The impact of corporate governance on REITs performance
around the financial crisis
Master thesis
Bart Joris - 11423021
University of Amsterdam Faculty Economics and Business
MSc Finance Real Estate Finance Track
Thesis Supervisor: Dr. F.P.W. (Frans) Schilder
2
Statement of Originality
This document is written by student Bart Joris who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
This study investigates the impact of corporate governance on the performance of Real Estate Investment Trusts (REITs) around the global financial crisis of 2008. Some researchers did already investigate the impact of corporate governance on firm performance and specifically for REITs, but the influence of a financial crisis on this impact has not been studied in detail. Three different models are used to test if the impact of corporate governance on REIT performance differs when the market has to deal with an economic crisis. A regression model based on the paper of Bauer, Eichholtz and Kok (2010), a difference-in-difference model and an adjusted version of the Carhart (1994) four factor model are used to test whether the impact of corporate governance is significant. The results show that there is no significant difference in operating performance of REITs with different corporate governance levels for the total sample period. The operating performance is measured by the funds from operations. Also during the financial crisis, there is no statistically significant difference in performance between REITs with different corporate governance structures. Anti-takeover indicators have a negative impact during the crisis, but this was the only significant result for this period. All REITs underperformed in terms of stock returns compared to the broader market portfolio and REITs with high corporate governance levels did even worse in comparison to the low corporate governance REITs.
Keywords: Real Estate Investment Trusts (REITs), corporate governance, board characteristics, anti-takeover, ownership concentration
4
Table of contents
Abstract ... 3 1. Introduction ... 5 1.1 Research Question ... 5 1.2 Background information ... 6 1.2.1 REIT ... 6 1.2.2 Corporate Governance ... 6 1.3 Thesis structure ... 7 2. Literature review ... 8 2.1 REIT performance ... 8 2.2 Corporate governance ... 9 2.2.1 Board characteristics ... 9 2.2.2 Ownership concentration ... 102.3 Relationship corporate governance and REIT performance ... 11
2.4 Thesis contribution ... 12
3. Methodology and Data ... 14
3.1 Overview methodology ... 14 3.2 Model specification ... 15 3.3 The Data ... 17 3.3.1 Summary Statistics ... 18 3.3.2 Data Shortfalls ... 20 4. Results ... 22 4.1 Results model 1 ... 22
4.1.1 One corporate governance indicator ... 22
4.1.2 Four corporate governance indicators ... 24
4.1.3 Mean corporate governance values ... 25
4.2 Results model 2 ... 28
4.3 Results model 3 ... 29
5. Conclusions ... 31
5.1 Limitations of this study / Possibilities for further research ... 33
6. References ... 34
5
1. Introduction
1.1 Research QuestionThe collapse of the housing market in 2007, in combination with the sub-prime mortgage crisis in the US, were two of the major causes of the credit crunch. This financial crisis changed a lot in the global financial markets all over the world. There already exists a significant body of literature about the impact of this crisis. The real estate market played an important role, especially in the case of decreasing property values and in the mortgage crisis. The value of a lot of firms also declined during this crisis and this firm value is influenced by a lot of different factors.
Jensen (1993) was one of the first researchers who found a relationship between ownership structure and the value of a firm. Later, also Erkens, Hung and Matos (2012) showed that corporate governance has a major impact on the performance of financial firms. They found that firms with more independent boards and a higher institutional ownership had worse stock returns during the global crisis. Firms with these characteristics took larger risks during the pre-crisis period which resulted in larger losses during the crisis. Firms with more independent boards also raised more equity capital compared to debt during the crisis and this led to a transfer of wealth from the existing shareholders to debt holders.
The real estate market played a major role in the financial crisis. A phenomenon that gets increasing attention nowadays in the real estate market is the existence of Real Estate Investment Trusts (REITs). The NAREIT1 is an association that presents detailed information about REITs and other publicly traded companies in the real estate market. Their main focus is to give information about the real estate market in the United States. NAREIT defines a REIT as: “a company that owns, operates or finances income-producing real estate. Modeled after mutual funds, REITs historically have provided investors of all types regular income streams, diversification and long-term capital appreciation.”2 A more detailed explanation of REITs will be given in the background information section (1.2).
An interesting subject to investigate would be the combination of corporate governance and the performance of REITs with a sample period around the global financial crisis. This sample will be split into three different periods: pre-crisis, during the financial crisis and after the crisis.
The main research question that follows is:
“To what extent does corporate governance influence the financial performance of REITs in the United States during the period around the global financial crisis? ”
To be able to answer this main research question, there are some challenges which have to be tackled. The first one is to choose specific model(s) to give an insight in the difference in performance between the REITs with different corporate governance structures. By using these models, we first have to investigate what the overall relationship between corporate governance and the performance of REITs in the United States is during
1National Association of Real Estate Investment Trusts (NAREIT) 2
6
the total sample period between 2000 and 2015. To answer the main question, differences in this relationship between the three different periods (before, during and after the crisis) have to be investigated.
Studying the corporate governance structure and the influence on REIT performance is relevant because of the growing interest in REITs. Increasingly more international property flows are allocated through indirect property vehicles and REITs are one of the major examples of such a vehicle. Directly owned property investments become less attractive for investors nowadays. Investigating the impact of the corporate governance structure on REITs with a recent sample period could be a useful addition to the already existing literature.
1.2 Background information
This section will give a briefly description of what REITs are and what is meant by the corporate governance structure. The relationship between the two and the related literature about these concepts will be described in detail in the second chapter of this master thesis. Some theoretical background is needed to be able to understand the concepts of REITs and the corporate governance structure in a better way.
1.2.1 REIT
The impact of corporate governance on Real Estate Investment Trusts (REITs) will be investigated. There are two main types of REITS, which are Equity REITs and Mortgage REITs. These two types give investors the possibility to invest in either the equity or the debt financing parts of the real estate markets. Equity REITs acquire commercial properties (for example offices or apartment buildings) and lease the space to different tenants. Income generated by an Equity REIT stems from rents or sales of the properties they own. These REITs pay out most of their income to their shareholders as dividends. On the other hand, mortgage REITs invest in real estate mortgages or mortgage-backed securities (MBS). The interest on these investments as well as the sales of these mortgages, create the income stream for the mortgage REITs. Like Equity REITs, Mortgage REITs also pay the bulk of their generated income as dividends to their shareholders annually.
