Corporate governance and EU company
performance during the financial crisis
Geu Luik (s1884018)
ABSTRACT
This research uses firm-level data from five European countries (from the UK, Germany, France, Spain and the Netherlands) to examine whether corporate governance has played a role in company performance during the financial crisis of 2008. This research is primarily based on Bae, Baek, Kang and Liu (2012). Empirical evidence from East Asia shows that companies with a high quality of corporate governance performed better during the East Asia crisis than companies with a poor quality of corporate governance. This research investigates whether this relationship also holds for Europe during the financial crisis of 2008. Conclusion is that this is partially the case. Companies with a good quality of corporate governance performed significantly better than companies with a poor quality of corporate governance during the financial crisis of 2008, based on several expropriation variables. On the other hand, companies with a poor quality of corporate governance recovered significantly faster after the crisis.
Double Degree MSc International Financial Management
University of Groningen & University of Uppsala
1. INTRODUCTION
While more and more advanced economies in Europe are currently recovering from the financial and economic crisis it faced during the last four to five years, an increasing number of academics, politicians and citizens are searching for an answer and the cause of one of the largest crises in modern history. It is generally accepted by economists that the international financial crisis has started in 2008, when several renowned financial institutions collapsed. From then on, the crisis spread from the US towards Europe. Financial institutions and non-financial companies came in troubled water. While later on, whole economies, such as Greece, Italy and Ireland faced serious liquidity problems.
previous research regarding financial crises can be investigated. For example, Mitton (2002) and Bae et al. (2012) both researched the financial crisis in East Asia from 1997 until 1998. Where Mitton (2002) has taken several emerging East Asian countries into account regarding his research, Bae et al. (2012) have mainly focused on South Korea. Both papers show that corporate governance plays a role during the East Asia crisis. Subsequently, several academics argue the importance of corporate governance, not only in emerging economies such as the East Asian countries. For example, La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998, 1999b, 2000) conclude that corporate governance, across countries, is a key element in establishing firm value and financial market development. As a definition of corporate governance, Mitton (2002) uses the following: “it is the means by which minority shareholders are protected from expropriation by managers or controlling shareholders”. The oft-cited paper Gompers, Ishii and Metrick (2003) show that ‘good’ corporate governance has a positive impact on firm performance.
According to Mitton (2002), although some economists suggest that corporate governance can have set the East Asia crisis in motion, corporate governance practices in East Asia could have made countries and companies more vulnerable to the financial crisis, and could even aggravated the crisis as soon as it started. Corporate governance can become more important and critical within a financial crisis, for at least two major reasons (Mitton, 2002). First, expropriation of minority shareholders can become more overwhelming during a crisis. Johnson, Boone, Breach and Freeman (2000) argue that a crisis can lead to greater expropriation, since managers are persuaded to expropriate more as the expected return on investment declines. And second, a crisis can force investors to recognize and benefit from deficiencies in corporate governance that existed all along (Johnson et al., 2000).
This is in accordance with Bae et al. (2012), in which the focus is on one of the possible explanations why corporate governance plays a role during the financial crisis in East Asia. One potential explanation is that during a financial crisis period, incentives of controlling shareholders to expropriate their counter minority shareholders have the inclination to go up, since expected return on investments decrease (Johnson et al., 2000; Mitton, 2002; Baek, Kang and Park, 2004). This implicates that incentives of controlling shareholders to expropriate minority shareholders is the fundamental direction through which corporate governance affects firm value during a crisis period. Bae et al. (2012) refers to this theory as the expropriation hypothesis. Alternative explanations of corporate governance will be discussed later in the theoretical section of this paper.
governance in relationship with the financial crisis of 2008. However, Erkens et al. (2012) only focused on corporate governance practices within financial institutions and the role it played during the financial crisis. Therefore, it would be hazardous to make, in general, direct conclusions regarding the financial crisis of 2008 for companies with different backgrounds (in terms of country coming from, industry, etc.). Moreover, the financial crisis of 2008 has had a significantly larger impact than the East Asian crisis of 1997-1998, in terms of affecting company and country performance and having an impact on the number of companies and countries.
“If corporate governance has been a significant factor in the crisis, then corporate governance should explain not only cross-country differences in performance during the (East Asian) crisis, but also cross-firm differences in performance within countries” (Mitton, 2002). Therefore, this research uses firm-level data to explain whether corporate governance has indeed had an impact during the financial crisis on the performance of European countries (from United Kingdom, Germany, France, Spain and the Netherlands). This differentiates this research from Bae et al. (2012) who mainly focused on South Korean companies. The research framework is loosely based on Mitton (2002) and Bae et al. (2012), with some nuance differences regarding data and variables that are being used. Moreover, where Bae et al. (2012) compares post-crisis period with the crisis period, this research broadens the comparison with adding a pre-crisis period (as also done by Mitton, 2002).
This paper contributes to an emerging body of research that attempts to identify the mechanisms that influenced how severely firms were impacted by the financial crisis of 2008, and to what extent corporate governance has played a role in it. It contributes in primarily three ways. First, not only cover financial companies are covered in this research, but also non-financial companies, contrary to for example Erkens et al. (2012). To be more accurate, this research will show whether there is a difference in company performance between financial and non-financial firms. Second, three reference periods(pre-crisis period, crisis period, post-crisis period) are compared with each other, to capture the trend in corporate governance and other relevant variables. Finally, the focus will be on Europe, while other research has mainly focused on East Asian companies during the East Asian crisis.
research. Finally, a conclusion can be made. This includes implications of this research, its limitations and future research directions.
2. THEORETICAL FRAMEWORK
2.1: Definition of a financial crisis
Before defining and discussing corporate governance, it is evidently important to define what a financial crisis is. This research follows the definition of Mishkin (1992). According to Mishkin (1992), a financial crisis is:
“A disruption to financial markets in which adverse selection and moral hazard problems become much worse, so that financial markets are unable to efficiently channel funds to those who have the most productive investment opportunities.”
With the definition of Mishkin (1992) of a financial crisis in mind, it is clear that the crisis of 2008 is a financial crisis. As starting point of the financial crisis – although this is arbitrary – this research uses, just as several other renowned economists did in their research, January the 1st of 2008. From then on, stock markets declined significantly (see for example figure 1). The financial crisis has caused a change in the market value of companies. This can be due to a collapse on the stock market. Several stocks, and thus stock markets, dropped as soon as the financial crisis moved towards the European continent. Contagion effects could affect other stocks, and therefore also decrease. Hence, market value of a significant number of companies dropped.
Subsequently, besides problems on the stock market, the financial crisis also caused problems among citizens. Customer confidence and consumption decreased. Finally, this causes lower sales, and thus a decrease in profits for a company. Shareholders know this process, and expect that in the end, the cash flows (i.e. dividends) to shareholders decrease. In return, the stock price of the specific company must decline. As a result, market value of a company will decrease.
