs on firm performance and that abnormal returns in “bad” years are caused by other factors with which corporate governance might be correlated. Moreover, a possible explanation for the rejection of this hypothesis is that investors prefer to invest in safe assets during a period of economic decline and therefore signals of bad governance tend to undervalue the stocks in
3 Index construction: for every firm, they add one point for every provision that restricts shareholders rights
(2002, p. 8)
4
8
“bad” years. In contrast, investors will start assuming higher risk and invest in riskier assets, reducing their returns, during periods of economic growth.
Later studies in non-US markets report conflicting findings. Dobretz et al. (2004, p. 278) find an abnormal return of 12% per year for a small sample of German listed firms. In the construction of their governance index they used a broad, multifactor corporate governance rating based on objective survey responses from 253 German firms in different market segments (2004, p. 271). In contrast, Bauer, Gunster and Otto (2004, p. 98) report
insignificant abnormal returns for a large sample of European firms in the period 1997-2002. A drawback of their analysis is that they use corporate governance ratings for only two years (2000 and 2002) in constructing their index, while returns are measured over a more
extensive time of period. Insignificant abnormal returns for the Japanese market can also be found in the study of Aman and Nguyen (2008, p. 659), who build their index based on fifteen attributes known to mitigate or exacerbate agency conflicts, and thus associated with corporate governance. Even though their results are consistent with the efficient market hypothesis, they did find evidence that a portfolio of Japanese firms with poor governance has significantly higher return and risk than a portfolio of well governed-firms (2008, p. 660).
Carvalhal and Nobili (2011) take a slightly different approach in testing whether corporate governance plays any significant role in abnormal equity returns. In order to measure firm-level governance among Brazilian companies, they construct a governance index based on fifteen questions that can be answered from public information disclosed by listed companies (2011, p. 250). Most of the studies mentioned above sort firms into
portfolios based on the level of governance and apply the Fama and French5 three factor model to find abnormal returns. In contrast, Carvahal and Nobili (2011, p.295) extend the Fama and French model by adding a fourth governance factor. With this approach, they found an abnormal return of 10% per year in the Brazilian market from buying poor governance firms and selling good ones (2011, p. 258), an approach consistent with the market efficiency hypothesis.
Although Bauer et al. (2004) did not find a significant evidence for a relationship between governance and firm valuation in Europe, nor in the UK, it is worth analyzing this
relationship under a different approach. Since they measured corporate governance only in 2000 and 2001, but portfolio returns in the period 1997-2002, their research suffers both from
5 Asset pricing model designed to describe stock:
9
survivorship bias and look-ahead bias. Moreover, given the fact that most studies found proof for an abnormal governance return, the following hypothesis will be tested:
Hypothesis 2: Corporate governance differences lead to abnormal equity returns in the
European market
The hypothesis above wants to test whether significant abnormal returns can be earned by pursuing a long-short investment strategy in governance-sorted portfolios. Under this
hypothesis, investors will buy well-governed portfolios and sell short poorly governed portfolios. This will be tested by sorting European firms into portfolios based on governance related variables and by analyzing whether there is a significant difference in the abnormal rate of return on a governance-sorted portfolio in excess of what would be predicted by the capital asset pricing model. In the case the hypothesis is rejected and abnormal returns are zero this means that there is a negative relationship between firm-level governance and expected stock returns, which is in line with the market efficiency.
3. Data
3.1 Firm level governance data
The main data source is the Director Insight platform, developed by AMA Partners, which publishes executive pay and corporate governance data on the largest European markets. The database contains hand-collected information from various public sources including annual reports, bylaws and other documents regarding corporate governance published by 1,200 European listed companies in 2008-2014.
Since I cannot use companies with incomplete or unavailable information, the sample was restricted to 194 companies from various indices covering different European countries. The following indices were chosen in the analysis: AEX, AMX, AScX (The Netherlands), BEL20 (Belgium), DAX (Germany) and FTSE100 (The United Kingdom). Furthermore, the
information used in the index construction is selected for the time period of 2009 and 2014, two different years from an economic point of view. This decision was motivated by the fact that governance is less likely to change in the short-time, but it might change after the economy was hit by a negative shock since investors become more aware of its importance.