As mentioned already, the NAREIT is an association that represents REITs and other publicly traded real estate companies which focuses on the real estate and capital markets in the United States. Members of the NAREIT are companies throughout the whole world that own, operate or finance real estate that produce income streams. But this association also contains companies who advise, study and service those real estate businesses.
1.2.2 Corporate Governance
Shleifer and Vishny (1997) published a survey research on corporate governance which formed the basis for a lot of papers about this subject. It focuses on the importance of legal protection and ownership concentration. Shleifer and Vishny state that corporate governance deals with the different ways in which financiers assure themselves a good return on their investment. They look at corporate governance from the agency perspective, which separates the ownership and control of a firm. Investors want to get their invested money back from the managers of the firm and in addition want to receive some additional payments (dividends for example). Differences in investors’ interests and the interests of
7
the managers could lead to agency problems, which could result into lower returns on investments.
Major points of interests in the topic of corporate governance are the observable attributes of board governance. Examples are independent chairmen, the size of the board and the percentage of outside directors on the board. Corporate governance systems which are successful are characterised by a significant legal protection of some investors at least, with an important role for large investors. Limited legal protection could minimize the attractiveness of an investment and could lead to limited external financing possibilities, especially when the firms are led by family or insider-dominated managers.
Shleifer and Vishny (1997) based their paper on the one of Jensen and Mecklin (1976) which also focused on the agency theory that separates the ownership and control of a specific firm. The use of this separation of control and ownership as a firm has led to greater growth in market value of the corporations that used this form. A drawback of this separation is the fact that agency costs also arise when you use this form of a firm structure. During the financial crisis the market environment demanded a lot from the corporate governance structures of the firm. The boards of the firms had to set up a clear strategy to manage the risk the company faced during this crisis. Efficient reporting systems were required to respond adequately and on time to these different risks. But while the demand for a good corporate governance system was high, evidence points out several weaknesses of these structures in a company, as shown by Kirkpatrick (2009). The corporate governance arrangements did not serve their purpose to control for excessive risk taken by managers of different financial services companies. Whether the corporate governance structure influenced the REIT performance around the financial crisis is investigated in this paper.
1.3 Thesis structure
The next chapter of this thesis gives an overview of the existing literature related to the research on REIT performance, corporate governance, the relationship between those two concepts and the contribution of this thesis to the already existing literature. After the related literature has been discussed, the methodology used to estimate the relationship between corporate governance and REIT performance will be described. The data used and the main results of this research are presented after the model specification. The discussion and overall conclusion form the last part of this thesis. Also the limitations of this paper and possibilities for further research will be discussed at the end.
8
2. Literature review
There is already some relevant literature about the relationship between corporate governance and real estate investment trusts (REITs), but the existing literature is not that extensive. Most of the literature focuses on the relationship between corporate governance and the financial performance of other companies instead of REITs. The existing literature about the relation between corporate governance and REITs use sample periods from before the crisis. It is therefore useful to investigate the change in this relationship around a global financial crisis.
The aim of this section is to provide a theoretical background and explain the insights given by prior literature about REITs and corporate governance. First, information about REITs, and especially their performance, will be provided. After that the corporate governance structure of a firm will be described as well as the impact of this structure on the performance of different kind of firms. This literature review will be used to derive the hypotheses that will be tested in our models.
2.1 REIT performance
REITs are companies that operate in a legally restricted setting. Because REITs are listed property companies, their managers are limited in their managerial freedom. This setting possibly reduces the agency problems as mentioned by Bauer, Eichholtz and Kok (2010). Agency problems could affect the financial performance of companies, so we have to keep in mind this legally restricted setting in which REITs operate. But the setting is not the only factor that influences the performance of REITs, other factors are also mentioned in the existing literature.
A lot of this literature pays attention to the performance of REITs in general. The impact of different firm-specific characteristics that play a role in this performance is also subject of a lot of related literature already. Han and Liang (2009) investigated the long-term performance of REITs during the period 1970-1993. Especially the stability of this performance was studied with the Jensen index as a specific performance measure in their study. Their findings suggest that the performances of REITs in their sample was similar to the performance of a passively managed portfolio containing 3-months treasury bills and a stock market portfolio. Equity REITs did outperform mortgage REITs and the REIT performance was not stable over the whole sample period. Especially the short-term performance did vary substantially over time and during different market situations.
The article written by Kuhle, Walther and Wurtzebach (2009) gives an update to the performance evaluation of REITs. Main reasons for their update are the fact that REITs are considered as well-performing firms on a risk-adjusted basis, which could suggest an imperfection which is not in line with the efficient market hypothesis. And secondly, the literature about REIT performance covered time periods from 1963 to 1979, so an update with newer time periods gives an useful insight into the performance of REITs. Significant alpha or excess returns occurred during ten of their thirteen-years sample period. REITs did significantly over- or underperform compared to the Standard and Poor’s 500 Index during the major part of their sample period.
The performance of REITs from 1984 through 1997 was investigated by Liang and McIntosh (1998). They used a performance style model developed by Sharpe. In their paper,
9
Liang and McIntosh find that all REITs and equity REITs have remarkably stable style attributes during this period. The performance of mortgage REITs fluctuate more over time. All REITs in their sample underperformed before 1995 and outperformed their style portfolios after that time. Alpha or excess returns were also the case during this sample period.
The three papers mentioned above give an overall review of REIT performance during different time periods. They compared the performance of REITs with the performance of different portfolios. In this thesis we want to investigate the impact of corporate governance on this performance. Performances of firms with different corporate governance structures will be compared, but the performance will not be linked to portfolios as is the case in the articles above.
2.2 Corporate governance
Gompers, Ishii and Metrick (2003) found that firms with stronger shareholder rights had higher firm value, higher profits, higher sales growth, lower capital expenditures and has made fewer corporate acquisitions during the nineties of the last century. They used the Carhart (1997) four-factor model and their sample included a new constructed “governance index” for about 1500 large firms during the 1990s, the so-called G-Index. This index uses provisions for takeover as a proxy for the level of shareholder rights. A limitation of this method is that only one aspect of corporate governance is the basis of the G-Index.
The impact of corporate governance on bank performance during the financial crisis is studied by Aebi, Sabato and Schmid (2012). The same Carhart four-factor model as in the previous study was used during their sample period from July 2007 to December 2008. Standard corporate governance variables do insignificantly or negatively affect the banking performance during the crisis. These standard variables include for example CEO ownership, board independence or shareholder rights. In addition, they investigated the impact of “risk governance” characteristics on the bank performance. Banks performed slightly better during the financial crisis when the Chief Risk Officer (CRO) directly reported to the board of directors, compared to banks where the CRO reported to the CEO first.