2.2: Definition of corporate governance
Corporate governance is a concept that has been widely used and discussed in modern academics. Subsequently, it has been used for explaining several movements in the economy. Moreover, corporate governance can be measured in many ways. However, before different ways of measuring corporate governance are being discussed, it is important to define the concept of corporate governance.
Corporate governance is referring to the system how a company is being directed and controlled. It is the “system of rules, practices and processes by which a company is directed and controlled. Corporate governance essentially involves balancing the interests of the many stakeholders in a company - these include its shareholders, management, customers, suppliers, financiers, government and the community. Since corporate governance also provides the framework for attaining a company's objectives, it encompasses practically every sphere of management, from action plans and internal controls to performance measurement and corporate disclosure.1” When focusing on the group of shareholders, this definition is in accordance with for example Mitton (2002) and Gompers et al. (2003). Mitton (2002) defines corporate governance as the “means by which minority shareholders are protected from expropriation by managers or controlling shareholders”. Whereas, Gompers et al. (2003) loosely defines corporate governance as the “degree of ultimate authority among voters” (to what extent shareholder rights are not restricted). According to La Porta et al. (2002) “corporate governance is, to a large extent, a set of mechanisms through which outside investors protect themselves against expropriation by the insiders.” Additionally, Johnson et al. (2000) define corporate governance “the effectiveness of mechanisms that minimize agency conflicts involving managers, with particular emphasis on the legal mechanisms that prevent the expropriation of minority shareholders”, which is also discussed by Shleifer and Vishny (1997).
Corporate governance is a concept that is becoming increasingly important within companies. Especially after the introduction of Sarbanes-Oxley Act (SOX) and several local acts such as code Tabaksblat (the Dutch corporate governance code) in the Netherlands, companies have changed or have been mandated by (inter)national governments to change their corporate governance structure. Additionally, the increase of hostile takeovers, for example by investment and hedge funds, have restricted shareholder rights during the last two or three decades (Gompers et al., 2003).
2.3: Measuring corporate governance
As mentioned earlier, there are different ways of measuring corporate governance. First, this can be done on a country-level basis, and second, on a firm-level basis (Mitton, 2002). La Porta et al (1997, 1998, 1999b, 2000) has demonstrated that across countries, corporate governance is indeed an important factor in financial market development and firm value. They have examined legal rules covering protection of corporate shareholders and creditors among dozens of countries. Their results show that common-law countries have the highest legal protection (La Porta et al., 1998). According to La Porta et al. (2000) investor protection is crucial, since in many countries expropriation of minority shareholders and creditors by the controlling shareholders is extensive. When outside stakeholders are financing the company, the chance of redeeming their returns on investments reduces dramatically, due to the fact that controlling shareholders or managers are expropriating. In general, more legal protection (based on a country-level basis) improves corporate governance.
Expropriation can occur in many ways as discussed by La Porta et al. (2000). The most basic way is that insiders (controlling shareholders and managers) simply steel the profits from inside the organization. Other possibilities are transfer pricing, asset stripping and investor dilution, which have, although often legal, the same result as theft (La Porta et al., 2000). In those cases, insiders, respectively, sell the output, the assets or additional securities from the company they control to companies outside the organization of which they also own, for prices far below the current market price. Other ways of expropriation are installing, potentially, unqualified family members in controlling and managing positions within the organization, diversion of corporate opportunities from the firm and overpaying executives. Without prejudice to the approaches of expropriation, expropriation is directly linked with the agency problem as discussed by Jensen and Meckling (1976), meaning that insiders use the profits of the company to benefit themselves (and maybe their relatives) instead of returning the benefits to the outside investors (for example shareholders and debtholders) (La Porta et al., 2000). The research of La Porta et al. (1997, 1998, 1999b, 2000) is definitely linked with Johnson et al. (2000), who present evidence that weaknesses of legal institutions has had an important impact on (exchange rate) depreciations and stock market declines during the East Asian crisis. If expropriation by managers increases when the expected rate of return decreases (such as during a crisis), an adverse shock to investor confidence will lead to increased expropriation as well as lower capital inflow and greater capital outflow for a country. In the end, this translates into lower stock prices and depreciated exchange rates. Johnson et al. (2000) found convincing support for their hypothesis that corporate governance explains to a great extent exchange rate depreciations and stock price decreases during the East Asian crisis.
diversification, although it is not by its very nature a corporate governance mechanism, according to Mitton (2002). Lower transparency of diversified companies allows managers and/or controlling shareholders to take advantage of minority shareholders, due to higher levels of asymmetric information (Lins and Servaes, 2000; Lins, 2000). Thus, during a financial crisis, expropriation may increase, but will mainly harm diversified companies (Mitton, 2002). Therefore, diversified companies are associated with significantly worse stock price performance.
Bae et al. (2012) measures corporate diversification in a somewhat different way, but its research is nevertheless loosely based on the foundations of Mitton (2002). Additionally, Bae et al. (2012) describes and measures more possible explanations than only the expropriation hypothesis, as discussed earlier in this paper. One of the alternative explanations is the information-based argument. Based on this argument, “poor performance of firms with weak corporate governance during the East Asian crisis is not necessary due to increased expropriation, but due to investors’ paying more attention to corporate governance problems that have been hidden” (Bae et al., 2012). Another alternative explanation is that “governance measures (used in previous studies) are closely correlated with firms’ sensitivity to business conditions.” This implies that performance of poorly governed firms is more sensitive to changes in market conditions. Hence, companies with weaker corporate governance will suffer more when the market performs badly. A final explanation is that investors overreact to shocks, and that this degree of investors’ overreaction is most noticeable regarding poorly governed companies.
more during the recovery period.” These results are consistent with the expropriation hypothesis. Controlling shareholders’ incentives to expropriate minority shareholders are a key factor implying a direct link between corporate governance and company performance during the crisis period.
Besides Mitton (2002) and Bae et al. (2012), there are several other academics having researched corporate governance in combination with company performance. Gompers et al. (2003) found that companies with stronger shareholder rights, have higher firm value, higher profits, higher sales growth, lower capital expenditures, and made fewer corporate acquisitions. It is important to note that around twenty years ago companies had almost no reason to restrict shareholder rights, since proxy fights and hostile takeovers were rare (Gompers et al., 2003). The increase in hostile takeovers required several companies to restrict shareholder rights and impose defence mechanisms (Gompers et al., 2003).
Bhagat and Bolton (2008) takes into account the inter-relationships between corporate governance, corporate performance, corporate capital structure and corporate ownership structure. In total, Bhagat and Bolton (2008) used seven different measures of governance, which are GIM indices, BCF indices, stock ownership of board members, CEO-chair separation, Brown and Caylor governance score, The Corporate Library benchmark score and board independence. The first four variables are positively correlated with operating performance, whereas board independence is negatively correlated with operating performance (Bhagat and Bolton, 2008). However, these results should be interpreted with ultimate precaution, since correlation does not, per se, imply a causal relationship.