The dataset used to construct the governance index contains fifteen attributes associated with good or poor corporate governance. Five out of the fifteen variables are binary. The
10
binary variables are Board structure, Severance policy, Clawback provision, Shareholding guideline and CEO duality. Appendix A presents a table which contains a detailed description of each variable, its type and how they were constructed. Table 1 provides summary statistics for each separate attribute for both years, 2009 and 2014. These attributes are divided into three broader categories: board diversity, board composition and incentive alignment.
Table 1. Summary statistics of governance attributes
The table reports the average, the standard deviation and the maximum and minimum of the different governance attributes used in the index construction. See Appendix A for more details.
2009 2014
Variables Mean Std. Dev. Min/Max Mean Std. Dev. Min/Max
A. Board
diversity Average Age 56.52 3.45 44/63 58.37 2.91 51/66 %Females 9.1 0.08 0/33.33 18 0.096 0/45.45 %Non-local 44.33 0.26 0/100 41.37 0.25 0/100 B. Board
composition Board structure 0.59 0.49 0/1 0.6 0.49 0/1 Nr. committees 3.66 1.6 0/8 3.64 1.61 0/8 Board size 14.77 7.74 1/42 15.12 7.48 4/41 CEO duality 0.94 0.24 0/1 0.95 0.23 0/1 C. Incentive alignment Severance 0.27 0.45 0/1 0.53 0.5 0/1 Clawback 0.06 0.23 0/1 0.1 0.29 0/1 Shareholding guideline 0.41 0.5 0/1 0.52 0.5 0/1 % Shareholding CEO 0.57 0.031 0/33 0.5 0.03 0/33 % Shareholding TB 1.5 0.086 0/100 1 0.044 0/34 % STI 18 0.19 0/77 19 0.17 0/75 % Shares granted 16 0.21 0/72 18 0.21 0/76 % Options granted 10 0.19 0/94 6.2 0.14 0/77
The first category “ Board diversity ” evaluates whether diversity among board member leads to better corporate governance. The percentage of females in the supervisory board is generally low, but previous studies documented that females on board enhance governance quality in boardroom by providing different perspectives, skills and knowledge, thus leading to better firm performance (Ruigrok, Peck and Tacheva, 2007, p. 547). Other dimensions of board diversity are nationality and the age of board members. Choi, Park and Yoo (2007) indicate that outside directors have a significant and positive effect on firm performance, while Horvath and Spirollari (2012, p. 479) argue that a younger board has a positive impact on firm performance since its members are willing to bear more risk as well as to undertake major structural changes.
Another important category for corporate governance is “ Board composition ”.
11
under a two-tier structure, while board committees help to efficiently advance the business of the board. They mention that a two-tier board structure consists of a management board that manages the firm‟s operations, plus a separate supervisory board that excludes managers and is charged with overseeing the firm‟s activities, including the appointment and monitoring of corporate managers (2014, p. 365). In contrast, CEO duality, which refers to the fact that the CEO is also the Chairman of the board, seems to have a negative impact on firm performance (Yang and Zhao, 2014). Moreover, O‟Connell and Cramer (2010) find that board size
exhibits also a negative association with firm performance.
The last category, “ Incentive alignment ”, suggests that the key to aligning managerial compensation with shareholder interest is to increase the sensitivity of executive
compensation to firm performance. According to Almazan and Suarez (2003, p. 521),
severance pay is important when the board wants find a substantially better CEO and replace the old one since a monetary incentive will make the old CEO more likely to cooperate. Further research provides empirical evidence that the adoption of clawback provisions enhances firm value and can be effective in reducing incentives for earnings manipulation (Datta and Jia, 2013, p. 195). Additionally, Gopalan et al. (2014, p. 2812) report that shorter-duration executive compensation contracts are associated with greater managerial incentive to manipulate short-term performance, while long-term compensation serves better in
incentivizing CEOs to act in the best interest of shareholders. They also report that executive pay duration is longer in less risky firms and firms with better stock performance (2014, p. 2814). Moreover, the empirical results of Lin and Liu (2013) reveal that managerial shareholding is significantly and negatively associated with audit fees in the low and high regions of managerial ownership.