2.2.1 Board characteristics
The empirical research findings of Zahra and Pearce (1989) show that the board of directors has an impact on corporate financial performance. The board of directors is one of the most influential instruments of a firm’s corporate governance structure. Directors have a big impact on the operations of a firm and have the goal to protect the interests of different shareholders by effectively controlling and managing the firm.
Dalton, Daily, Johnson and Ellstrand (1999) show that the relationship between board size and the corporate financial performance is different for various firm sizes. The relationship is statistically significant for both small and large firm groups in their sample. The impact tends to be larger for smaller firms. This may be due to the fact that small firms have less competition compared to large firms. The analysis used in this paper does not give a rational explanation for this difference.
A paper written by Core, Holhausen and Larcker (1999) find that board and ownership structure explain a significant amount of variation in CEO compensation, after controlling for the standard economic determinants of CEO compensation. Their results
10
suggest that CEOs compensation is higher when governance structures are less effective. CEO compensation is negatively related to the percentage of inside directors in the board, but is positively related to the size of the board and the percentage of outside directors appointed by the CEO. A firm’s operating and stock return performance are both negatively correlated with the predicted component of compensation which arises from the board and ownership variables. Firms with weaker corporate governance structures tend to have bigger agency problems based on their findings. Within these firms with larger agency problems, the CEOs receive a greater amount of compensation. These firms also tend to perform worse, so the agency problems that arise (when the corporate governance structure is weak) have a negative effect on the financial performance, directly as well as indirectly through the CEO compensation.
Mehran (1995) shows that the form of compensation plays an important role in what motivates the managers to increase firm value. The level of compensation has little impact on the manager’s motivation. Firms perform on average better when the percentage of equity held by the managers is high or when their compensation is equity-based. This kind of compensation is often used in firms with more outside directors in the board. Equity-based compensation is less used in corporations with insiders or outside block holders.
The model of Simsek (2007) shows that CEO tenure influences the level of risk taken by the top management team and also relates to the performance of the firm via the entrepreneurial initiatives undertaken by the firm. CEO tenure has a positive effect on the risk taking. These findings suggest that the boards of directors benefit from loyal managers, by an increasing number of successful entrepreneurial initiatives.
Markets also react to the degree in which board directors are independent. The number of independent directors affect the value of companies as shown by Lefort and Urzúa (2008). Only the proportion of outside directors seems to affect the company value, the proportion of professional directors do not have a significant impact. Bhagat and Black (2001) showed in their paper that firms with more independent boards do not perform better in comparison with other firms. Despite of these results, low-profitability firms increased the independence of their board of directors to improve their financial performance.
2.2.2 Ownership concentration
Ownership concentration contains information about the ownership structure of a company. In the case of this thesis, ownership concentration is described as the percentage of the total outstanding shares which are held by investors who own at least 5% of the total number of shares. The impact of ownership concentration on firm performance is described in many papers already. The paper of Demsetz and Villalonga (2001) finds no statistically significant relationship between the ownership structure and the performance of the firm. Their study describes the effect of two aspects of ownership structure, namely the fraction of shares owned by the five largest shareholders and the fraction of shares owned by the management of the company. Demetz and Lehn (1985) have already found that there is no significant relationship between ownership concentration and the accounting profit rates for U.S. corporations in their sample.
Anderson and Reeb (2003) investigated the influence of family ownership on the firm performance of different corporations listed on the S&P 500 index. Their main findings suggest that firms where founding-family members own a firm, perform better compared to
11
the so-called non-family firms. The relation between family holdings and the performance of the firm is nonlinear and when the CEO is one of the family members, performance is even better compared to firms with an outside CEO.
The results of Chen, Cheung, Stouraitis and Wong (2005) on the other hand do not show a positive relationship between family ownership and firm performance measured by the variables return on assets, return on equity and market-to-book ratio. CEO duality is more likely in firms controlled by families and this CEO duality is negatively related to performance. The board composition has little impact on the firm performance for their sample of 412 publicly listed firms in Hong Kong during the sample period from 1995 up and until 1998. In the Czech Republic, more concentrated ownership led to higher firm profitability and labour productivity as shown by Claessens and Djankov (1999). But these findings may be influenced by the fact that mass-privatization took place in the Czech Republic, which excluded managers and outside investors from gaining a high percentage of corporate ownership.
2.3 Relationship corporate governance and REIT performance
The relationship between the REIT market microstructure and stock returns is studied by Wang, Erickson, Gau and Chan (1995). As measurements for financial performance, they use market-adjusted returns, excess returns, Jensen’s alpha and Sharpe ratio. Their sample period contains data from 1970 up and until 1989. REIT stocks tend to have a lower level of institutional investor participation compared to the general stock market. But REIT stocks with higher percentages of institutional investors or REIT stocks followed by more security analysts, perform better in comparison to other REIT stocks. Wang et al. (1995) suspect that REITs underperform compared to normal stocks because of the agency problems that exist in the real estate security market mentioned earlier.
The effect of corporate governance structures on IPO valuations is studied by Hartzell, Kallberg and Liu (2008). Their sample includes 107 IPOs of REITs between 1991 and 1998 and shows that firms with stronger governance structures had a higher IPO valuation and also a better long-term operating performance compared to the peers. To value the IPO they used the Tobin’s Q at the time of the IPO as measurement and two measures of incentive alignment: the percentage of shares owned by insiders and the proportion of the total compensation that is tied to the performance.
A study by Hartzell, Sun and Titman (2006) did not find a strong relation between the corporate governance mechanisms and Tobin’s Q during the nineties of the last century. Tobin’s Q is a measurement of market value of a firm. They found a relationship though between the REITs’ investment expenditures reaction to their opportunities on the one hand and the corporate governance structures of the given REITs on the other hand. When the institutional ownership is greater, the investment choices made by a REIT are more closely tied to Tobin’s Q.
Bauer et al. (2010) found that the firm value of the 220 REITs in their sample is significantly related to the firm-level governance of REITs with low payout ratios only. For other REITs this relationship is not significant. In their study they used the CGQ index provided by the ISS database which is a measurement of corporate governance based on eight categories. They also investigated the effect for more than 5000 other companies as well as a control sample of firms with high corporate real estate ratios. Here they found a significantly positive relationship between the governance index they used and several
12
performance variables. These findings indicate that there is a partial lack of the relationship between corporate governance and the performance in the real estate sector and that this lack might be explained by a REIT effect. Their sample period contains information about REIT from 2003 to 2005.