Lemmon and Lins (2003) combine in their research ownership structure, corporate governance and firm value. Their study, again based on East Asian companies, shows that the crisis negatively impacted the number of investment opportunities companies had, resulting in an increase in incentives for controlling shareholders to expropriate minority shareholders (Lemon and Lins, 2003). Subsequently, “the large separation between cash flow ownership and control rights that arises from the use of pyramidal ownership structures in these markets suggests that insiders have both the incentive and the ability to engage in expropriation” (Lemon and Lins, 2003).
Erkens et al. (2012) were one of the first to link the financial crisis of 2008 with corporate governance. They based their research however on a limited number of observations per country and used only two measures of corporate governance (ownership structure and board independence). Another hiatus is that Erkens et al. (2012) only took financial institutions into account. Nevertheless, they show that “firms with more independent boards and higher institutional ownership experienced worse stock returns during the crisis period” (Erkens et al., 2012).
quality of (corporate) governance. They found empirical support from the European Union for their hypothesis that government ownership is associated with lower quality of corporate governance. Moreover, preferential voting rights of golden shares are significantly damaging to governance quality (Borisova et al., 2012).
2.4: Corporate governance in Europe and Asia
As discussed before, most studies investigating the link between corporate governance and company performance have been conducted in (East) Asia. However, corporate governance in Europe is different from corporate governance in Asia. Before the East Asia crisis, corporate governance quality in East Asia was relatively low, according to Wei and Zhang (2008). The crisis mandated governments to take necessary reforms. Quality of corporate governance mechanisms increased due to improved government regulations and scrutiny of shareholder rights activists (Wei and Zhang, 2008). For example, it became a requirement that independent directors would serve on company boards. Nevertheless, many corporate governance failures still existed, such as concentrated ownership and divergence between control rights and cash-flow rights of largest shareholders (Wei and Zhang, 2008).
Due to numerous factors, such as corporate scandals in the US (Enron, etc.) and outside the US (for example Parmelat), governments of advanced economies were forced to reform corporate governance policies and laws. The Sarbanes-Oxley Act (SOX) is one the most pronounced examples of these reforms (Kim and Lu, 2013). Also the countries in this research have seen similar reforms. In France the Law on Economic Regulations and Financial Security Law were passed (Kim and Lu, 2013). Germany installed a governmental commission creating a German Corporate Governance Code, to a great extent a German copy of SOX (Kim and Lu, 2013). After several scandals and acquiring a reputation for weak investor protection, the Ministry of Finance and Economic Affairs in the Netherlands released the Dutch Corporate Governance Code (Kim and Lu, 2013). This corporate governance code is a direct replacement of code Tabaksblat from 20032. Spain introduced the Corporate Governance Unified Code and earlier the Transparency Law to improve corporate governance practices (Kim and Lu, 2013). On the other hand, UK introduced the Cadbury report and Greenbury report, which was later merged to form the Combined Code (Kim and Lu, 2013). Kim and Lu (2013) show that all these different corporate governance reforms have had a significant impact on investor protection. They support this with cross-border acquisitions. If the bridge in terms of investor protection between acquirer and target country increases, foreign acquirers’ cherry picking tendency increases (Kim and Lu, 2013).
Enriques and Volpin (2007) describes the differences between corporate governance in continental Europe and corporate governance in the US. There are at least two key differences between both continents. First, most American companies are widely held, whereas their European counterparts have mostly controlling shareholders. Second, regulations on self-dealing are in general stricter in US companies, relative to continental European countries (Enriques and Volpin, 2007). As discussed before, if there are controlling shareholders within a company, the chance of expropriation of minority shareholders increases substantially (Mitton, 2002; Bae et al., 2012). Hence, appropriate corporate governance regulations are inevitable. Enriques and Volpin (2007) emphasize that large European countries, such as Italy, France and Germany have enacted several significant corporate law reforms, in order to “strengthen the mechanisms of internal governance, empower shareholders, enhance disclosure requirements, and toughen public enforcement”, with the main focus on empowering minority shareholders and disclosure. The latter two are according to Enriques and Volpin (2007) the most useful and effective tools to challenge misconducts by dominant and controlling shareholders.
2.5: Hypotheses
It is evident that there is abundant empirical support for the relationship between corporate governance and company performance. Companies with high quality corporate governance perform significantly better than companies with lower corporate governance quality during a financial crisis, according to La Porta et al. (1997, 1998, 1999b, 2000), Johnson et al. (2000), Mitton (2002) and Bae et al. (2012). During a financial crisis, incentives of controlling shareholders to expropriate their counter minority shareholders have the tendency to go up, as expected return on investments decrease (Johnson et al., 2000; Mitton, 2002; Baek, Kang and Park, 2004). With high quality corporate governance, the rights of minority shareholders can be (at least partially) protected. Thus, companies with poor corporate governance quality will perform worse during the financial crisis of 2008. Therefore, the first general hypothesis in this research should be:
Hypothesis 1: European companies with poor corporate governance quality (in terms of expropriation) will perform worse than European companies with high corporate governance quality during the financial crisis of 2008.
variables/mechanisms Bae et al. (2012) and Mitton (2002) used. First, as discussed previously in the theoretical section of this research, high disclosure quality leads to higher quality of corporate governance, and thus better performance during the financial crisis. Thus:
Hypothesis 2: European companies with high disclosure quality perform better than European companies with low disclosure quality during the financial crisis of 2008.
Moreover, as Bae et al. (2012) and Mitton (2012) argue, high disparity between cash flow rights (shareholder rights) and voting rights of (controlling) shareholders and a high percentage of block holdings, results in lower performance for companies during the crisis. Therefore:
Hypothesis 3a: European companies with low disparity between shareholder rights and voting rights of (controlling) shareholders will perform better than European companies with high disparity between cash flow rights and voting rights during the financial crisis of 2008.
And:
Hypothesis 3b: European companies with a low amount of block holdings (as a percentage) will perform better than European companies with a high amount of block holdings (as a percentage) during the financial crisis of 2008.
According to Mitton (2002) and Bae et al. (2012), the third mechanism is corporate diversification, although it is not per se a corporate governance mechanism. Lower transparency of diversified companies allows managers and/or controlling shareholders to take advantage of minority shareholders, due to higher levels of asymmetric information (Lins and Servaes, 2000; Lins, 2000). Hence, during a financial crisis, expropriation will possibly increase, but primarily arise among diversified companies (Mitton, 2002). Therefore, diversified companies are associated with worse stock price performance. Thus:
Hypothesis 4: European companies with no or low corporate diversification perform better than European companies with high corporate diversification during the financial crisis of 2008.
beginning, it is expected that these companies have performed worse than non-financial companies. Hence, as a final hypothesis:
Hypothesis 5: European non-financial companies have performed better than European Financial institutions during the financial crisis of 2008.