3.2 Stock returns and risk factors data
The data for stock market prices and stock returns are from Datastream for two different periods: 2008-2009 and 2013-2014. The market index used for the construction of the risk factor in the regression was built based on the companies‟ market capitalization downloaded also from Datastream. As a risk free interest rate the German ten year government bond yield was chosen, obtained from the same database.
As mentioned before, corporate governance is not likely to change in the short-term and therefore, in order to increase the statistical power of the analysis, the sample is extended over two consecutive years. Since the data is retrieved on a monthly basis, a total of
12
approximately 4,600 observations are obtained for each group of years. Table 2 presents summary statistics for the excess firm return and excess market return.
Table 2. Summary statistics of excess returns
The table presents the average excess return, its standard deviation and the maximum and minimum value for all stocks in the sample and for the market index.
2008-2009 2013-2014 Excess Returns Mean return (%) Std. Dev. Min/Max (%) Mean return (%) Std. Dev. Min/Max (%) Firm -4.13 0.1388716 -83.31/142.27 -0.35 0.0675543 -57.56/66.40 Market -4.89 0.056399 -16.82/ 5.95 -0.354 0.0223314 -6.001 /3.55
The table above shows that the average excess return of the sample chosen for this analysis closely follows the average excess market return. The mean excess return is higher for the period 2013-2014 than for the period 2008-2009, but still negative. The market return is more concentrated around zero than the sample, which makes the market less risky. This can be also observed by looking at the standard deviation, which is higher for the sample in both years.
4. Method
4.1 Construction of the governance index
Most previous studies that analyze whether differences in corporate governance leads to abnormal returns follow the method of Gompers et al. (2003, pp.8-9), who adds one point for every practice that reduces shareholders rights so that the best governed firms will be in the lowest governance index, or a method based on responses to objective survey questions. However, the first method does not take into account any variables that have an empirically demonstrated effect on corporate governance, while the second method might suffer from subjectivity and a low response ratio. Therefore, both methods are arbitrary.
This study proposes a new method to group firms into governance-sorted portfolios, by using cluster analysis. Cluster analysis, which has not been previously associated with corporate governance, identifies groups of objects that are similar to each other but different from other groups. There are numerous ways in which clusters can be formed, but since most of the variables used in this analysis are quantitative, Ward‟s hierarchical method seems most appropriate. Under this method clusters are merged in such a way as to reduce the variability within a cluster. Appendix B presents a more detailed description of this method.
13
Furthermore, Figure 1 represents a cluster dendrogram which shows three governance-sorted equity portfolios.
Figure 1: Dendrogram with the three governance-sorted portfolios
The vertical axis refers to the distance or dissimilarity measure between clusters and the horizontal axis shows the clusters formed based on governance dissimilarities
To distinguish the portfolios based on the corporate governance level the average variables of each cluster is compared to an overall average. Table 3 shows the average value for each variable per cluster and the total overall average value per variable. To see which portfolio exhibits the highest risk each average variable is compared to its total average.
Under the first category, “ Board diversity ”, CG3 seems to be the portfolio with the best governance characteristics and CG2 with the worst characteristics. Even though CG3 has a lower percentage of females on the board, it seems that the companies‟ boards are somewhat younger than in G1. Furthermore, the percentage of female and non-local members is the lowest in CG2.
The second category, “ Board composition ”, shows that CG3 is the riskiest portfolio since it is the one with the highest board size compared to the total average and with the highest number of board committees, both leading to a less efficient board. Moreover, the CEO duality variable has the lowest average in CG3 which means that the CEO is also the Chairman in a noteworthy number of companies. Moreover, CG1 is the least risky portfolio because the average values of all variables in this category are closest to their total average value.
14
The last category, “ Incentive alignment ”, shows similar results as the second category with CG1 the least risky portfolio and CG3 the riskiest one. The severance and clawback provision exists mostly among the firms in CG1. Moreover, CG1 shows the highest managerial shareholding and that the CEOs in this portfolio have the lowest percentage of short-term compensation and the highest percentage of long-term compensation. In contrast, CG3 displays the opposite values to CG1 making it the most risky portfolio.
Overall, the analysis above shows that CG1 is the safest investment portfolio based on governance characteristics while CG3 is the most risky investment choice.
Table 3. Average variables per cluster
The table reports the average of each governance attribute in each of the governance-sorted portfolio. CG1, CG2 and CG3 represent the governance sorted portfolios. The total average is used as a benchmark to assess the level of risk of each portfolio.
Governance portfolio CG1 CG2 CG3 Total
A. Board diversity Average age 57.84459 57.19136 56.96512 57.43002 %Females 0.15004 0.110789 0.1439032 0.134205 %Non-Local 0.491344 0.308183 0.519687 0.428522
B. Board composition Board structure 0.431138 0.659722 0.8133333 0.590674 Nr. Committees 3.886228 2.8125 4.773333 3.658031 Board size 15.4012 8.055556 27.26667 14.96632 CEO duality 0.964072 0.965278 0.84 0.940415
C. Incentive alignment Severance 0.473054 0.430556 0.1733333 0.398964 Clawback 0.113773 0.020833 0.0933333 0.07513 Shareholding guideline 0.556886 0.465278 0.2666667 0.466321 % Shareholding CEO 0.001696 0.011807 0.0000642 0.005151 % Shareholding TB 0.012262 0.01884 0.0002444 0.012381 % STI 0.167252 0.1906 0.2391567 0.189933 % shares granted 0.191584 0.175061 0.1195536 0.171424 % options granted 0.099578 0.069328 0.0572385 0.080067
4.2 Risk adjusted returns on governance sorted portfolios
Stock returns are computed by dividing the stock price in period t+1 to the stock price in period t, sourced from Datastream, and subtracting the value 1. In addition, an own market index was constructed by multiplying each companies‟ return to its market capitalization weight. To measure the risk-adjusted abnormal returns we apply the capital asset pricing model (CAPM), developed by Sharpe (1963) and Lintner (1965). Specifically, to account for the differences in risk between portfolios the following model will be applied to estimate a time series regression:
15
Before computing the abnormal returns for each portfolio, excess returns have to be calculated by subtracting the monthly 10 year German government bond yield from the monthly equity returns for each company. Furthermore, to measure the firm‟s exposure to economy-wide conditions the market‟s excess return over the risk free rate will be calculated. The market excess return is calculated in the same way as the companies‟ stock excess returns with the difference that the efficient market portfolio represents a market
capitalization weighted average of all firms‟ stock returns. After obtaining the abnormal returns for each portfolio in both periods, 2008-2009 and 2013-2014, the significance of a long-short position in the governance-sorted portfolios will be tested.
4. Results
5.1 Univariate analysis
Following the study of Gompers et al. (2003) and Dobetz et al. (2004) realized returns should differ systematically across portfolios if governance matters and it is not incorporated immediately into stock prices. However, the results in this thesis show that the poorly
governed portfolios have higher returns than the good governed ones. This result is in line with the efficient market hypothesis, which shows that poor corporate governance increases the expected return on equity to account for the higher risk due to agency costs. Based on the results in Table 4, the period of 2008-2009 shows that the portfolio with the medium
investment risk, CG2, exhibits a higher average return of -0.44% than the portfolio with better governance (CG1), which has an average return of -0.51%. The negative returns might be explained by the fact that the market was in financial distress due the global financial crisis. Furthermore, the same result is found for the period 2013-2014, CG2 has an average return of 1.1% relative to the 1% average return for the CG2 portfolio.
However, what is surprising is that the portfolio with the worse governance (G3) has an average return of -0.58% for the first period and an average return of 1.05% for the second period. These values show that the low governance portfolio does not outperform the better-governed one for the first period, but it does outperform the best-better-governed portfolio (G1) by a 0.045% difference for the second period. However, this difference is insignificant as shown by Table 4, which displays insignificant differences between the mean returns of governance-sorted portfolios.
16
Table 4. Portfolio performance by governance rating
The table shows the mean return for each governance-sorted portfolio, its standard deviation and their minimum and maximum value. The number of observations per portfolio is also displayed. The second part of the table shows the returns earned by an investor who buys a better-governed portfolio and sells short a bad governed equity portfolio in 2008-2009 and 2013-2014. T-statistics are presented in parentheses. CG1, CG2 and CG3 represent the governance sorted portfolios. CG1 is the least risky investment choice.CG2 is more risky than CG1, but less risky than CG3. CG3 represent the riskiest investment choice.
*** Significance at the 1% level. ** Significance at the 5% level. * Significance at the 10% level.
Mean return (%) St. dev Min (%) Max (%) Nr. obs.
A. 2008-2009 CG1 -0.51 0.126 -70 143.7 1944 CG2 -0.44 0.14 -56.06 145.7 1560 CG3 -0.58 0.157 -79.3 124.5 888 B. 2013-2014 CG1 1 0.062 -54.02 67.2 1919 CG2 1.1 0.077 -55.94 39.5 1728 CG3 1.05 0.059 -36.19 37 889 Time period/ Long-short strategy 2008-2009(%) 2013-2014 (%) CG3-CG1 -0.07868 0.04513 (-0.14) (0.18) CG3-CG2 -0.14575 -0.09273 (-0.24) (-0.31) CG2-CG1 0.06706 0.13787 -0.15 (0.6)
Although the findings above are conflicting with Gompers et al. (2003) and Doretz et al. (2004), they are in line with the results of Aman and Nguyen (2008) and Carvahal and Nobili (2011) who provide evidence that returns are negatively related to firm-level corporate governance.
5.2 Multivariate analysis
An important factor to describe is the market beta. Table 5 reveals that the market beta increases with a decrease in firm-level governance. This is not surprising since a lower governance level implies a higher exposure to market conditions and thus, a higher
systematic risk. However, for the second period, 2013-2014, CG2 exhibits a lower beta than CG3, which is surprising. This might be explained by the fact that other factors, with which corporate governance might be highly correlated, make CG2 riskier than CG3. In addition, the market betas in 2013-2014 are lower than in 2008-2009. This result might be due to the higher regulations imposed after the financial crisis that leads to a less risky sample and thus, to a less volatile one.
17
Furthermore, Table 5 shows whether poorly governed firms have significantly higher returns than well-governed firms after adjusting for the firm‟s exposure to economy-wide conditions. We can interpret that by looking at the intercept (α), which shows the risk-adjusted measure of the excess return on an investment in one of the three portfolios presented in Table 5. The results clearly show that firm-level corporate governance is negatively related to the expected stock returns. For the period 2008-2009 the poorly-governed portfolio (CG3) has a significant higher alpha of 2.96% than the well-poorly-governed portfolio (CG1), which has a significant alpha of 1.78%. All these results are significant at a 1% level6, including the alpha value of 2.4% for CG2. Moreover, the value of 2.4% for the alpha of portfolio CG2 shows that CG2 is outperformed by CG1, the poorly governed portfolio, but underperformed by CG1, the well-governed portfolio. This strengthens the conclusion that corporate governance and stock returns are negatively related. Therefore, the second hypothesis of this study is rejected and no abnormal returns can be gained by buying well-governed portfolios and by selling short poorly governed ones.
Table 5. Risk-adjusted returns of value-weighted portfolios
Dependent variable: Governance-sorted stock returns
The table shows the excess returns of each governance-sorted portfolio after adjusting for risk for two time periods. Rmkt-Rf is the excess return on the market index. α is the excess return earned after adjusting for risk. T-statistics are stated in parentheses. CG1, CG2 and CG3 represent the governance sorted portfolios. CG1 is the least risky investment choice.CG2 is more risky than CG1, but less risky than CG3. CG3 represent the riskiest investment choice.
*** Significance at the 1% level. ** Significance at the 5% level.
* Significance at the 10% level.
Time period 2008-2009 2013-2014 Governance level CG1 CG2 CG3 CG1 CG2 CG3 Rmkt-Rf 1.208896*** 1.323131*** 1.466555*** 1.015599*** 0.8323052*** 1.080323*** (27.96) (24.56) (18.33) (17.23) (10.30) (13.52) α 0.0177687*** 0.0240255*** 0.0295815*** -0.0005251 0.0002156 0.0001942 (5.51) (5.98) (4.95) (-0.39) (0.12) (0.11) R-squared 0.2871 0.2791 0.2749 0.1341 0.0579 0.1709
In the period 2013-2014 the poorly governed portfolio with has an alpha of 0.019% still outperforms the well governed portfolio, which has a negative alpha of -0.053%. However, the portfolio with the medium-governance level exhibits the highest alpha with a value of 0.022%. These results are not completely in line with the negative relationship mentioned above. A possible explanation is that companies change governance between the two periods
6
18
chosen for this study and therefore, there might be other factors which explain the excess returns, more important than the governance level, and that have not taken into account during the analysis. Furthermore, all the results for the second period are statistically insignificant.
Furthermore, R-squared is below 0.3 for all portfolios in both time periods. The highest values are among the poorly-governed portfolios for the second time period as those have a higher market beta7, but there are no notable differences in the R-squared value for the first time period. However, to better explain the variation in the returns governance-based
portfolios should be formed by country and matched with the specific market index. Another explanation for the low R-squared could be due to the model choice. Further research could analyze whether the Fama and French three factor model could lead to different outcomes and to a higher R-squared since a model with multiple risk factors would lead to higher explained variation in the stock returns.
Table 6 shows the return that could have been achieved by an active investment strategy. The intercept is the return on an investment strategy that buys the poorly-governed portfolio and sells short the well-governed portfolio. In the period 2008-2009 an investor could earn a significant monthly return of 1.18%, which translates in an annual return of approximately 14.16%. Moreover, an investor could obtain, though insignificant, a monthly return of 0.072%, which would be 0.86% per year for the period 2013-2014.
Table 6. Long-short strategy for different governance portfolios
This table presents the significance the excess return earned by an investor who buys bad-governed portfolios and sells short governed portfolios. CG1 is the least risky investment choice and considered the well-governed portfolio.CG2 is more risky than CG1, but less risky than CG3. CG3 represent the riskiest
investment choice and considered the poorly governed portfolio. αi- αj is the difference in the excess returns of
governance sorter portfolios and βi- βj is the difference between their market beta. T-statistics are stated in
parentheses.
*** Significance at the 1% level. ** Significance at the 5% level. * Significance at the 10% level.
Long-short strategy 2008-2009 2013-2014 αi- αj βi- βj αi- αj βi- βj CG3-CG1 0.0118128* 0.2576595*** 0.0007194 0.0647239 (1.89) (3.08) (0.31) (0.63) CG3-CG2 0.005556 0.1434244 -0.0000214 0.2480175** (0.80) (1.53) (-0.01) (1.96) CG2-CG1 0.0062568 0.114235* 0.0007408 -0.1832936* (1.23) (1.67) (0.33) (-1.86)
19
In line with these results, Figure 2 shows that the poorly-governed portfolio is more skewed to the right than the good-governed one, thus more significant for period 1 than for period 2. By comparing these two long-short strategies between the two periods chosen in this analysis, we can conclude that corporate governance is more important for periods of economic downturn. Moreover, no abnormal returns can be gained in 2008-2009 since the effect of corporate governance is reflected by the stock returns. However, for 2013-2014 the results are mixed, though insignificant. Therefore, the first hypothesis of this study is
rejected. This can be explained by the fact that signals of bad governance tend to undervalue the stocks in “bad” years since investors are more risk averse and prefer to invest in safer assets. They will start assuming higher risk and invest in riskier assets, reducing their returns during periods of economic growth.
Figure 2: Outperformance by poorly-governed portfolios
G3 represents the poorly governed portfolios, while G1 represents the well-governed portfolio.
Overall, evidence is found that the corporate governance level is negatively associated with excess returns, which is in line with the market efficiency and with the findings of Aman and Nguyen (2008) and Carvahal and Nobili (2011). Additionally, an investor should take into account the level of governance of a firm when making an investment decision. There is evidence that a significant return of 14.16% per year for the period 2008-2009 can be gained on a long-short investment strategy and, though insignificant, a yearly positive return of 0.86% for the period 2013-2014.
20
6. Conclusion
While the relationship between good corporate governance and higher firm valuation has been widely recognized there is still doubt whether stock returns fully reflect differences in governance. According to the efficient market hypothesis, excess return can only result from identified risk factors. If governance-related information is already incorporated into a firm‟s valuation there should be limited space for any additional returns. However, past research found significant evidence that there is a slow price adjustment to corporate announcements and this issue can be exploited by pursuing an active investment strategy, rather than a passive one.
Given the fact that for the period 2013-2014 insignificant results are obtained, a limitation of this paper is the number of companies used in analyzing returns. To get a better
representation of the European market the number of countries from which the sample was chosen should be increased. This would lead to greater statistical significance of the results and would play an essential role in examining the robustness of the conclusions. Furthermore, this research suffers from omitted variables in the index construction. Since corporate
governance is hard to measure, there is a possibility that other governance-related variables were left out. An idea for further research would be to include more variables in the index construction that were empirically demonstrated to have an effect on corporate governance. Furthermore, the variables might suffer from measurement errors since all of them were hand-collected. As a solution for this limitation annual reports and other public available information could be consulted. Finally, the capital asset pricing model has its own limitations. The model is based on a number of unrealistic assumptions like the fact that a highly liquid government security is considered as a risk free security or that the lending or borrowing rates are equal. Moreover, it is difficult to test the validity of the model and since the betas do not remain stable over time implies that historical betas are poor indicators of the future securities‟ risk.
Notwithstanding these limitations, it is important to note that the results of this paper do not imply that governance ratings are useless and that they contribute to the information available to market participants. Consistent with the market efficiency, investing in securities based on publicly available information cannot be expected to yield excess returns. Only unexpected changes in the firm governance level, which are not reflected in asset prices, could lead to abnormal profits.
21
References
Almazan, A. and Suarez, J. (2003). Entrenchment and severance pay in optimal governance structures. The Journal of Finance, 58(2), 519-547.
Aman, H. and Nguyen, P. (2008). Do stock prices reflect the corporate governance quality of Japanese firms?. Journal of The Japanese and International Economies, 22, 647-662. Bauer, R., Gunster, N. and Otten, R. (2004). Empirical evidence on corporate governance in
Europe. Journal of Assent Management, 5(2), 91-104.
Bebchuk, L., Cohen, A. and Ferrell, A. (2009). What matters in corporate governance?. The
Review of Financial Studies, 22(2), 783-827.
Beiner, S., Drobretz, W., Schmid, M.,M. and Zimmermann, H. (2006). An integral framework of corporate governance and firm valuation. European Financial Management, 12(2), 249-283.
Belot, F., Ginglinger, E., Slovin, M., B. and Sushka, M., E. (2014). Freedom of choice between unitary and two-tier boards: An empirical analysis. Journal of Financial
Economics, 112, 364-385.
Carvalhal, A. and Nobili, C. (2011). Does corporate governance matter for stock returns?. Estimating a four-factor asset pricing model including a governance index.
Quantitative Finance, 11(2), 247-259.
Choi, J., J., Park, S., W. and Yoo, S., S. (2007). The value of outside directors: Evidence from corporate governance reform in Korea. Journal of Financial and Quantitative
Analysis, 42(4), 941-962.
Dobretz, W., Schillhofer, A. and Zimmermann, H. (2004). Corporate governance and expected stock returns: Evidence from Germany. European Financial Management,
10(2), 267-293.
Gompers, P., A., Ishii, J., L. and Metrick, A. (2003). Corporate governance and equity prices.
Quarterly Journal of Economics, 118(1), 107-155.
Gopalan, R., Milbourn, T., Song, F. and Thakor, A., V. (2014). Duration of executive compensation. The Journal of Finance, 69(6), 2777-2817.
Horvath, R. and Spirollari, P. (2012). Do the board of directors‟ characteristics influence firms‟ performance? The U.S. evidence. Prague Economic Papers, 4, 470-486. Iskandar-Datta, M. and Jia, Y. (2013). Valuation consequences of clawback provisions. The
22
Klapper, L., F. and Love, I. (2004). Corporate governance, investor protection, and performance in emerging markets. Journal of Corporate Finance, 10, 703-728. Kouwenberg, R., Salomons, R. and Thontirawong, P. (2014). Corporate governance and
stock returns in Asia. Quantitative Finance, 14(6), 965-976.
La Porta, R., Lopez-De-Silanes, F., Shleifer, A. and Vishny, R., W. (2002). Investor protection and corporate valuation. The Journal of Finance, 57(3), 1147-1170. Lin, Z., J. and Liu, M. (2013). The effect of managerial shareholding on audit fees: Evidence
from Hong Kong. International Journal of Auditing, 17(3), 227-245. Lombardo, D. and Pagano, M. (2000). Legal determinants of the return on equity.
Unpublished working paper, Centre for studies in Economics and Finance, University of Naples, Italy.
O‟Connell, V. and Cramer, N. (2010). The relationship between board characteristics and firm performance in Ireland. European Management Journal, 28, 387-399. Ruigrok, W., Peck, S. and Tacheva, S. (2007). Nationality and gender diversity on Swiss
corporate boards. Corporate Governance: An International Review, 15(4), 546-557. Shleifer, A and Vishny, R., W. (1997). A survey of corporate governance. The Journal of
Finance, 52(2), 737-783.
Yang, T. and Zhao, S. (2014). CEO duality and firm performance: Evidence from an exogenous shock to the competitive environment. Journal of Banking and Finance,
23
Appendix A: Index variables definition
Variable Definition Type Construction
1. Board diversity
Average age The average age of the supervisory board members
quantitative
Total age/Number of members %Female
The percentage of female members in the supervisory board
quantitative
Number of females/Total members %Non-local
The percentage of foreign members in the supervisory board
quantitative
Number of non-locals/Total members
2. Board composition
Board structure Unitary or two-tier structure (separation between the management and supervisory board)
binary
1= two-tier structure 0= unitary structure Nr. committees The number of board
committees
quantitative
The actual number of committees Board size How many members are in
the supervisory board
quantitative
The actual number of members in the supervisory board
CEO duality Whether the Chairman of the supervisory board holds also the role of the CEO
binary
1= the CEO is not the chairman 0= the CEO is also the Chairman
3. Incentive alignment
Severance The existence of severance (termination benefit) payment
binary 1= severance is present 0= severance is missing
Clawback The existence of clawback provision (the option to limit bonuses for executives)
binary 1= clawback is present 0= clawback is missing
Shareholding guideline Requirement for the board to hold company shares
binary 1= shareholding guideline is present 0=shareholding guideline is missing % Shareholding CEO The percentage of firm
shares owned by the CEO
quantitative Number of share owned by CEO/Total number of shares % Shareholding TB The percentage of company
shares owned by the supervisory board
quantitative Number of share owned by the supervisory board/Total number of shares
% STI The percentage of the CEO's short-term compensation out of total compensation
quantitative Short-term compensation/Total compensation
% Shares granted The percentage of
performance share granted to the CEO
quantitative Number of share granted*share price/ Total compensation
% Options granted The percentage of options granted to the CEO
quantitative Number of options granted*fair value/ Total compensation
24
Appendix B: Ward‟s minimum variance clustering method
This method is distinct from other methods because it uses an analysis of variance approach to evaluate the distances between clusters. Cluster membership is assessed by calculating the total sum of squared deviations from the mean of a cluster.
∑ ∑ ∑ ̅̅̅̅
Where represents the value for variable k in observation j, belonging to cluster i and ESS represents the Error Sum of Squares.
The equation above sums over all variables and all of the units within each cluster. The individual observations for each variable are compared against the cluster means for that variable. When the Error Sum of Squares is small this suggests that the data is close to their cluster means, implying a cluster of like units.