Campbell, Ghosh, Petrova and Sirmans (2011) examined the effect of corporate governance on mergers and acquisitions by public US REITs between 1997-2006. The bidder returns seem to be higher for REITs with smaller boards and more experienced CEOs, but with shorter tenure. Greater ownership by the CEO and other directors are associated with significantly higher acquirers’ announcement returns.
2.4 Thesis contribution
This master thesis contributes to the already existing literature in the first place by using a sample period including the financial crisis. The study of Aebi et al. (2012) uses a sample period including the financial crisis, but investigated the impact of corporate governance on the performance of banks while this paper investigates the impact on REITs in the United States. Sun, Titman and Twite (2015) found that the share prices of equity REITs were much more volatile during the financial crisis, compared to the underlying commercial real estate prices. The share prices of REITs with higher debt-to-asset ratios and shorter maturity debt decreased more during the 2007-2009 crisis period. The financial distress costs had a permanent impact on REIT values. This study adds to the paper of Sun et al. (2015) by investigating what the role of corporate governance is, in this market situation with decreasing share prices of REITs and higher volatility of these shares during the financial crisis.
A possible reason for the fact that the relationship between corporate governance and REITs has not been investigated before, is the data-availability. Data about corporate governance levels of different REITs is often incomplete and difficult to retrieve. Some variables are not available to everyone and during a sample period it lacks information about specific years during a specific sample period. Another reason for the lack of research could be the fact that, especially since the last decade, more international property flows are allocated through indirect property vehicles. Before that time, indirect property vehicles of which REITs are one of the major examples, did not get that much attention from the researchers. Nowadays directly owned property investments become less attractive, so it gets more interesting to investigate the indirect property vehicles.
The main objective of this paper is to test whether the corporate governance structure has a significant impact on the performance of REITs in the United States around the financial crisis. This objective leads to the first overall hypothesis. Based on the existing literature (for example Bauer et al., 2010), we do not expect a significant impact of the corporate governance on the REIT performance for the whole sample period.
H1: There is no significant impact of the corporate governance structure on the performance of REITs in the United States.
Based on the paper of Johnson, Boone, Breach and Friedman (2000) we construct the second hypothesis to be tested in this paper. In their paper they investigate the impact of corporate governance during the Asian financial crisis. Their findings suggest that the extent of exchange rate depreciation as well as the stock market decline during the end of
13
the last century were influenced by corporate governance. In countries with weak corporate governance, worse economic prospects were associated with larger fall in asset prices because the expropriation by managers was bigger in these countries. Based on these findings we define the next hypothesis:
H2: During the financial crisis, REITs with weaker corporate governance performed worse compared to REITs with stronger corporate governance in the United States.
14
3. Methodology and Data
This chapter will be used to introduce the methodology used in this thesis and to give a first insight in the dataset that is used. The first part gives an overall overview. The second paragraph will explain the model specification and the last part will give information about the dataset, including summary statistics and the shortfalls regarding the dataset used in this research.
3.1 Overview methodology
Three regression models will be used to give answers to the research questions mentioned and the hypotheses stated will be tested by using these models. In the next paragraph, the models will be explained extensively. The sample period used, includes the financial crisis and a period before and after this crisis. It contains the years from 2000 up and until 2015, to compare the pre-crisis period with the crisis-period itself and the after-crisis years. The global financial after-crisis will be indicated by the years 2007 and 2008 as also stated by Erkens et al. (2012). After 2008 the economy recovered slowly, but the impact of the financial crisis will be noticeable also in the first years after the crisis period. Pre-crisis factors could also have an impact on the performance of REITs just before the crisis started in 2007. By taking a large sample period around the global crisis, this research wants to control for these effects and we test the effect of the financial crisis on performance of REITs. So there will be three different time periods in the whole sample period from 2000-2015. The first contains the period before the crisis (2000-2006), the second the period of the global financial crisis (2007-2008) and the last one contains data from the years 2009-2015 after this crisis.
The data contains information about the corporate governance structure of the different REITs in our sample. As described before, several measures for the level of corporate governance are used in the related literature, described before. The CGQ (Corporate Governance Quotient) measure which was used in Bauer et al. (2010) is no longer available. In this study we will construct a new measure for corporate governance which is based on the paper of Aggarwal, Erel, Stulz and Williamson (2007). They use 44 attributes to indicate the level of corporate governance structure, divided in four sub-categories namely: board, audit, anti-takeover and compensation/ownership. Based on the availability of data we will construct a measurement which includes as many of these 44 attributes as possible. These attributes are gathered from the ISS database in Wharton Research Data Services (WRDS) which is available for students of the University of Amsterdam.
Appendix A shows the data of these 44 attributes which is available through WRDS. As you can see in this appendix, some data is available for the board-category and all the information for the anti-takeover-category is available. For both the audit-category and the compensation & ownership-category, no data is available in the ISS-database.
In addition to the board-category and the anti-takeover-category, board size and ownership concentration will be added to the models that will be described in the next section. Based on the related literature which is available for the impact of the board size (Mak & Kusnadi, 2005; Guest, 2009) as well as ownership concentration (Cheng et al., 2005)
15
on the firm performance and especially for REITS, these two variables will be included as well to indicate the corporate governance level of the different REITs in our sample.
3.2 Model specification
This research will use an empirical study to test the relationship between the corporate governance structure and two performance measurements for REITs in the United States. The performance will be measured with two different methods: funds from operations and the stock price fluctuations of the different REITs. The corporate governance measurements will be used as explanatory variables in combination with some control variables to estimate the causes of differences in REIT performance. Three different models will be used to investigate the influence of corporate governance on REIT performance around the financial crisis of 2007/2008.