In the upcoming section, the methodology will be discussed. Proxies regarding the concept of corporate governance will be given. Moreover, every variable in the model will be conceptualized and discussed in terms of measurement.
3. METHODOLOGY
3.1: Data
In total, companies from five large European countries are used. These countries are: United Kingdom, Germany, France, Spain and the Netherlands. Since these five countries are significantly different from each other, they give a good representation of Europe. As Faccio and Lang (2002) highlight in their study regarding ultimate ownership between Western European countries, is that there are differences in ownership control, primarily between continental Europe and Anglo-Saxon UK. In order to examine the role of corporate governance in the financial crisis of 2008 it is therefore most useful to examine several different countries in terms of ownership control, since this is one of the aspects of corporate governance according to Mitton (2002) and Bae et al. (2012).
In order to have the highest chances of success regarding data availability, companies from the major stock exchanges are selected in this research sample. Companies are selected from seven major stock exchanges. Table 1 gives an overview of these stock indices.
Table 1: Data sample
Country Stock Exchange Number of observations
United Kingdom FTSE-100 101
Germany SBF-120 120
France HDAX (XETRA) 110
Spain IBEX-35 35
IBEX-Medium Cap 20
The Netherlands AEX Index 25
AMX Index 25
Total 436
No available returns/Cross-listing 40
The Financial Times Stock Exchange Index covers the 1003 largest, most actively traded companies on the London Stock Exchange. On the other hand, the Société des Bourses Françaises 120 Index (SBF-120) includes the 120 most actively traded companies on the stock exchange of Paris. It includes all 40 stocks from the CAC-40 and the remaining are from the medium-cap indices. The HDAX is the German counterpart of it. It is a combination of the 110 largest companies traded on the German stock exchanges. On the other hand, the AEX and AMX represent the main Dutch Stock Exchange and Medium-Cap index, respectively. The IBEX-35 is the index of the Spanish Stock Exchange. As the name already suggests, this index consists of 35 largest and most actively traded companies. At the same Stock Exchange, the IBEX-Medium Cap represents 20 smaller, but still actively traded companies.
In total, the seven indices sum up to 436 observations. Each of the indexes has been selected as a constituent list on Datastream. Due to a lack of available returns (dependent variable) and reduction of companies with a cross-listing4, this number reduced to 396. Returns were not available in some cases, since companies have had their initial public offering (IPO) somewhere after the starting period of the research, or left the exchange within the period of research. Besides returns, also all other variables or elements of variables are retrieved from Datastream. Sometimes, specific variables were not available for a specific company. In those cases, these companies were left out. Therefore, for each different model as illustrated in the results section of this research, the total number of observations fluctuates.
As discussed before, the starting point of the financial crisis in Europe is the 1st of January 2008. Although it is debatable whether this should be the exact day, somewhere around the 1st of January 2008 the financial crisis in the US evidently spread to Europe as can been seen in figure 1. To capture the trend regarding corporate governance in relationship with company performance, three periods are compared with each other:
Pre-crisis period: 1st of January 2007 – 31st of December 2007 Crisis period: 1st of January 2008 – 31st of December 2008 Post-crisis period: 1st of January 2009 – 31st of December 2009
3
Although it lists the 100 largest companies, it has 101 listings, since Royal Dutch Shell has listed both A and B shares on this exchange. Nevertheless, Royal Dutch Shell is only counted once in the observation/sample. 4
Figure 1: Shows prices of the five main stock price indices (AEX Index, FTSE-100, HDAX, IBEX-35 and SBF-120) from January 2007 to December 2009. The Y-axe represents the opening price of each individual stock index in their local currency. The X-axe coves the four years which are part of the three periods investigated in this research.
Figure 1 shows the five main indices5. What is most striking on figure 1 is that it can be seen that every Stock Exchange has faced a dramatic fall in value. This fall already started in the end of 2007,
5
AMX Index and IBEX-Medium Cap are not included in the figure since they follow a similar trend as the AEX, respectively the IBEX-35.
100 200 300 400 500 600
I II III IV I II III IV I II III IV
2007 2008 2009 AEX 3,000 4,000 5,000 6,000 7,000
I II III IV I II III IV I II III IV
2007 2008 2009 FTSE 100 1,500 2,000 2,500 3,000 3,500 4,000 4,500
I II III IV I II III IV I II III IV
2007 2008 2009 HDAX 6,000 8,000 10,000 12,000 14,000 16,000 18,000
I II III IV I II III IV I II III IV
2007 2008 2009 IBEX 1,500 2,000 2,500 3,000 3,500 4,000 4,500 5,000
I II III IV I II III IV I II III IV
2007 2008 2009
although then the crisis was not that severe in Europe, but its main focus was in 2008. Stock exchanges had their all-time low in the beginning of 2009. After that, stock exchanges recovered already at a slow pace. Therefore, the three reference periods have the beginning and end dates as mentioned above.
3.2: Dependent variable
Returns can be calculated in different ways. This research uses two different measures. First, as dependent variable in this research, the daily logarithmic returns of each individual stock on the seven different stock exchanges are used. These daily returns are then segmented for each of the three periods (pre-crisis, crisis and post-crisis period). For each period, an average daily logarithmic return can be calculated for each company. Hence, this results in 396 average daily logarithmic returns for each individual period. Formula 1 gives a mathematical representation of the daily logarithmic return:
)) (1)
(Where is the closing price of security i at day t)
The individual returns are the proxy of corporate performance. If individual returns are high, the individual company is performing significantly well.
Another way to calculate returns for each company is to use period returns. This is simply the increase or decrease in stock price with the first period day and last period day as reference points. Formula 2 shows how to calculate these period returns. Both ways are separately used in this research as a measure of corporate performance.
) (2)
(Where is the stock price of security i at end of period and the stock price of security
i at the beginning of the period)
3.3: Independent variables
separately. These two variables are whether a company has an American Depository Receipt (ADR) and corporate diversification. Since ADR is a substitute for legal protection as mentioned in La Porta et al. (1997, 1998, 1999b, 2000), according to Mitton (2002) and corporate diversification is not per se a corporate governance mechanisms (Mitton, 2002; Bae et al., 2012), these two variables are taken into account separately. These two variables are therefore discussed in the control variables section of this paper.
3.3.1: Variables regarding difference between shareholder rights and voting rights
The two main variables regarding this set of variables are the score on difference between shareholder rights and voting rights and voting cap percentage. The first variable measures whether there is a difference between the rights of the (controlling) shareholder, such as cash flow rights, and the right to vote. The variable is a percentage of shares that provide equal voting rights. The higher this number, the less likely it will become that controlling shareholders will expropriate, since there is less chance that controlling shareholders have specific shares, such as preferred stock, in which they can use their effected power to expropriate minority shareholders. Thus, companies with a relatively high amount of shares with equal voting rights will perform better (during a financial crisis) than companies lacking shares with equal voting rights.
Voting cap percentage refers to the level at which the voting cap has been set. It is the percentage of maximum voting rights allowed or ownership rights6. This variable will have values between 0 and 100. The higher the voting cap, the more power controlling shareholders can potentially have. Therefore, a negative relationship is expected.
Companies with voting cap refer to whether the company has a voting cap. The main question is: does the company have shares with a voting cap (ceilings) clause, ownership ceilings or control share acquisition provision7? This variable is a dummy variable. The company gets a value of 1 if it has a voting cap, and it gets a value of 0 if it has not a voting cap. A voting cap can prevent (controlling) shareholders of misusing their power to take decisions which are specifically harmful for minority shareholders. Hence, a voting cap can have a positive effect on corporate performance.
Companies with confidential voting policy is also a dummy variable. It refers to whether shareholders can vote confidentially or not, at for example, specific shareholder meetings. The exact definition refers to whether the company has a confidential voting policy (i.e., management cannot view the results of shareholder votes)8. As discussed before, also management can be a significant
6
part of the controlling shareholders block. If a company has a confidential voting policy this dummy variable will have a value of 1. If the company lacks a confidential voting policy, this dummy variable will have a value of 0. It is expected that companies with confidential voting policy have better quality of corporate governance, and thus higher firm performance, since with a confidential voting policy, minority shareholders have (in theory) less chance that they are pressured from controlling shareholders from management. Thus, chances of expropriation by controlling shareholders are reduced.
Companies with shareholder rights policy elements/equal voting right is a dummy variable referring to whether the company applies the one-share, one-vote principle. If a company does not apply this, this indicates that some shares have more power (one share is then combined with a number of votes higher than one) than other shares. This can increase inequality between controlling shareholders and minority shareholders. Hence, companies applying the one-share, one-vote principle will perform better.
Companies with shareholder/voting rights is related to the first independent variable. It is a dummy variable that defines whether all shares of an individual company provides equal voting rights. If this is the case the value becomes 1. If not all shares provide equal voting rights this value becomes 0. The expected results are the same as for the first variable. Companies with equal voting rights are expected to perform better during a financial crisis than companies who have differences in their voting rights policy.
This set of variables is a proxy of hypothesis 3a. They are all related to the disparity between shareholder rights and voting rights of shareholders.
3.3.2: Variables regarding related to block-holdings
Main variable regarding the set of variables related to block holdings is the voting rights of the single biggest owner. This variable represents the percentage of voting rights the biggest owner of the individual company has. The higher this number, the more power a shareholder has. Thus, it is a proxy of how controlling the shareholder can be. Moreover, the higher this number, the higher the chance that the (controlling) shareholder has the ability to expropriate minority shareholders. Hence, companies in which their biggest owners have relatively high voting power will perform worse than companies in which the biggest owners have relatively less voting power (or in which the biggest owners are relatively small, compared with other shareholders).
Companies with multiple or double voting rights shares is a dummy variable that takes the value of 1 if the company multiple (double) voting right shares9. If the individual company does not
possess multiple or double voting right shares, this dummy variable takes the value of 0. If a company possess multiple voting right shares, there is a significant chance that these shares are securities belonging to controlling shareholders. Under those circumstances, controlling shareholders can again use their power to expropriate minority shareholders, since the latter group shareholders have less voting rights. Thus, companies with multiple voting rights shares will perform worse during the financial crisis than companies who do not have multiple voting right shares outstanding.
Companies with non-voting shares refers to whether the company has non-voting rights common shares. This is loosely related with the previous variable. Also this variable is a dummy variable. If the company has non-voting shares, this variable takes the value of 1. If the company does not possess non-voting shares, this variable will have the value of 0. Non-voting shares can implicate that significant power belongs to some specific shareholders, since these shareholders have shares which have voting power. With this voting power, specific shareholders, who can be of course controlling shareholders, can expropriate for example minority shareholders with non-voting shares. Hence, therefore it is expected that companies with non-voting shares will perform worse than companies who do not have non-voting shares outstanding.
This set of variables is a proxy of hypothesis 3b. They are all related to block holdings and ownership of significantly large shareholders.
3.4: Control variables
As in every other research, some general and more specific control variables are included in order to acquire the optimal results. One of the main risks of not including control variables is that there are outside factors that have a significant effect on the dependent variable, or are correlated with one or more independent variables. This would significantly affect the results and would make conclusions evidently less useful to interpret. The following control variables are included, just as in Mitton (2002), Baek et al. (2004) and Bae et al. (2012), that can effect firm performance during each of the three periods (pre-crisis, crisis and post-crisis period.
As discussed in Mitton (2002) and Bae et al. (2012), firm risk is important to take into account. Beta is a renowned measure of firm risk. The beta of a security or a company measures to what extent the individual stock return is affected by swings from the market (market return). Hence, beta measures the volatility of a stock, or so-called systematic risk. Instead of using the beta for each individual company stated in Datastream, the beta for each individual security has been calculated10.
10
Formula 3 gives a mathematical definition of the relationship of the stock return with the market return (as a regression):
(3)
(In which is the daily stock return of security i, is the daily stock index return11, ,
called beta, the slope of the regression, and the intercept of the regression)
Thus, in total, for each company, three betas have been calculated; one beta for the pre-crisis period, one beta for the crisis period and one beta for the post-crisis period. It is expected that companies with a high beta will perform worse during the crisis (high risk results in high decreases in stock returns) but will stronger recover after the crisis.
As Baek et al. (2004) and also Bae et al. (2012) highlight, results should be controlled for firm size. “Larger firms tend to have easy access external finance and suffer less from information asymmetry,” according to Bae et al. (2012). Hence, it is expected that larger companies will be less vulnerable to external shocks, such as a financial crisis. Size is measured as the logarithm of total assets, which is in accordance with Bae et al. (2012).
Firms with high leverage will find it difficult to attract new ways of external financing during a crisis. This is logical, since the chance of bankruptcy will be higher than their counterparts with relatively low leverage. Therefore, financial institutions, who also have (had) their financial problems during the financial crisis of 2008, are more willing to lend money to companies with low leverage. Hence, since highly leveraged companies will find it difficult to attract new external financing during a crisis, it is expected that they will experience a larger drop in equity value/company performance. There is also empirical support for this hypothesis. According to, Ivashina and Scharfstein (2008) the financial crisis of 2008 caused a drop in long term-debt to large borrowers by 47 percent in the fourth quarter of 2008, relative to the third quarter of 2008. Relative to a year ago, bank lending dropped by even 79 percent (Ivashina and Scharfstein, 2008). Hence, it becomes significantly harder to finance activities with long-term debt. Leverage is measured in the same way as Bae et al. (2012) did: total debt divided by total assets.
11
Daily stock index returns are calculated as follows:
)) (4)
Moreover, liquidity (or cash flows) can have an effect on firms’ stock returns. Firms with higher liquidity are less dependent on external financiers. Hence, their stock returns are expected to be higher during the financial crisis, since external financiers are during these periods more anxious to lend money, as discussed previously with the empirical support of Ivashina and Scharfstein (2008). Companies with high liquidity will have at least still opportunities to invest, unlike companies with almost no liquidity. Liquidity is measured as the ratio of cash flow (operating income12 plus depreciation) relative to total assets (Bae et al., 2012).