The first model that will be used, checks if the corporate governance has an impact on the financial operating performance of the different REITs in our sample. This model is mainly based on the paper of Bauer et al. (2010). The dependent variable is the natural logarithm of the funds from operations. Two different explanatory variables (CG) will be used in this model because of the limited availability of data during the three periods (before, during and after the crisis). The first one is a mean variable which is the average corporate governance per REIT for all three different periods mentioned. For each REIT we get three different corporate governance levels, one before the financial crisis (2000-2006), one during the financial crisis (2007-2008) and one corporate governance level after the financial crisis (2009-2015). The effect that these average levels have on the REIT performance will be tested empirically. Because these three periods are not that large, the corporate governance levels are assumed to be relatively constant. These levels normally do not change dramatically during a short period of time. All the variables that are taken into account are represented in Appendix B. Only the board-category and anti-takeover-category will be included in this mean variable. Another possibility is to include the ownership concentration or board-size categories as well. But when you do this, these two categories would have too much impact on the overall corporate governance level used in the regressions in comparison with the board- and anti-takeover-category.
The second corporate governance variable splits the overall corporate governance into four different categories and the impact of these categories will be investigated separately. The four categories used, can be found in appendix C. They include three ratio variables (a board variable without the board size indicator, an anti-takeover variable and an ownership concentration indicator) and one numeric variable (board size separately). Including the board size in the board variable as well, would lead to a double-impact of this size on the REIT performance. Indirectly through the board variable and directly as a separate indicator.
These corporate governance variables could lead to endogeneity problems because of the potential reverse causality. It could be the case that firms with better performance are likely to establish higher corporate governance in their REIT. Reasons to show high corporate governance levels to the market could be to obtain a lower cost of capital in case of financing as mentioned by Klapper and Love (2004). The limited sample period makes it difficult to deal with this endogeneity problem, but control variables are included in our model to mitigate the problem of reverse causality. A dummy variable indicating the time span of the financial crisis as well as an interaction term of the corporate governance with the crisis dummy variable are included in the difference-in-difference model. This
16
interaction term compares the effect of corporate governance on the financial performance during the crisis with the time periods before and after the crisis.
Another control variable is the book-to-market ratio, which is included following the paper of Core, Guay and Rusticus (2006). The firm size is also included as control variable by adding the log of the book value of assets to our model (e.g. Bauer et al., 2010). The control variables about the debt ratio and the dividend are included based on the paper of Bianco, Ghosh and Sirmans (2007). This paper showed that higher debt ratios can affect performance in a negative way, therefore it is included in our model.
The second model to test the effect of the financial crisis on the relationship between corporate governance and REIT performance is the research method of differences-in-differences. Model 2 is a more extensive way to investigate the impact of the crisis compared to the model above, with an additional distinction between two groups: the control group and the treatment group. The control group contains REITs with low corporate governance. The treatment group contains REITs with high corporate governance levels. The distinction between high and low corporate governance levels will be made in two different ways,. The first method splits the sample in two groups, the upper half is the high corporate governance group and the lower half contains the REITs with low corporate governance levels. The second method uses the above 75%-percentile as high corporate governance indicator and the below 25%-percentile for the low corporate governance levels. Both separations are based on the descriptive statistics.
At last, a model will be constructed to measure the effect of corporate governance on the stock prices of different REITs in the United States. A combination of the Fama and French (1993) three-factor model and the Carhart (1997) four-factor model is used to calculate this effect.
Rit is the total return from the REIT stock and Rft is the risk-free rate return, calculated by the US Treasury Bill rate. RMRF is the return of the S&P 500 minus the risk-free rate. SMB and HML are introduced by Fama and French (1993) in which SMB stands for
17
small minus big market capitalization and HML indicates the high minus low book-to-market ratio. The momentum factor UMD is suggested by Carhart (1997) as an improvement to the Fama and French three-factor model. UMD indicates the one year momentum of the stocks to show the difference between winners and losers. The data is available at the site of Kenneth French.3 Appendix D shows the 5 Fama and French variables needed for Model 3. The results of this model will focus on the different coefficients in front of the four factors, but the major point of interest is the alpha-coefficient. This coefficient indicates whether a specific REIT portfolio outperforms the broader market or not.
3.3 The Data
Data about the performance of the different REITs will be divided in two main subgroups: operating performance and stock price performance. Stock price data is available at the Center for Research in Security Prices (CRSP) which could be accessed via WRDS. To measure the value of a REIT, standard measures like growth and price-to-earnings ratios do not apply. Net income could be a good measure, but when you think about depreciation expenses you can conclude that this is not a proper way to estimate the value. Net income includes depreciation expenses, but real estate properties rarely lose value and sometimes even appreciate. In most papers, funds from operations (FFO) is used to estimate the operating performance of a REIT. NAREIT (National Association of Real Estate Investment Trusts) adopted this definition of FFO in 1991 and their current definition follows:
FUNDS FROM OPERATIONS means net income (computed in accordance with generally accepted accounting principles), excluding gains (or losses) from sales of property, plus depreciation and amortization, and after adjustments for unconsolidated partnerships and joint ventures. Adjustments for unconsolidated partnerships and joint ventures will be calculated to reflect funds from operations on the same basis (NAREIT, 2002, p. 2).
The performance and control variables used in our model are obtained from the Capital IQ database. This database contains financial annual reports for all REITs in our sample. Because these reports can only be downloaded separately, it costs some time to combine them in one dataset containing information about the financial performance. Afterwards, this database was combined with the stock performance database which was retrieved from the CRSP Ziman REIT database. In the existing literature there is often a distinction between internal and external advised REITs (Bauer et al., 2010). Our sample could contain REITs from both groups, but we do not make a distinction between internal and external advised REITs in our regression models. The internal or external board members however, are included in our board-category to measure the level of corporate governance.
The sample period used in this study contains data from 2000 up and until 2015. The period from 2000 up and until 2006 is the pre-crisis period, the crisis period is during the years 2007 and 2008 and from 2009 up and until 2015 is named the after-crisis period.
18
3.3.1 Summary Statistics
The summary statistics for the dependent, independent and control variables are described in Table 1. The total sample contains information about 81 REITs and the sample period contains information about 16 years (2000-2015). The funds from operations are not available for every REIT during the whole sample period and therefore the information of REITs is not complete for all 16 years.
Because of the limited availability of corporate governance indicators during the sample period, the number of observations is lower for these variables compared to the dependent variable and the control variables. The board-category variables, anti-takeover variable and ownership concentration variable are ratio variables which can take a value between 0 and 1. Board size is a numeric variable.
The performance variable FFO shows a lot of variety during the sample period. The funds from operations is on average $274.21 million, but the standard deviation is $360.94 million. The book-to-market ratio indicates the relationship between the book value and market value of the REITs. A book-to-market ratio below 1 is an indication for an undervalued stock, above 1 indicates that the stock is overvalued. The average in our sample is 0.61, so on average the stocks of the REITs in our sample are undervalued. Especially during the crisis the market value of the stocks drops below the book value, which will be shown in table 2 later on.
A lot of variety can also be observed when looking to the size of the different REITs in our sample, measured by the book value of assets. The mean of this book value is 4.9 billion dollars, but the standard deviation is about 5.87 billion dollars. The average dividend yield is 0.45 and the average payout ratio for a given REIT during a given year is 85.7%. This payout ratio is calculated on a per share basis, namely the dividends paid per share outstanding divided by the earnings indicated by the FFO per share.
Table 1
Sample summary statistics for the total sample period 2000-2015.
The table shows summary statistics for key variables for a sample of 81 REITs for the years 2000 up and until 2015. The data are taken from the CRSP-Ziman REIT Database and the financial performance variables are retrieved from the Capital IQ Database. Funds from operations and assets are measured in millions of dollars. The board size is measured in numbers of members for each board of directors. The book-to-market ratio and the debt-to-equity ratio are ratios and the other variables are percentages.
Variable Obs Mean Std. Dev. Min. Max.
FFO 1104 274,21 360,94 -65,25 3.571,24
Book-to-Market Ratio 1094 0,6099 0,7014 0 11,5932
Book Value of Assets 1104 4.904,93 5.870,77 35,60 33.324,57
Dividend Yield 1092 0,4532 1,1217 0 23,2356
Payout Ratio 1055 0,8574 3,2078 0 60,3294
Debt-to-Equity Ratio 1104 1,4372 1,5639 0 23,2356
Board incl. Board Size 871 0,7122 0,3771 0 1
Board excl. Board Size 872 0,7067 0,3784 0 1
Anti-Takeover variable 852 0,3361 0,2046 0 0,8333
# of Board Members 738 8,5718 1,9962 1 14
19
The size of the board of directors varies from 1 to 14 members. The average board of directors consists of 8.6 members, with a standard deviation of 2. The mean of the board-categories are respectively 0.71 for both including and excluding the board size. It does not make a large difference whether we include the board size in this corporate governance variable or not. The average anti-takeover variable has a value of 0.34 and the standard error is 0.20. On average, the REITs have an ownership concentration of 6.34%, with a standard deviation of 16.4%.
Table 2 describes the means of all variables included in our models, per different time period in our total sample period. The period before the crisis (2000-2006), during the crisis (2007-2008) and after the financial crisis (2009-2015) are separated. The funds are on average higher after the crisis, compared to the periods during and before the crisis. It increases from 333 million dollars to about 396 million dollars after the crisis. The book to market ratio is 0.50 before the crisis, 0.49 after the crisis but tops at 1.06 on average during the crisis. So this means that before and after the crisis, the stocks of the REITs in our sample are undervalued and overvalued during the crisis period. This is mainly due to the drop in market prices and not because of an increase in book values. During the crisis, the market prices of the shares dropped significantly.
Both the dividend yield and the payout ratio has its maximum average values during the crisis. During this period, the dividend yield is 0.73 which is much higher than before (0.30) or after the crisis (0.46). The dividends per share are higher during the crisis compared to the funds from operations per share (ratio 1.17). Before the crisis (0.54) and after the crisis (0.96) the dividends per share are lower than the FFO per share.
Table 2
Sample summary statistics per period.
The table shows the different means for key variables for a sample of 81 REITs during the sample period 2000-2015. The variables are split into three different periods. The period before the crisis (2000-2006), during the crisis (2007-2008) and after the crisis (2009-2015). The data are taken from the CRSP-Ziman REIT Database and the financial performance variables are retrieved from the Capital IQ Database. Funds from operations and assets are measured in millions of dollars. The board size is measured in numbers of members for each board of directors. The book-to-market ratio and the debt-to-equity ratio are ratios and the other variables are percentages.
PERIOD
Pre-Crisis Mean During Crisis Mean After Crisis Mean
FFO 333,4187 FFO 304,5625 FFO 396,4681
BM-Ratio 0,4983 BM-Ratio 1,0619 BM-Ratio 0,4859
BV Assets 5720,2750 BV Assets 6063,8070 BV Assets 7070,5680
Div Yield 0,2977 Div Yield 0,7313 Div Yield 0,4650
Payout Ratio 0,5389 Payout Ratio 1,1659 Payout Ratio 0,9586
D/E Ratio 1,6633 D/E Ratio 1,7607 D/E Ratio 1,3578
Board Incl. 0,6180 Board Incl. 0,8606 Board Incl. 0,8993
Board Exc. 0,6037 Board Exc. 0,8614 Board Exc. 0,8955
Anti-Takeover 0,4482 Anti-Takeover 0,1039 Anti-Takeover 0,2976
Board size 9,0388 Board size 8,6122 Board size 8,5340
Own. Concentr. 0,1385 Own. Concentr. 0,0643 Own. Concentr. 0,0478
The board variable (including as well as excluding the board size indicator) increases during the three time periods. After the crisis the board indicators of corporate governance increased compared to the period before the crisis. The anti-takeover variable shows a
20
different pattern. It has the highest value before the crisis (0.45), decreases to 0.10 during the crisis and climbs again after the crisis to an average level of 0.30. The board size decreases slightly from on average 9 members before the crisis to 8.6 during and 8.5 after the crisis. The ownership concentration decreases at a higher pace. Before the crisis this variable has an average level of 13.85%, during the crisis 6.43% and after the crisis even 4.78%.
Before we run the different regressions, we first take a look at the correlations between the different variables in our models (Table 3). Correlations give information about whether the relationship between two variables is positive or negative and about the strength of these relationships. The table shows that all corporate governance indicators are positively related to the funds from operations, but only the correlations of the board indicator and the board size are significant. The size of the firm is positive and significant correlated with the funds from operations and also with the two board indicators and the boards size. Larger REITs thus tend to have a higher corporate governance level based on the board indicator and also have on average a greater board of directors.
The book-to-market ratio is the only variable which is negatively and significantly correlated with the dependent variable funds from operations. A higher book-to-market ratio could be an indication for an undervalued stock. Both the dividend yield and the payout ratio have a positive correlation with the funds from operation, but only the correlation of the dividend yield is significant on a 90% confidence interval. In the results section of this thesis we will talk more about the different relationships between the variables based on our regression models.
Table 3
The correlation matrix.
The table shows the different correlations between key variables for a sample of 81 REITs during the sample period 2000-2015. The correlation indicates the relationship between two variables and whether this relationship is positive or negative. The * is an indication of significance on a 90% confidence interval.
1 2 3 4 5 6 7 8 9 10 11 1 FFO 1 2 BTM -0,1057* 1 3 Assets 0,8520* -0,0046 1 4 DivYield 0,0679* 0,0976* 0,2490* 1 5 PayoutRatio 0,0410 0,0207 0,0214 -0,0284 1 6 DE Ratio 0,1425* -0,0074 0,2097* 0,4886* -0,0134 1 7 Board Incl. 0,1549* 0,0305 0,1753* -0,0032 0,0452 -0,1585* 1 8 Board Excl. 0,1551* 0,0035 0,1738* -0,0053 0,0440 -0,1630* 0,9963* 1 9 Anti-Takeover 0,0210 -0,1636 0,0007 -0,0222 -0,0020 0,0161 -0,3878* 0,3891* 1 10 Boardsize 0,2789* -0,0316 0,3147* 0,0808* -0,0240 0,1134* 0,1120* 0,0501 0,0673* 1 11 Concentration 0,0444 -0,1003* -0,0086 -0,0276 -0,0389 0,2060* -0,2564* -0,2626* 0,1005* 0,0757* 1 3.3.2 Data Shortfalls
The dataset of this master thesis has some pitfalls that has to be taken into consideration when looking at the regression results and the conclusions later on. The first and major problem is the data-availability considering the corporate governance levels of the different REITs during our sample period. ISS (former RiskMetrics) has a database which
21
contains information about the corporate governance indicators for all kind of corporations, including REITs. This database is split into two different parts: the governance part and the directors part. The governance part contains information about the structure of the company itself, for example whether it deals about a dual class company or whether the firm has a poison pill or not. The directors part gives information about the board of directors and the specific information about the individual members of this board for each firm in the database.
Two major problems exist when you use this database. At first, it does not contain information about the different indicators for every REIT during every year in our sample period. So there are some years that do not contain any information about the corporate governance indicators at all. For the first model, the corporate governance is an average of the different corporate governance indicators. In the second model, 4 different indicators of the corporate governance structure are used separately in the regression model. When a year lacks information about the corporate governance structure and there is information about the year before, the assumption is made that this structure has not changed since last year. So the missing year gets the same corporate governance value as the previous one, which is based on this assumption and not on actual data available in the corporate governance database ISS.
Another problem when using this database is the split-up of the governance database. The database has a part regarding directors and one part regarding the governance itself, but both parts contain a legacy section and a “normal” section. The legacy section describes the corporate governance up and until 2006 and the normal section gives information from 2007. The problem is that both parts do not contain exactly the same variables. Some are only available before 2007, others only after this period. This situation makes it difficult to compare the REITs around the financial crisis. It will be necessary to counter these data-problems with some robustness checks. These robustness checks will be made by means of the different model specifications mentioned. Because we use three different models and also different corporate governance indicators, this paper tries to deal with the data-availability problems. When using different models and indicators, it makes the results and conclusions based on the models more reliable but also more difficult to compare. Therefore the study focuses on the statistically significance of the coefficients as well as the sign of it. It is hard to compare the strength of the relationships when using different models and corporate governance measurements. The difference-in-difference model uses a crisis-dummy variable and an interaction term variable to deal with this.
To make our results and conclusions less vulnerable for unavailable data, we will also examine the empirical model by using an average corporate governance level for the years before, during and after the financial crisis. When you take the average values, the missing years do not count anymore and therefore the reliability of our results will increase. More about this additional regression will be explained in the next section where we will describe the results of the different models used.
22
4. Results
This section will evaluate the results and give an analysis of the results from the three different models which are described in the previous methodology section. This chapter is divided into three different sections, where each section will contain an analysis about one of the three models.
4.1 Results model 1
4.1.1 One corporate governance indicator
Results from the first model, with one independent corporate governance variable, are displayed in Table 4 below. The table is divided in 4 different sets of columns, each indicating another time frame in the sample period: the period before, during and after the crisis and the last set contains the total sample period. The odd columns represent the regressions with only the corporate governance variable as independent variable, while the even columns also contain the five control variables explained in the methodology section.
Table 4
The natural logarithm of the funds from operations, one corporate governance measure and control variables.
The table shows results from regressions of the natural logarithm of the funds from operations for REITs from January 2000 to December 2015 on the corporate governance levels and some control variables. The dependent variable is the natural logarithm of the funds from operations. The most important independent variable is the corporate governance variable, which is a variable that includes board-indicators as well as anti-takeover-indicators for the level of corporate governance. The book-to-market ratio is the book value of the shares divided by the market value of the shares outstanding. For the variable with the book value of assets, a log transformation is applied. The payout ratio indicates the proportion of the FFO per share which is paid out as dividends and the dividend yield is the dividends per share divided by the price per share.
Pre-Crisis During the crisis After the Crisis Total sample period
VARIABLES lnFFO lnFFO lnFFO lnFFO lnFFO lnFFO lnFFO lnFFO
(1) (2) (3) (4) (5) (6) (7) (8) Corp. Governance -0.219 -0.116 0.0539 -0.450 1.686*** 0.0761 0.805*** 0.0266 (0.311) (0.141) (0.607) (0.276) (0.315) (0.129) (0.197) (0.0862) BTM - Ratio -0.477*** -0.0462* -0.661*** -0.119*** (0.113) (0.0276) (0.0669) (0.0225) Log (BV Assets) 1.878*** 1.934*** 2.059*** 2.018*** (0.0785) (0.0867) (0.0392) (0.0340) Dividend Yield -0.0264 0.0112 -0.0534*** -0.0389*** (0.0280) (0.0295) (0.0166) (0.0134) Payout Ratio -0.386*** -0.00375 0.00621 0.00277 (0.111) (0.00786) (0.00516) (0.00462) D/E - Ratio -0.0435** -0.0398* -0.0140 -0.00335 (0.0204) (0.0224) (0.0133) (0.0101) Constant 19.18*** 1.772** 18.99*** 0.907 18.29*** -0.234 18.76*** -0.0841 (0.151) (0.773) (0.292) (0.815) (0.188) (0.366) (0.109) (0.320) Observations 174 173 117 112 483 453 774 738 R-squared 0.003 0.813 0.000 0.833 0.056 0.874 0.021 0.835
23
Due to the lack of data on the corporate governance variable before the crisis, you can see that columns (1) and (2) contains less observations compared to the period after the crisis (columns 5 and 6). Columns (3) and (4) have the least number of observations, but this period only contains information about the years 2007 and 2008. A small decrease is visible when including the control variables, so not every year has information available for every individual control variable as well.
The mean variable of interest in this first regression model is the corporate governance indicator, which is a ratio variable that can vary from 0 to 1. In fact it is a percentage variable which describes the percentage of corporate governance requirements met by a specific REIT during a specific year in our sample period. These indicators can be found in Appendix B and when a specific requirement is met it gets a value of 1, otherwise a 0. When more requirements are met, this results in a higher corporate governance variable which indicates a higher corporate governance level for a specific REIT. The negative sign in column (1) indicates a negative relationship between corporate governance and the natural logarithm of the funds from operations but this relationship is not significant. Columns (3), (5) and (7) give a different result, namely a positive relationship between the corporate governance level and the funds from operations. Only during the period after the crisis and during the total sample period, this positive relationship is significant. It should be noted that each of these regressions does not include control variables.
When we add these control variables in our regression model, we conclude that there is no significant relationship between the corporate governance level and the funds from operations. Before and after the crisis and for the total sample period the sign of the relationship is still the same. Only during the crisis the sign changes from positive to negative when adding the control variables, but because none of the relationships is significant, there is nothing we can conclude about this.
At last, we look at the control variables and their relationships with the natural logarithm of the funds from operations. The relationship between the book-to-market ratio is significantly negative for each of the four periods. When the book-to-market ratio is higher, the funds from operations decrease based on this model. The negative sign has the highest value after the crisis, so the impact of the book-to-market ratio is higher after the crisis than before or during the crisis. This result is in line with the paper of Fama and French (1995). Their findings suggest that high book-to-market ratios signal persistent poor performance in terms of earnings. Between the book value of assets and the performance variable there is a positive and significant relationship for all four periods. When the book value of the asset increases, the funds from operations also tend to increase. The same paper of Fama and French (1995) shows a similar pattern. Small firms in their sample tend to have lower earnings compared to big firms, especially during the small-stock depression of the 1980s.
Before the crisis, the sign for the dividend yield is negative and this sign is positive for the crisis-period itself but both are insignificant. After the crisis and for the whole sample period, the sign is significantly negative. When the dividend yield increases, the performance of the REIT decreases during those two periods. The relationship between the payout ratio and the FFO is only significant before the crisis and the sign is negative. The debt-to-equity ratio has a negative impact on the funds from operations, but this impact is only significant before and during the crisis.
24
4.1.2 Four corporate governance indicators
As a robustness check for the test of the relationship between the corporate governance level and the funds from operations, we also use four different corporate governance variables as independent variables in our regression model. The results of this regression is shown in Table 5 on the next page. Also in this table, the odd columns show regressions without and the even columns with control variables. The four different independent variables which are used to indicate the corporate governance level are: board indicator excluding board size, the natural logarithm of the board size, the anti-takeover indicator and the ownership concentration.
Table 5
The natural logarithm of the funds from operations, four corporate governance indicators and control variables.
The table shows results from regressions of the natural logarithm of the funds from operations for REITs from January 2000 to December 2015 on the corporate governance levels and some control variables. The dependent variable is the natural logarithm of the funds from operations. The most important independent variables are four corporate governance variables. The first indicates the corporate governance of the board excluding the board size. This board size is the second independent variable and for this variable a log transformation is applied. The third is a variable that summarizes the anti-takeover indicators and the fourth variable indicates the ownership concentration. This is the percentage of the total outstanding shares which are held by shareholder that own more than 5% of the shares outstanding. The book-to-market ratio is the book value of the shares divided by the market value of the shares outstanding. For the variable with the book value of assets, a log transformation is applied. The payout ratio indicates the proportion of the FFO per share which is paid out as dividends and the dividend yield is the dividends per share divided by the price per share.
Pre-Crisis During the crisis After the Crisis Total sample period
VARIABLES lnFFO lnFFO lnFFO lnFFO lnFFO lnFFO lnFFO lnFFO
(1) (2) (3) (4) (5) (6) (7) (8) Board excl. -0.0274 -0.228 -0.329 -0.320 1.405*** -0.0812 0.466*** -0.0709 (0.206) (0.160) (0.425) (0.208) (0.361) (0.147) (0.161) (0.0759) ln Board Size 1.963*** 0.123 1.843*** -0.0447 1.355*** -0.188** 1.573*** -0.105 (0.274) (0.225) (0.330) (0.194) (0.188) (0.0829) (0.143) (0.0752) Anti-Takeover -0.952** -0.132 -0.228 -0.526** 1.108*** 0.168 0.577*** 0.0795 (0.398) (0.267) (0.498) (0.258) (0.261) (0.109) (0.181) (0.0861) Own. Conc. -0.793*** -0.0509 -0.484 -0.183 -0.806** -0.238* -0.878*** -0.0841 (0.293) (0.208) (0.499) (0.254) (0.327) (0.137) (0.210) (0.102) BTM-Ratio -0.434*** -0.0588* -0.721*** -0.126*** (0.150) (0.0297) (0.0705) (0.0248) Log(BV Assets) 1.838*** 1.905*** 2.073*** 2.014*** (0.151) (0.114) (0.0442) (0.0420) Dividend Yield -0.0508 -0.0121 -0.0537*** -0.0445*** (0.0766) (0.0439) (0.0175) (0.0163) Payout Ratio 0.0548 -0.00281 0.00528 0.00308 (0.219) (0.00804) (0.00524) (0.00481) D/E-Ratio -0.0270 -0.00185 -0.00617 0.0111 (0.0314) (0.0497) (0.0144) (0.0129) Constant 15.44*** 1.769 15.50*** 1.405 14.85*** 0.129 15.38*** 0.226 (0.647) (1.279) (0.728) (0.943) (0.475) (0.384) (0.332) (0.355) Observations 105 105 103 98 448 422 656 625 R-squared 0.433 0.795 0.247 0.841 0.192 0.878 0.199 0.836
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Before and during the crisis, the relationship between the board exclusive-indicator and the FFO is negative and insignificant, but after the crisis and for the total sample period this relationship is positive in the regressions without control variables. The impact of the board size on the funds from operations is positive and significant in all these four