Subsequently, future investment opportunities can have an impact on company performance. To control for future investment opportunities, Tobin’s Q is added as a variable. Tobin’s Q is calculated as follows:
13
(5)
If Tobin’s Q is below 1, the company has not many future investment opportunities. If the variable takes values of higher than 1, the company has several future investment opportunities which can be attractive to increase sales and thus returns. Hence, companies with a high Tobin’s Q are expected to have higher stock returns.
Percentage of firms with an ADR is also a dummy variable, and is a proxy of hypothesis 2. The dummy variable takes a value of 1 when the company has indeed an American Depository Receipt. This is mostly the case for the largest companies, since the costs of doing so are often prohibitive according to La Porta et al. (1999a). If a company has not an ADR, the dummy variable takes a value of 0. As discussed before, ADR is a proxy of disclosure quality. Companies with an ADR are expected to have better disclosure quality, formally through mandated disclosure requirements of the listed (stock) exchange and informally through investors demanding higher quality of disclosure (Coffee, 1999; Mitton, 2002). Hence, an ADR will improve the corporate governance framework of firms (Gozzi et al., 2008). Thus, companies with an ADR are expected to suffer less than companies lacking an ADR14.
12
Operating income represents in this case the difference between sales and total operating expenses. 13
Another mathematical description of Tobin’s Q:
(6) 14
Moreover, companies are divided in two groups regarding diversification. One group of companies that is diversified and one group of companies that is undiversified. Diversification of business activities is based on SIC codes. Datastream provides sales for in total eight general segments, based on their SIC code. If more than 90 percent of sales are coming from one of these eight segments, companies are defined as undiversified. In that case, the dummy variable takes the value of 0. If less than 90 percent of sales are coming from one of the eight segments, thus, if more than 10 percent of sales are coming from one or more other segments, the company is defined as diversified. Then the
Table 2: Variables
Datastream name Datastream code Stated name in paper Expected sign
Stock Price P Stock price and used for beta -
Confidential Voting Policy CGSRDP062 Confidential voting policy (dummy) Positive Multiple or Double Voting Rights
Shares CGSRDP024
Multiple or double voting rights
shares (dummy) Negative Non-Voting Shares CGSRDP023 Non-voting shares (dummy) Negative Score - Shareholder Rights/Voting
Rights CGSRO02S Score shareholder rights/voting rights Positive Shareholder Rights Policy
Elements/Equal Voting Right CGSRDP0011
Shareholder rights policy
elements/equal voting right (dummy) Positive Single Biggest Owner Voting Rights CGSRDP046 Voting rights biggest owner Negative Voting Cap (Dummy) CGSRDP026 Companies with voting cap (dummy) Positive Value - Shareholder Rights/Voting
Rights CGSRO02V
Shareholder rights/voting rights
(dummy) Positive
Voting Cap Percentage CGSRDP027 Voting cap percentage Negative
INDICATOR - ADR X(WC11496) ADR (dummy) Positive
Total liabilities X(WC03351) Used for leverage
Total assets X(WC02999) Used for leverage, size and Tobin's Q Operating income X(WC01250) Used for liquidity
Depreciation X(WC01148) Used for liquidity Market capitalization X(WC08001) Used for Tobin's Q
General industry classification X(WC19596) Used for financial (dummy) SIC CODE 1 X(WC19516) Used for diversified (dummy) SIC CODE 2 X(WC19526) Used for diversified (dummy) SIC CODE 3 X(WC19536) Used for diversified (dummy) SIC CODE 4 X(WC19546) Used for diversified (dummy) SIC CODE 5 X(WC19556) Used for diversified (dummy) SIC CODE 6 X(WC19566) Used for diversified (dummy) SIC CODE 7 X(WC19576) Used for diversified (dummy) SIC CODE 8 X(WC19586) Used for diversified (dummy)
Other (control) variables Expected sign
Liquidity Positive
Leverage Negative
Size Positive
Tobin's Q Positive
Beta
Negative (during the crisis)
Diversified (dummy)
Negative (during the crisis)
Financial (dummy)
dummy variable takes a value of 1. The degree of expropriation of minority shareholders by controlling shareholders can be more severe among highly diversified companies. This is because these diversified firms are characterized by more information asymmetry and agency costs (Lins and Servaes, 1999; Mitton, 2002; Bae et al., 2012). Lower transparency of diversified companies allows managers and/or controlling shareholders to take advantage of minority shareholders, due to higher levels of asymmetric information during a financial crisis (Lins and Servaes, 2000; Lins, 2000). Hence, diversified companies are expected to suffer more than undiversified companies during a financial crisis. Therefore, the corporate performance of diversified firms will be lower than their undiversified counterparts. This variable is testing hypothesis 4.
In order to measure whether there is a difference in performance between financial and non-financial companies, a new dummy variable15 has been added. Based on the general industry classification provided by Datastream, companies are either divided in the group financials or the group non-financials16. The dummy variable takes a value of 1 if the company has its main activities in the financial industry. If the company is primarily active in a non-financial industry, the dummy variable takes a value of 0. This variable measures hypothesis 5.
Table 2 summarizes all the variables used from datastream, including their name in this research, datastream code and expected sign/effect on the dependent variable (stock return). The latter part of the table summarizes variables which rely on datastream data, but are not direct variable in datastream. Hence, they had to be calculated separately. Their signs are also summarized in the latter part.
3.5: Cross-sectional regression
Besides measuring whether there are significant differences in each individual variable during the three defined periods of research, different cross-sectional regressions are done. Such a regression can be done, since the data collected for each of the three periods provides cross-sectional data. Cross-sectional data means that there is data on a set of different variables collected at a single point in time (Brooks, 2008:5). This is the case in this research, since for each period (although consisting of a various daily data), an average has been calculated. With these adjustments/calculations,
15
sectional data arises. For each time period (pre-crisis, crisis and post-crisis period), there is one single value for each variable (dependent, independent and control variables). With this data, cross-sectional regressions can be done. The linear regressions measure whether the independent and control variables have a significant impact on the dependent variable (corporate performance).
3.6: Outliers
As in several other financial research papers, outliers can be a significant problem in (linear) regression models. One way to overcome this problem is by winsorizing the outliers. Winsorizing sets outliers to a pre-determined percentile of the data, for example 98 percent. In this case, all data below the 1st percentile are set to the 1st percentile, whereas all data above the 99th percentile are set to the 99th percentile. In this light, the winsorizing technique, which is often used by financial researchers, is different from trimming or truncation. In the latter technique, outliers are simply removed, and thus the number of observations decline. With winsorizing, this is not the case. Also in this research, winsorizing is used to adjust the regressions for outliers. This is done for the 1% lowest outcomes in returns and the highest 1% in returns (this is both done for average daily returns and period returns). In the next section, the results are presented.
4. RESULTS
4.1: Descriptive statistics and correlation
Table 3 summarizes the main descriptive statistics17. When taking a better view at both dependent variables (average daily returns and period returns) it can be seen that the means and medians of each period are significantly different from each other. Moreover, the returns are positive in the period before and after the crisis, and they are negative in the crisis. This was already illustrated by figure 1.
For the independent variables, almost no significant results can be found. Only the score between shareholder and voting rights has significant different medians. The remaining of the independent variables shows no significant differences between its means or medians.
On the other hand, more significant results can be found among the control variables though. First of all, the beta has a significant difference in means and medians between the pre-crisis period and the crisis period, whereas it also has a significant difference in medians between the crisis period and the post-crisis period. Moreover, Tobin’s Q values decreased significantly during the crisis. So,
17
***) Significant at a significance level of 1 % **) Significant at a significance level of 5 % *) Significant at a significance level of 10 %
Table 3: Descriptive statistics
A. Pre-crisis period B. Crisis period C. Post-crisis period Test of difference (A-B): p-value Test of difference (B-C): p-value
Mean Median Mean Median Mean Median t-test
Wilcoxon
Z-test t-test
Wilcoxon Z-test
Dependent variable
Average daily return 0.0000 0.0000 -0.0024 -0.0022 0.0013 0.0010 0.0000*** 0.0000*** 0.0000*** 0.0000*** Period return 0.0014 0.0129 -0.6135 -0.5597 0.3320 0.2715 0.0000*** 0.0000*** 0.0000*** 0.0000***
Expropriation variables
Score shareholder rights/voting rights 36.4308 70.0400 36.7476 69.3300 36.3881 67.7100 0.7123 0.0001*** 0.9966 0.0000*** Voting rights single biggest owner 16.3335 8.8950 17.3708 9.8900 17.7616 9.8200 0.4818 0.4923 0.7899 0.8898 Voting cap percentage 68.8711 100.0000 72.6431 100.0000 73.2623 100.0000 0.2557 0.3688 0.8422 0.8729 Percentage of companies with confidential
voting policy (dummy) 1.0135 2.1277 2.7027 0.2803 0.9082
Percentage of companies with multiple or double
voting rights shares (dummy) 12.5320 12.8141 12.3457 0.9054 0.8416
Percentage of companies with non-voting shares
(dummy) 2.3018 2.2444 1.9656 0.9569 0.7828
Percentage of companies with shareholder rights
policy elements/equal voting right (dummy) 59.8465 62.6866 65.1961 0.4124 0.4577 Percentage of companies with voting cap
(dummy) 5.3708 5.5138 5.7072 0.9296 0.9054
Percentage of companies with value -
shareholder rights/voting rights (dummy) 50.8951 53.2500 53.9409 0.5081 0.8443
Control variables Beta 0.6663 0.6136 0.6074 0.5382 0.7543 0.6601 0.0060*** 0.0014*** 0.0000*** 0.0000*** Size 6.7805 6.7061 6.8104 6.7341 6.8259 6.7252 0.6227 0.6086 0.7971 0.7774 Leverage 0.6327 0.6306 0.6567 0.6612 0.6416 0.6305 0.157 0.0524* 0.3917 0.1756 Liquidity/Cash flow 0.1043 0.0960 0.0940 0.0906 0.0867 0.0789 0.1518 0.2188 0.3027 0.0627* Tobin's Q 1.7824 1.4319 1.3531 1.0778 1.5168 1.1915 0.0000*** 0.0000*** 0.0151** 0.0000*** Percentage of firms with an American Depository
Receipt (dummy) 33.0275 33.0275 33.0275 1.0000 1.0000
Percentage of diversified firms (dummy) 91.5138 91.5138 91.5138 1.0000 1.0000
companies had probably less investment opportunities during the crisis. This is as expected beforehand. After the crisis, Tobin’s Q increases significantly. The other variables show almost no significant differences between each period (except for median differences in liquidity/cash flow and leverage). Hence, conclusions cannot be made regarding these variables. The percentage of firms having an ADR, being diversified, or being mainly active in the financial industry, have not changed over the three periods. Hence, these results are, of course, insignificant. Notwithstanding, these variables can be still a significant factor affecting company performance. Since the number of companies with confidential voting policy and the non-voting shares is significantly low, reliable conclusions regarding the effect of these variables on company returns are difficult to make. Therefore, these two variables are not taken into account in the regression analysis.
Table 4 summarizes the correlation coefficients for all independent variables in the pre-crisis period18. The different independent variables show no strong correlation with each other. Hence, all variables can be used in the regression. There is only one exception. Voting cap percentage shows in some cases strong correlation with some of the other independent variables. Since voting cap percentage is one of the explanatory independent variables, this variable will be used as a single variable in the linear regression (together with the controlling variables), but without the other explain variables. Thus, in that case, the results are not affected by the high correlation between voting cap percentage and the other explanatory variables.
4.2: Cross-sectional regression
Table 5 shows the first set of independent variables (those related with the difference between shareholder rights and voting rights) in relationship with company performance for each of the three periods. For each period it uses three different models. The first model is linear regression measuring to what extent the first main independent variable (score of difference between shareholder rights and voting rights) influences corporate performance. This model includes also the control variables. The second model includes the second main independent variable (voting cap percentage), including the control variables. The third model is the most extensive one. This includes besides the two main independent variables and control variables, also the other independent variables related with shareholder rights, voting rights and voting policy. As dependent variable in all three regression models, period returns are used, since the explanatory variables show higher coefficients when using period returns as dependent variable. This is logical, since the mean in period returns is significantly higher than the mean in average daily returns (as can be seen in table 3). Moreover, winsorization of
18
Table 4: Correlation coefficients independent variables (pre-crisis period) Pre-crisis period ADR (dummy) Beta Voting rights biggest owner Shareholder rights policy elements/equal voting right (dummy) Financial (dummy) Diversified (dummy) Leverage Liquidity/Cash flow Multiple or double voting rights shares (dummy) Period returns Shareholder rights/voting rights (dummy) Score shareholder rights/voting rights Size Tobin's q Companies with voting cap (dummy) Voting cap percentage ADR (dummy) 1.0000 Beta 0.3022 1.0000
Voting rights biggest
owner -0.0150 0.0253 1.0000
Shareholder rights policy elements/equal voting right (dummy) 0.2840 0.3156 0.4768 1.0000 Financial (dummy) 0.0238 0.2387 -0.1257 0.0321 1.0000 Diversified (dummy) 0.0423 0.0900 0.0979 0.0428 -0.0269 1.0000 Leverage 0.0422 0.0357 0.2435 0.2481 0.2244 0.1581 1.0000 Liquidity/Cash flow 0.0154 -0.0585 0.2006 0.2109 -0.3091 0.0325 0.1632 1.0000
Multiple or double voting
rights shares (dummy) 0.1738 -0.0752 0.0556 -0.1811 -0.0588 0.0962 -0.0066 0.0318 1.0000
Period returns 0.1094 0.0303 0.1884 0.2086 -0.1868 0.0776 0.0471 0.3240 0.0292 1.0000 Shareholder rights/voting rights (dummy) 0.1746 0.2652 0.2299 0.7138 0.0434 -0.0048 0.2004 0.1014 -0.2492 0.0881 1.0000 Score shareholder rights/voting rights 0.1853 0.2723 0.2445 0.7288 0.0447 -0.0007 0.2065 0.1064 -0.2341 0.0939 0.9995 1.0000 Size 0.5293 0.2919 0.2798 0.5145 0.3999 0.1406 0.3517 -0.0566 0.1811 0.1115 0.3021 0.3206 1.0000 Tobin's q -0.1065 -0.0328 0.0816 0.0335 -0.2322 -0.0338 0.1526 0.6400 -0.1044 0.1810 0.0534 0.0505 -0.3279 1.0000
Companies with voting cap
(dummy) 0.1755 -0.0565 -0.0114 -0.0332 0.1010 0.0873 0.1140 0.0378 0.4161 0.0437 -0.2259 -0.2122 0.2570 -0.0395 1.0000
Table 5: Return and shareholder rights/voting rights (period returns)
A. Pre-crisis period B. Crisis period C. Post-crisis period
[1] [2] [3] [4] [5] [6] [7] [8] [9]
Constant -0.7844 -0.7192 -0.7217 -0.7589 -0.6287 -0.4948 0.7848 0.7659 0.6869
-4.5964 *** -3.7740 *** -3.5080 *** -3.5447 *** -2.7718 *** -2.0770 ** 4.8691 *** 4.5565 *** 3.8380 ***
Expropriation variables
Score shareholder rights/voting rights 0.0001 -0.1211 0.0006 -0.0002 -0.0002 -0.0013
0.1131 -1.5524 0.8188 -0.0597 -0.5076 -0.6981
Voting cap percentage 0.0001 0.0032 0.0014 0.0025 -0.0003 -0.0003
0.1839 1.4438 2.2745 ** 2.4847 ** -0.7032 -0.4758
Companies with shareholder rights policy
elements/equal voting right (dummy) 0.0440 -0.0040 -0.0581
0.6632 -0.0480 -0.9252
Companies with voting cap (dummy) 0.2897 0.2000 -0.0786
1.5520 1.7569 * -0.9509
Companies with value - shareholder
rights/voting rights (dummy) 0.0081 -0.0412 0.0118
Table 5 shows the coefficient in the cross-sectional regression with period returns as the dependent variable. Moreover, winsorization has been used to handle outliers in the dependent variable. The value underneath each regression coefficient represents the individual t-value. The asterisks behind the t-value highlight to what extent the variable/coefficient is significant in the regression model. There are three significance levels:
***) Significant at a significance level of 1 % **) Significant at a significance level of 5 % *) Significant at a significance level of 10 %
Adjusted R-squared 0.1324 0.1349 0.1325 0.1617 0.1720 0.1711 0.1738 0.1761 0.1773
F-statistic 6.2035 *** 5.6439 *** 4.1481 *** 8.0097 *** 8.3150 *** 5.9869 *** 8.8093 *** 8.7884 *** 6.3556 ***
outliers as discussed in the methodology section has already been done. Compared with the original results, this has not led to significantly other outcomes.
Regarding the first period, the pre-crisis period, the first model shows that the score of the difference between shareholder rights and voting rights does not influence stock price return at all. Even the coefficient is near zero. The second model shows that also the second main independent variable (voting cap percentage) has no significant effect. This does not change when adding the additional independent variables (model three). Also the additional variables show no significant support. Three other (control) variables show significant support during the pre-crisis period. First, larger firms tend to perform better than smaller companies. Second, companies with a high Tobin’s Q, and thus more future investment opportunities, perform better than companies with a low Tobin’s Q. And finally, financial firms perform significantly worse than non-financial companies. This is not strange, since the financial crisis was triggered by financial institutions. Apparently, these companies were already harmed during 2007 (pre-crisis period).
During the crisis period, the difference between shareholder rights and voting rights remains insignificant (model one). Nevertheless, voting cap percentage becomes a significant factor. Companies with a high voting cap percentage perform better than companies with a relatively low voting cap percentage. This is somewhat counterintuitive, since it was expected that a high voting cap percentage would have a negative effect on company performance. However, at the same time, model three shows that companies who have at least a voting cap, perform significantly better than companies lacking a voting cap. Subsequently, it can be seen that both beta and leverage have a significantly negative effect on company performance. This is as expected. First, companies with a high beta will outperform the stock market during good times, but underperform the stock market during bad times (such as the financial crisis of 2008). Second, highly leveraged companies find it difficult to attract new capital for investment opportunities, especially during a financial crisis. Hence, they perform significantly worse. Moreover, as expected, highly liquid companies and companies with a high Tobin’s Q perform significantly better.
Not a single explanatory variable is significant during the post-crisis period. Nevertheless, several control variables are significant. First, companies with a high beta perform better. Hence, after the crisis, companies with high systematic risk recover faster than more conservative companies. Additionally, large companies find it apparently difficult to recover from the crisis, since they perform significantly worse than smaller companies. As expected, companies with a high liquidity/cash flow and low leverage perform significantly better.
Table 6: Return and corporate ownership (period returns)
A. Pre-crisis period B. Crisis period C. Post-crisis period
[1] [2] [3] [4] [5] [6]
Constant -0.6604 -0.6821 -0.9906 -1.0154 0.9539 0.9343
-3.5698 *** -3.6451 *** -4.4006 *** -4.4446 *** 5.5232 *** 5.4312 ***
Expropriation variables
Voting rights single biggest owner 0.0011 0.0011 -0.0031 -0.0028 0.0020 0.0020 1.0199 1.0706 -2.5018 ** -2.2595 ** 2.3482 ** 2.3116 ** Companies with multiple or double voting rights
-3.9719 *** -4.0162 *** 0.5505 0.6418 -0.3713 -0.1532
R-Squared 0.1563 0.1589 0.1967 0.1981 0.2160 0.2140
Adjusted R-squared 0.1279 0.1271 0.1730 0.1716 0.1933 0.1885
F-statistic 5.4957 *** 4.9891 *** 8.3242 *** 7.4623 *** 9.5223 *** 8.3882 ***
Number of observations 277 275 316 313 321 319
Table 6 shows the coefficient in the cross-sectional regression with period returns as the dependent variable. Moreover, winsorization has been used to handle outliers in the dependent variable. The value underneath each regression coefficient represents the individual t-value. The asterisks behind the t-value highlight to what extent the variable/coefficient is significant in the regression model. There are three significance levels:
