Management Ownership and Capital Structure Decisions:
The Impact of the Financial Crisis
SANDER C. BORGERS
ABSTRACT
This paper analyzes the impact of management ownership on the capital structure of the firm, with results generally suggesting that CEO equity ownership is negatively related to the leverage ratios of firms. This study extends recent literature on capital structure theory in three ways. First, it extends the “core model of leverage”, described in Frank and Goyal‘s (2009) paper, by adding an additional, significant management ownership factor into the model. Second, this study investigates whether a non-linear relationship exists between leverage ratios and management ownership. Third, it examines the interaction between the financial crisis and the management ownership factor. In a panel analysis, the empirical findings are not consistent with the hypothesis that the risk aversion of CEOs affects the capital structure of firms.
Name: Sander Carl Borgers Student id: 10534903
E-mail address: sander.borgers@student.uva.nl Supervisor: Torsten Jochem, Ph.D.
E-mail address supervisor: t.jochem@uva.nl
Bachelor’s Thesis
University of Amsterdam Economics and Business Finance and Organizations
Statement of Originality
This document is written by Sander Carl Borgers 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.
Introduction
In recent years, search for an optimal capital structure has led to a large number of theoretical and empirical studies, examining the factors that influence the leverage ratio of a firm. The optimum is driven by several prominent theories of capital structure deduced from the perfect Miller-Modigliani world. For instance, the trade-off theory states there is a trade-off between the tax shield of debt and the costs of financial distress when the firm faces bankruptcy. Firms choose levels of debt in order to balance this trade-off. Jensen and Meckling (1976) extent the trade-off theory by adding agency costs to the model. Furthermore, Myers (1984) explains the pecking order theory, which describes a fixed ranking of three sources of funds based on signaling and the asymmetric information problem between insiders and investors. In
addition, Graham and Harvey (2001) find evidence for the market timing theory. This theory argues that the capital structure depends on the conditions in both debt and equity markets.
Frank and Goyal (2009) state that no unified model exists in which all the main capital structure theories come together to explain the capital structure of firms, so they provided a model. This model mixes the existing theories by including several core factors that are reliably important, described as the “core model of leverage”. However, this study finds evidence that the model excludes one of the real important determinants of capital structure because the model leaves out the management ownership factor. In this study management ownership is described as the proportion of the firm’s equity owned by the CEO. A large number of previous studies examined the influence of management ownership on capital structure decisions of a firm. In addition, the literature shows mixed results about the impact of management ownership on firm leverage.
Jensen and Meckling (1976) introduced this subject by arguing that the leverage ratio of firms is negatively related to management ownership through their agency theory. Because of the separation of ownership and control, external shareholders might have to implement disciplining mechanisms, which could prevent CEOs from non-profit-maximizing behavior. According to Jensen and Meckling (1976), high debt levels constrain CEOs to pursue personal goals, whereas management ownership acts to align the preferences of the CEO and external shareholders. They explain that if management ownership increases, less debt is needed in the firm so there exists a negative relationship between these factors. Furthermore, the ownership of equity will have an increasing impact on the CEO’s interest in the firm, according to Jensen and Meckling (1976), so their risk aversion will become more influential.
Since default risk increases when the firm is more levered, a higher proportion of management ownership decreases the leverage ratio due to the CEO’s risk aversion. However, several other studies find a positive impact of management ownership on leverage due to managerial entrenchment (e.g.; Berger et al., 1997; Stulz, 1988). These studies explain this positive relation by referring to CEOs who increase leverage to maximize firm’s value, inflate their voting power, or block takeovers.
The purpose of this paper is to extend recent literature on capital structure and management ownership in three ways. First, it extends the “core model of leverage”, described in Frank and Goyal‘s (2009) paper, by adding an additional, significant management ownership factor into the model. The sample used for this research can be seen as a continuation on the sample of Frank and Goyal (2009), consisted of US nonfinancial firms. Second, this study investigates whether the non-linear relationship between leverage and management ownership, found by Bruslerie and Latrous (2012) for French firms, is also applicable to US nonfinancial firms. Third, since the literature finds contradictory evidence about the influence of management ownership on leverage, this study examines the interaction between management ownership and the financial crisis. The empirical results generate new insights into the management ownership problem, by assuming that default risk increased during the financial crisis. The financial crisis should have a negative impact on leverage among high management ownership firms if the dominant channel in the theory of Jensen and Meckling (1976) was managerial risk aversion rather than debt as a disciplining device.
The empirical results suggest that management ownership is negatively related to the leverage ratios of firms. Some of our results are therefore consistent with the theory of Jensen and Meckling (1976), which explains a negative relationship due to the risk aversion of CEOs and the decreased agency costs between CEOs and external shareholders. However, this study also provides evidence that is not in line with the risk aversion channel of Jensen and Meckling (1976). In particular, the management ownership factor is not affected by the financial crisis, what could be seen as new evidence against the risk aversion channel. So this study invalidates one channel, but does not invalidate the whole theory of Jensen and Meckling (1976). Furthermore, unlike Bruslerie and Latrous’ (2012) work, the empirical findings of this study do not find a U-shaped relationship between leverage and management ownership for US nonfinancial firms.
The remainder of this paper is organized as follows. Section I elaborates on the relationship between capital structure, management ownership, and several other control
factors. Section II describes the sample, variables, and method used to impute missing values. Next, section III presents the analysis of the impact of management ownership on capital structure, including the empirical method used and empirical results. Section IV presents similar information for the financial crisis analysis. Section V describes the limitations of this research. The conclusions are presented in section VI.
I.
The Relationship between Capital Structure, Management
Ownership, and Other Factors – Prior Research and Hypotheses
A. Capital Structure and Management Ownership
The relationship between management ownership and leverage was first introduced by Jensen and Meckling (1976) and followed by several other empirical studies. Management ownership exists when the CEO owns a part of equity in the firm. In addition, literature often finds a significant relationship between firms with management ownership and their debt levels beside of the other factors (Friend and Lang, 1988; Firth, 1995; Berger et al., 1997; Stulz, 1988). However, literature shows that management ownership affects a firm’s debt level, but there is still not a certain theory in what way the debt level is affected.
Jensen and Meckling (1976) argue that ownership by CEOs acts to align the preferences of the CEO and external shareholders. Both parties prefer to maximize shareholder value. Debt constraints are viewed as a mechanism to align the manager, so there will be less need for debt when management ownership is also viewed as an agency instrument. Furthermore, the ownership of equity will have an increasing impact on the CEO’s interest in the firm according to Jensen and Meckling (1976). Since CEOs are on average risk averse, they would be less likely to take excessive risk. For instance, managers will be more opposed to the default risk a firm faces through debt financing. Given these explanations, Jensen and Meckling (1976) find a negative relationship between leverage and management ownership. In addition, Friend and Lang (1988) and Firth (1995) also demonstrate the existence of a negative relationship between leverage and management ownership due to the risk aversion of managers. CEOs adjust the capital structure more to their own preferences when management ownership increases, because they have greater discretion and control over the debt policy of the firm. Since Friend and Lang (1988) and Firth (1995) follow the theory stated by Jensen and Meckling (1976), CEOs will lower the leverage ratio if their ownership becomes higher.
Conversely, Berger et al. (1997) find evidence that management ownership is positively related to the capital structure levels because of managerial entrenchment. They argue that the theory described by Jensen and Meckling (1976) focus only on the reduced agency problems between managers and shareholders, but they ignore the agency problem of debt itself. The regressions displayed in in the study of Berger et al. (1997) show a positive and generally significant relationship between firm leverage and management ownership. They explain these findings that managers, whose financial incentives are more closely related to shareholder value, will hold higher debt levels to benefit from a higher tax shield. As a result they maximize the value of the firm and therefore the value of their own shares, whereas there will be more default risk for the firm (Berger et al., 1997). These opposite findings of Jensen and Meckling (1976) and Berger et al. (1997) might be resulted from the difference in the empirical model or sample used to estimate the relationship.
Stulz (1988), however, suggest other entrenchment motives as explanations for the positive relationship. He states that management ownership enhances managers to set the debt level to their own interest, which might not maximize shareholder value. Managers may deviate from choosing the optimal debt level as a result of the agency costs of managerial discretion. Managers will increase the debt level to inflate the voting power of their equity shares and reduce the possibility of takeover attempts, according to Stulz (1988). For example, CEOs increase their equity stake in the firm by changing the capital structure of a firm through share repurchases financed with debt. Therefore managers are able to inflate their voting rights in the company and thereby it becomes easier for management to become isolated from shareholder pressure and become entrenched.. Moreover, Palepu (1986) provides evidence that highly levered firms are less likely to be acquired than unlevered firms. Stulz (1988) explains this outcome by arguing that a high debt level decreases the acquirer’s gain from control. The target’s debt may, for instance, decreases the acquirer’s control over the target’s assets due to the debt covenants. Therefore the entrenched manager could increase leverage so the firm will be less attractive and face lower takeover risk.
In conclusion, the literature shows mixed results in the effect of management ownership on capital structure. Some studies provide evidence for a negative relationship due to the risk aversion of CEOs, whereas other research states that management ownership is positively related because of the occurrence of managerial entrenchment. Therefore, the hypothesis postulated in this paper is that there is a significant relationship between management ownership and leverage, but it is not clear in what way the debt ratio is affected.
H1: There exists a significant relationship between leverage and management ownership.
B. A Non-Linear Relationship between Leverage and Management Ownership
Recent research of Bruslerie and Latrous (2012) finds evidence for a U-shaped relationship between leverage and management ownership for nonfinancial French firms. In addition, their results suggest that leverage is negatively related to management ownership when they hold more than 40% of the firm’s equity. Conversely, this result suggests that leverage is positively linked to the management ownership when they generally hold less than 40% of the firm’s common shares outstanding. Bruslerie and Latrous (2012) explain this U-shaped relationship by referring simultaneously to the entrenchment effect and to the incentive effect. At low levels of management ownership, CEOs become entrenched and increase leverage to inflate their power and protect themselves against takeovers. However, at high levels of management ownership, CEOs interests become more aligned with those of outside shareholders as their risk aversion replaces the entrenchment motives so they are less likely to take excessive risk. CEOs would prefer, for instance, to use less leverage to limit their default risk. According to Bruslerie and Latrous’ (2012) study, these two competing effects suggest a non-linear relationship between the level of leverage and management ownership. Formally, this study tests if the U-shaped relationship between leverage and management ownership is also applicable to US nonfinancial firms.
H2: There exists a significant U-shaped relationship between leverage and management ownership.
C. The Impact of the Financial Crisis on the Management Ownership Factor
The preceding literature finds contradictory evidence as to how leverage is related to management ownership. This study investigates whether the impact of management ownership is affected by the recent financial crisis so the empirical research will generate new insights into the management ownership problem. Namely, under the assumption that during the financial crisis in 2008 to 2009 default risk increased, CEOs with lots of equity may change the leverage level. Default risk could have increased because the leverage ratios of firms were too high, and could have increased beyond the CEO’s preferred level. To illustrate, if the ownership of equity will have an increasing impact on the CEO’s interest in the firm, as stated by Jensen and Meckling (1976), CEOs in firms with a high fraction of
management ownership would decrease the leverage ratio even more during the financial crisis due to the CEO’s risk aversion. For this reason, I expect leverage to be negatively related to the management ownership during the financial crisis.
Moreover, realizing that the next crisis could happen again, a CEO with a high ownership of equity may want to maintain a lower leverage level after the crisis, relative to the pre-crisis level, to avoid a repeat of such a situation. Hence, it would generate new evidence for the theory stated by Jensen and Meckling (1976), if leverage is negatively affected by management ownership during and after the crisis.
H3: Leverage is negatively affected by management ownership during and after the crisis.
D. Predictions By Other Capital Structure Theories
This section contains predictions of the relationships between leverage and the included control variables that are derived from Frank and Goyal’s (2009) study. Frank and Goyal (2009) describe these factors as the “core model of leverage” that account for more than 27% of the variation in leverage. These factors are added to the regression model of this study to control for other influences than management ownership on leverage. The predictions for the control variables are mainly based on two competing financial models of financing decisions, the trade-off theory of and the pecking order theory. While many of these theories seem uncontroversial, the results may deviate from the predictions stated by theory.
1. Leverage and Profitability
Recent studies suggest that profitable firms tend to have lower leverage levels despite their low likelihood of financial distress and the value that could be generated from a tax shield, known as a contradiction to the trade-off theory (Shyam-Sunder and Myers, 1999; Elsas, Flannery, and Garfinkel, 2014). In addition, Fama and French (2002) also find a negative relationship between profitability and leverage, and confirm therefore the pecking order theory. Namely, the pecking order theory argues that firms prefer internal finance to external funds. Since profitable firms generate more internal financial sources, these firms will become less levered. Therefore I expect firms that are more profitable tend to have a lower leverage ratio.
2. Leverage and Firm Size
Several papers demonstrate a positive relationship between large firms, in terms of assets, and leverage (Fama and French, 2002; Frank and Goyal, 2009; Danis, Rattl, and Whited, 2013). Smaller firms face higher default risk, according to Danis et al. (2013), because these firms are likely to be less diversified. Moreover, Frank and Goyal (2009) state that larger firms face lower debt-related agency costs because of the better reputations they have in debt markets. Hence, larger firms tend to have a higher leverage ratio, according to the predictions off the trade-off theory.
3. Leverage and Growth
Conversely to large firms, growing firms face a higher probability on financial distress costs so the trade-off theory would predict a negative relationship between leverage and growth. Consistent with the trade-off prediction, Barclay et al. (2003) find that leverage is negatively related to growing firms. Billett et al. (2007) explain this relationship by referring to the increasing debt covenants for growing firms to mitigate the agency costs of debt. These debt covenants make it less attractive for growing firms to finance its activities with debt.
In addition, the capital expenditure-to-book assets ratio is used as a proxy for the growth opportunities to exclude the impact of the financial crisis during the period 2008 to 2009. Frank and Goyal (2009), however, recommend the market-to-book asset ratio as a proxy for the growth opportunities, but this ratio is more affected by the financial crisis as market values are most influenced.
4. Leverage and Industry Growth
The median industry growth factor is included in the model to control for industry-specific variation. Firms in an industry face the same forces, like for example regulation and the nature of competition, that affect their financing decisions. Therefore, Frank and Goyal (2009) also include this factor to mitigate the problem of omitted variables. By including the median industry growth factor, a negative relationship with leverage is predicted according to the trade-off theory.
5. Leverage and Tangibility
Literature often finds a positive relation between tangibility and leverage (e.g., Titman and Wessels, 1988; Huang and Ritter, 2009). To illustrate, firms with a high proportion of tangible assets, such as machines and plants, are likely to face relatively lower expected
distress costs because these assets can be collateralized. Conversely, firms with unique, intangible assets, like high R&D expenses and the value of goodwill, could have high bankruptcy costs (Titman and Wessels, 1988). In addition, firms with a high proportion of tangible assets face less debt-related agency costs. Frank and Goyal (2009) state it is more difficult for shareholders to substitute high-risk assets for low-risk assets. Therefore the information asymmetry between debtholder and firm decrease as well as the debt-related agency costs. Since the expected distress costs and debt-related agency problems are lower for firms with a high proportion of tangible assets, the trade-off theory predicts that leverage is positively related to tangibility.
6. Leverage and Expected Inflation
Taggart (1985) argues that the real value of the tax shield, obtained from leverage, increases when inflation is expected to be high in the future. Since expected inflation increases tax rates to lower the inflation, the real value of the tax shield will increase. Therefore, the trade-off theory predicts a positive relation between leverage and expected inflation. However, the expected inflation factor is likely to be the least reliable factor, according to Frank and Goyal (2009). It is the only macroeconomic factor to be included and so instead of having over 5,410 firm-year observations, there are just 13 annual observations for expected inflation.
II.
Data Description
A. Sample
The sample consists of 417 US industrial companies between 2002 and 2014, which is derived from the dataset used by Frank and Goyal (2009). The dataset used in this study can be seen as a continuation on their dataset that consists of US industrial firms for the period from 1950 to 2003. Therefore the adjustments that have been applied to my dataset are approximately equal to the adjustments made in Frank and Goyal (2009).
The accounting data, for the core factors described by Frank and Goyal, is drawn from the COMPUSTAT database. In addition, the ISS dataset provides data for the number of common shares owned by CEOs in the firm. Data for the expected inflation factor is collected from the Livingston Survey.1
These datasets are merged and has been made independent from annual
fluctuations in external factors, such as inflation. Within this dataset, firms with missing
values of assets and management ownership are dropped from the sample. Also financial firms are excluded because these companies have to comply with stricter legal regulations than industrial firms (Frank and Goyal, 2009). Therefore it would be impossible to draw conclusions over all kinds of firms. Moreover, our analysis is limited to large firms only that are in the dataset for the whole period. This approach is used for simplicity so it would be easier to interpret the results that measure the impact of the financial crisis. Since this dataset is used, the conclusions of my analysis do not necessarily have to apply to smaller firms, what can be described as an external validity issue.
Furthermore, the applied adjustments to eliminate outliers are done by winsorizing the leverage ratio at the 0.50% level in both tails of the distribution. These adjustments replace outliers and the most extremely misrecorded data, and is also applied by Frank and Goyal (2009). Winsorizing is the transformation of any value above the 99.5 percentile and any value below the 0.5 percentile to the value of the observations at the cutoffs. The assumption made is that the outlier is misrecorded and estimates will be improved if the outlier replaced by other fitted data.
B. Variables for Analyzing the Capital Structure
Table I lists the dependent and explanatory variables for the analysis of capital structure levels. This study examines the impact of management ownership on capital structure levels of a firm. But there are several other factors that appear in other corporate governance studies, which influence the capital structure of a firm and for which I have to control for. In this study I use the core factors described by Frank and Goyal (2009) as control factors in my model. The ratios that serve as the control factors are therefore calculated in the same way as in the study of Frank and Goyal (2009).
Moreover, this study measures the book value of leverage at the end of each fiscal year to assess the influence of management ownership and controls upon capital structure levels, whereas Frank and Goyal (2009) use several definitions of leverage. In addition, I take the book value of assets and leverage to exclude the impact of the financial crisis during the period 2008 to 2009. The financial crisis could, for instance, decline the market value of equity and affect the firm's funding costs for debt. Therefore it would make the interpretation of the coefficients easier by using the book values of leverage.
Table I.
Data Description
Definitions and descriptive statistics for variables used in analysis of capital structure levels. The variables are based on the variable definitions described in Frank and Goyal (2009), except for the additional management ownership factor. The sample consists of 5,410 observations for 417 companies in the 2002 to 2014 period after eliminating financial firms, firms that are not in the dataset for a given year and firms with missing values on management ownership and total assets. In addition, the leverage ratio is winsorized at the 0.50% level in both tails of the distribution to drop outliers.
Dependent Variable Definition Mean Standard
Deviation
Leverage ratio Total debt (book value) ÷
total assets (book value) 0.224 0.147
Independent Variables Definition Mean Standard
Deviation
Management ownership Shares owned by CEO ÷
common shares outstanding 0.023 0.056 Profitability
Operating income before depreciation ÷ total assets
(book value)
0.131 0.089
Firm size Log of total assets (book
value) 8.630 1.730
Growth Capital expenditure ÷ total
assets (book value) 0.046 0.049 Industry growth Median of growth by SIC
code and year 0.041 0.038
Tangibility
Net property, plant, and equipment ÷ total assets
(book value)
0.562 0.397
Expected inflation
Expected change in the consumer price index over
the coming year
2.299 1.087
During crisis management ownership
Management ownership
factor during 2008-2009 0.003 0.021 Post crisis management
ownership
Management ownership
factor during 2010-2014 0.007 0.033
C. Impute Missing Data
The standard regression approach drops the firms out of the sample if they are suffering from missing necessary values. However, the missing values could be missing for a reason so that biases arise in the regression results from dropping missing values. Therefore I included an additional column titled ‘Impute Missing Values’ in every regression table that presents regression estimates based on a dataset where multiple imputation is applied for imputing
missing values. Multiple imputation replaces the missing values by making predictions based on the data that has been recorded.
The estimates from this model contain less bias and the model is implemented using the ICE procedure in Stata, which is similar to the treatment of missing values in the study of Frank and Goyal (2009). Using the univariate normal model as imputation model to impute the missing values, I assume that missing values are missing at random and that the variables are normal distributed.
III. Analysis of Capital Structure Levels and Management Ownership
This section presents the empirical results of OLS coefficient estimates for the relationship between leverage and management ownership. As noted in the preceding discussion, the literature finds contradictory evidence as to how leverage is related to management ownership. Jensen and Meckling (1976) find a negative relation because of the reduced agency costs between the CEO and external shareholders. Conversely, Berger et al. (1997) suggest that management ownership is positively related to leverage through the increasing proportion of managerial entrenchment. Bruslerie and Latrous (2012) add a new insight to this discussion by finding evidence for a non-linear relationship. The research strategy of this study is to investigate if management ownership significantly influences the leverage ratio of a firm, and in what way these are related to each other.Section III.A describes the empirical models used to investigate the relationship between leverage and management ownership, which are estimated by the OLS regressions. Section III.B explores the linear relationship, whereas section III.C presents OLS estimates for a non-linear relationship.
A. Empirical method
Table I shows the descriptions of the dependent and explanatory variables that are included in the empirical models. These models consist of a methodology similar to that of Frank and Goyal (2009) to control for several other influences on leverage, complemented by a management ownership factor. Within the empirical method, standard errors are corrected for heteroscedasticity and are clustered at the firm level. Moreover, in every regression table is an additional column titled ‘Impute Missing Values’ included that presents regression estimates based on a dataset where multiple imputation is applied for imputing missing values.
The first stage of the analysis examines the linear relationship between leverage and management ownership. The regression equation used for the OLS estimates of a linear relationship is:
𝑳𝒆𝒗𝒆𝒓𝒂𝒈𝒆 𝒓𝒂𝒕𝒊𝒐𝒊𝒕
= 𝜷𝟎+ 𝜷!𝑀𝑎𝑛𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!"+ 𝜷!𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦!" + 𝜷!𝐹𝑖𝑟𝑚 𝑠𝑖𝑧𝑒!"+ 𝜷𝟒𝐺𝑟𝑜𝑤𝑡ℎ!"+ 𝜷𝟓𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑚𝑒𝑑𝑖𝑎𝑛 𝑔𝑟𝑜𝑤𝑡ℎ!" + 𝜷!𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦!"+ 𝜷𝟕𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑖𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛!+ 𝜺𝒊𝒕
The second stage of the analysis examines the non-linear relationship between leverage and management ownership. The empirical model to be estimated is:
𝑳𝒆𝒗𝒆𝒓𝒂𝒈𝒆 𝒓𝒂𝒕𝒊𝒐𝒊𝒕 = 𝜷𝟎+ 𝜷𝟏𝑀𝑎𝑛𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!" + 𝜷𝟐𝑀𝑎𝑛𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!"! + 𝜷 𝟑𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦!" + 𝜷𝟒𝐹𝑖𝑟𝑚 𝑠𝑖𝑧𝑒!"+ 𝜷𝟓𝐺𝑟𝑜𝑤𝑡ℎ!"+ 𝜷𝟔𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑚𝑒𝑑𝑖𝑎𝑛 𝑔𝑟𝑜𝑤𝑡ℎ!" + 𝜷𝟕𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦!"+ 𝜷𝟖𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑖𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛!+ 𝜺𝒊𝒕
These models contain subscripts i and t, which represent firms and years. The estimated coefficients of the included factors are represented by the betas.
B. Regression Results of a Linear Relationship
Table II presents regression estimates of the capital structure coefficients to investigate the impact of the management ownership factor on leverage. As the coefficient displays in the second column of table II, the management ownership coefficient is negative and statistically significant at 0.01 levels of confidence. The negative coefficient implies that there exists a negative relationship between leverage and management ownership. Because of the use of percentages for the leverage and management ownership factor, the interpretation of the coefficients is different from the absolute values in the table. Taking the derivative indicates that an increase of 1% in management ownership results in a decrease of 0.296% in the leverage ratio. In addition, the economic effect of management ownership on leverage can be measured by calculating the typical changes in leverage. Typical changes involve the standard deviations in their calculations, and thereby put the absolute changes into perspective. To
illustrate, the typical changes show the effect of an increase of one standard deviation in management ownership on the leverage ratio compared by a change in leverage of a standard deviation. An increase in management ownership of one standard deviation, as shown in table I, leads to a decrease of 1.66% in the leverage ratio. Relative to the standard deviation of leverage, if management ownership increases by 1 standard deviation, then the leverage ratio drops by 11.3% of a standard deviation. Consequently, the estimate of the management ownership coefficient implies a sizeable and economically meaningful economic effect of CEO equity ownership on the leverage ratio of a firm.
Results in table II display a negative, significant coefficient for the management ownership factor, and thereby support the empirical findings of Jensen and Meckling (1976), Friend and Lang (1988) and Firth (1995). This implies that an increase in the proportion of equity owned by the CEO would align the CEO’s interest to external shareholder’s preferences. Less agency instruments are therefore required so the leverage level decreases. Jensen and Meckling (1976) also suggest a decrease in leverage due to the risk aversion of CEOs, but this theory is examined through the crisis regressions.
In addition, the OLS estimate of this coefficient become even more significant when missing values are imputed. These findings are consistent with the hypothesis, put forward in this paper, that management ownership is a significant factor for predicting the leverage ratio of a firm. Hence, the results demonstrate that the model, described by Frank and Goyal (2009), excludes one of the real important determinants of capital structure because the model leaves out the management ownership factor.
For the OLS coefficient estimates of the included control variables, the coefficients on the variables ‘Firm size’, ‘Growth’, and ‘Tangibility’ are estimated as predicted and significantly at the 0.05 level of confidence. The estimates show a positive relation between leverage and firm size, and supports several other capital structure studies (Fama and French, 2002; Frank and Goyal, 2009; Danis, Rattl, and Whited, 2013). The findings suggest that larger firms face lower debt-related agency costs because of the better reputations they have in debt markets, and are therefore more levered. Growing firms are likely to be less levered due to the higher probability on financial distress costs, according to the findings. The estimate of the tangibility coefficient is, however, only estimated as predicted for the imputed data. This difference was predictable for having 402 missing observations on tangibility. The estimates on the imputed sample are in line with the findings of Titman and Wessels (1988). They argue that with firms with a high proportion of tangible assets are likely to face relatively lower expected distress costs because these assets can be collateralized.
Table II.
Linear Regression Estimates of the Management Ownership Factor
This table reports OLS regression coefficients estimates of the linear impact of management ownership on capital structure levels. The sample consists of 5,410 observations for 417 companies in the 2002 to 2014 period. The t-statistics are shown in parentheses below the estimated coefficients. Moreover, the standard errors are robust, and clustered at the firm level. The multivariate regression controls for the core factors described by Frank and Goyal (2009). Using the univariate normal model as imputation model to impute the missing values, from which the results are displayed in column (3). This model assumes that missing values are missing at random.Dependent Variable: Leverage (Book Value) Univariate Regression (1) Multivariate Regression (2) Impute Missing Values (3) Management ownership -0.432*** (-4.48) -0.296*** (-2.71) -0.310*** (-2.82) Profitability -0.143* (-1.68) -0.081 (-1.03) Firm size 0.018*** (3.82) 0.011*** (2.94) Growth -0.275** (-2.15) -0.304** (-2.40) Industry growth -0.218 (-0.97) 0.066 (0.29) Tangibility 0.137*** (6.98) 0.124*** (6.86) Expected inflation -0.111 (-1.33) -0.131* (-1.73) Year indicator
variables Yes Yes Yes
Observations 5,410 5,000 5,410
Adjusted R-squared 0.031 0.149 0.120
F-statistic 15.45 15.49 15.25
P-value of F-statistic 0.000 0.000 0.000
No other coefficient estimate shows statistical significance at the conventional 0.05 level of confidence. Frank and Goyal (2009) already suggest that the profitability factor played a powerful role in determining leverage in the period before the 1980s, but became less important. They explain that equity financing became a more important source for unprofitable firms with good growth prospects, so the significant difference in leverage between these firms disappeared. Furthermore, the estimates suggest that it was reasonable of Rajan and Zingales (1995) to exclude median industry growth and expected inflation in their selection of control factors. The expected inflation factor was, however, mentioned by Frank and Goyal (2009) as the least reliable factor, because it is the only macroeconomic factor in the empirical model.
C. Regression results of a Non-Linear Relationship
Table III shows OLS coefficients estimates of a non-linear relationship between management ownership and leverage. As shown by the second column, the squared variable of management ownership is negative and statistically significant at 0.01 levels of confidence. In addition, the coefficient of the squared management ownership variable is more significant than the linear coefficient, according to the t-values. The adjusted R-squared values indicate that the squared factor also better fits the data. These results suggest that the theory of Jensen and Meckling (1976) is true, but could be better interpreted as a quadratic function.
However, after controlling for Frank and Goyal’s “core factors”, the coefficients of the non-linear regression, displayed in columns 4 and 5, are not significant for the management ownership factor. These results do not seem to support the view of Bruslerie and Latrous (2012) that there exists a U-shaped relationship between leverage and management ownership for US nonfinancial firms. Hence, this study does not find evidence for an increase in leverage at low levels of management ownership due to the CEO’s entrenchment. Regarding to the results shown in Table III, this study only finds evidence for a negative relation at all levels of management ownership.
Table III.
Non-Linear Regression Estimates of the Management Ownership Factor
This table presents OLS regression coefficients estimates of the non-linear impact of management ownership on capital structure levels. Columns (1) and (2) display the regression results of a squared relationship between management ownership and leverage. In addition, columns (3) and (4) test the existence of a parabolic curve. Furthermore, the sample consists of 5,410 observations for 417 companies in the 2002 to 2014 period. The t-statistics are shown in parentheses below the estimated coefficients. The standard errors are robust, and clustered at the firm level. Frank and Goyal (2009) described the core factors, which are used as control variables in this regression. Using the univariate normal model as imputation model to impute the missing values. This model assumes that missing values are missing at random.Dependent Variable: Leverage (Book Value) Squared Relationship (1) Squared + Controls (2) Non-Linear Relationship (3) Non-Linear + Controls (4) Impute Missing Values (5) Management ownership squared -0.951*** (-4.87) -0.763*** (-3.30) 0.391 (0.79) -0.548 (-1.05) -0.507 (-0.95) Management ownership -0.571** (-2.41) -0.094 (-0.40) -0.125 (-0.51) Profitability -0.142* (-1.67) -0.142* (-1.67) -0.080 (-1.01) Firm size 0.019*** (4.15) 0.018*** (3.92) 0.012*** (3.06) Growth -0.282** (-2.20) -0.281** (-2.20) -0.308** (-2.44) Industry growth -0.229 (-1.01) -0.225 (-1.00) 0.063 (0.28) Tangibility 0.140*** (7.16) 0.139*** (7.07) 0.126*** (6.96) Expected inflation -0.098 (-1.19) -0.102 (-1.21) -0.123 (-1.59)
Year indicator variables Yes Yes Yes Yes Yes
Observations 5,410 5,000 5,410 5,000 5,410
Adjusted R-squared 0.024 0.150 0.032 0.150 0.123
F-statistic 16.50 16.26 14.75 15.43 15.41
P-value of F-statistic 0.000 0.000 0.000 0.000 0.000
IV. Analysis of the Impact of the Financial Crisis
This section presents the OLS coefficient estimates of the impact the financial crisis on the management ownership factor. These coefficient estimates are assumed to generate new insights into the management ownership discussion. Jensen and Meckling (1976) state that the leverage level is negatively affected by a higher proportion of management ownership because of the risk aversion of the CEO and a reduction of the agency costs. In this section, an empirical model is estimated that investigate whether the risk aversion of CEOs is indeed a true explanation of the management ownership problem, by assuming that the default risk increased during the financial crisis. The analysis in this section examines the effect of the financial crisis during the crisis the period 2008 to 2009, but also after the crisis.
The included single dummy variables demonstrate the difference in leverage levels between periods. Duchin, Ozbas, and Sensoy (2010) find evidence for a negative coefficient estimate for the during-crisis dummy variable. Under the assumption that the crisis resulted in a decreased supply or higher costs of debt financing, the leverage ratio is expected to be relatively lower during the crisis. After the crisis, the Federal Reserve flooded the markets with money, so the costs of debt financing decreased. The coefficient estimate for the after-crisis dummy is therefore expected to be positive.
A. Empirical Method
The descriptions of the variables used in this empirical model are shown in table I. The model consists of interaction variables between management ownership and the effect during and post crisis, complemented by Frank and Goyal’s (2009) control variables. Within the empirical method, standard errors are robust to heteroscedasticity and clustered at the firm level.
The regression equation used to examine the impact of the financial crisis is: 𝑳𝒆𝒗𝒆𝒓𝒂𝒈𝒆 𝒓𝒂𝒕𝒊𝒐𝒊𝒕 = 𝜷𝟎+ 𝜷𝟏𝑀𝑎𝑛𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!"+ 𝜷𝟐𝐷𝑢𝑟𝑖𝑛𝑔 𝑐𝑟𝑖𝑠𝑖𝑠 𝑑𝑢𝑚𝑚𝑦! + 𝜷𝟑(𝐷𝑢𝑟𝑖𝑛𝑔 𝑐𝑟𝑖𝑠𝑖𝑠 𝑑𝑢𝑚𝑚𝑦! . 𝑀𝑎𝑛𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!") + 𝜷𝟒𝑃𝑜𝑠𝑡 𝑐𝑟𝑖𝑠𝑖𝑠 𝑑𝑢𝑚𝑚𝑦! + 𝜷𝟓(𝑃𝑜𝑠𝑡 𝑐𝑟𝑖𝑠𝑖𝑠 𝑑𝑢𝑚𝑚𝑦! . 𝑀𝑎𝑛𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝!") + 𝜷𝟔𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦!"+ 𝜷𝟕𝐹𝑖𝑟𝑚 𝑠𝑖𝑧𝑒!"+ 𝜷𝟖𝐺𝑟𝑜𝑤𝑡ℎ!" + 𝜷𝟗𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑚𝑒𝑑𝑖𝑎𝑛 𝑔𝑟𝑜𝑤𝑡ℎ!"+ 𝜷𝟏𝟎𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦!" + 𝜷𝟏𝟏𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑖𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛!+ 𝜺𝒊
The empirical model contains a ‘during crisis dummy’ variable, which equals one for the years 2008 and 2009, else 0. In addition, the ‘post crisis dummy’ variable equals one for the period 2010 to 2014, else 0.
B. Regression Results
Table IV presents regression estimates of the crisis and control coefficients to investigate the impact of the financial crisis on the management ownership variable. Regarding to the estimated coefficients of the crisis interaction variables, the estimates show insignificant results for both interaction variables. This result implies that the management ownership factor is not affected by the financial crisis at all. Moreover, these findings provide new evidence for the management ownership discussion. Jensen and Meckling (1976) state that the ownership of equity will have an increasing impact on the CEO’s interest in the firm. Since CEOs are risk averse, they would be less likely to take excessive risk. Under the assumption that default risk of firms increased during the financial crisis, these insignificant estimates provide evidence to reject the risk aversion theory of Jensen and Meckling (1976).
In addition, the empirical findings suggest that management ownership is not negatively related to leverage due to the risk aversion of CEOs. The insignificant coefficient of the interaction variables show that CEOs with lots of equity did not lower the leverage ratio during and post the crisis more than less owning CEOs. Therefore, the negative relationship between leverage and management ownership, presented in table II, could be explained as a reduction of the agency costs between CEO and external shareholders. Hence, when managerial ownership increases, less agency instruments are required to discipline managers via the debt market so the leverage level decreases (Jensen and Meckling, 1976).
For the single crisis dummies, the coefficient estimates show also insignificant results.2
These findings suggest that the financial crisis did not have a significant, direct impact on the leverage ratios of firms. Consequently, the higher costs of debt financing during the financial crisis did not influence the leverage ratios significantly.
2 Correlation values between these crisis dummies and the management ownership variable are below the 0.3
threshold. The possibility on multicollinearity between these variables is therefore assumed as negligible to explain the insignificance.
Table IV.
Crisis Regression Estimates of the Management Ownership Factor
This table reports OLS regression coefficients estimates of the crisis impact on capital structure levels, and whether different management ownership levels affect this impact. Therefore, each model includes an interaction term between a crisis dummy and the management ownership factor. Columns (1) and (2) display the regression results of the impact during the crisis on capital structure levels for the period 2008 to 2009. In addition, columns (3) and (4) test the influence of the crisis with additional post crisis factors. Furthermore, the sample consists of 5,410 observations for 417 companies in the 2002 to 2014 period. The t-statistics are shown in parentheses below the estimated coefficients. The standard errors are robust and clustered at the firm level. Frank and Goyal (2009) described the core factors, which are used as control variables in this regression. Using the univariate normal model as imputation model to impute the missing values. This model assumes that missing values are missing at random.
Dependent Variable: Leverage (Book Value)
During Crisis (1) During Crisis + Controls (2) Whole Crisis Effect (3) Whole Crisis Effect + Controls (4) Impute Missing Values (5) Management ownership -0.426*** (-4.66) -0.288*** (-2.75) -0.420*** (-4.81) -0.268*** (-2.82) -0.282*** (-2.95) During crisis dummy 0.005
(1.38) 0.000 (0.02) 0.005 (1.10) 0.000 (-0.03) 0.001 (0.27) During crisis x management ownership -0.013 (-0.19) -0.028 (-0.37) -0.020 (-0.24) -0.048 (-0.51) -0.036 (-0.40)
Post crisis dummy 0.001
(0.11) 0.000 (-0.04) -0.003 (-0.52) Post crisis x management ownership -0.017 (-0.24) -0.057 (-0.65) -0.065 (-0.79) Profitability -0.149* (-1.75) -0.148* (-1.74) -0.086 (-1.09) Firm size -0.017*** (3.86) -0.017*** (3.81) -0.011*** (2.91) Growth -0.258** (-2.04) -0.260** (-2.05) -0.287** (-2.28) Industry growth -0.234 (-1.05) -0.235 (-1.05) 0.046 (0.20) Tangibility -0.137*** (7.00) -0.137*** (7.00) -0.125*** (6.89) Expected inflation 0.000 (0.25) 0.000 (0.13) 0.000 (0.11)
Year indicator variables No No No No No
Adjusted R-squared 0.027 0.144 0.026 0.144 0.121
F-statistic 9.64 13.84 6.10 11.60 10.61
P-value of F-statistic 0.000 0.000 0.000 0.000 0.000
Significant at 1 percent (***), 5 percent (**), and 10 percent (*) levels.
V.
Limitations
This section describes the limitations of this study, which are important to account for when interpreting the empirical results. For now, it goes beyond the scope of this study to solve these limitations, but this may be interesting for future research. The limitations can be divided into two terms of validity. Internal validity explains the limitations inside the empirical model. In addition, external validity discusses if the empirical results can be generalized to other samples.
A. Internal Validity
The empirical method, used for the analysis of the relationship between leverage and management ownership, assumes the management ownership factor to be exogenous. However, Bruslerie and Latrous (2012) state that reverse causality is an empirical problem that arises between leverage and management ownership by referring to Jensen (1986) and may result in a simultaneous causality bias. Jensen (1986) suggests that the use of external equity financing reduces through debt financing. According to this theory, a higher leverage ratio would yield a higher proportion of management ownership. This endogenous relationship implies that management ownership can influence the leverage ratio of a firm, but the level of management ownership could also be affected by leverage at the same time.3
The empirical results of this study, estimated by simple OLS regressions, could therefore be deducted from a biased and inconsistent model that suffers from a simultaneous causality bias (Stock and Watson, 2011). Hence, future empirical research could use more sophisticated estimation technique to solve this empirical problem by making use of instrumental variables.
Furthermore, selection issues could arise because the sample used for this study only contains firms that are in the dataset for the whole period. By using this selection criteria, small firms and firms that went bankrupt are excluded from the sample so the sample may be
3 Several other ownership studies also ignore this potential for reverse causality, e.g.; Firth (1995) and Berger et al. (1997).
suffering from a survivorship bias. However, the number of dropped observations is relatively small and the databases used for this study consist of a well-diversified sample, so the selection issues should be limited.
As for the control variables used in this study, the empirical models could be suffering from omitted variable bias because it leaves out some factors. To illustrate, Frank and Goyal (2009) also find other significant factors, which influence the capital structure, beside of the “core model of leverage”, such as supply-side factors and stock market conditions. These factors are not controlled in the implemented empirical models of this study. In addition, Korajczyk and Levy (2003) find that firms facing financial constraints do not choose the same capital structure as unconstrained firms.They explain that unconstrained firms time their issue of debt to coincide with periods of favorable macroeconomic conditions, while constrained firms do not. Future research could investigate the impact of these other factors.
Finally, this study suggests that risk aversion is not the main channel behind the Jensen and Meckling (1976) theory because the leverage ratios were not significantly influenced by the financial crisis. This conclusion is based on the assumption that the firm’s default risk increased during and after the financial crisis. There is, however, no evidence provided by this study that validates this assumption. Moreover, by assuming that the financial crisis did increase firm’s default risk, CEOs may have reduced this default risk with other methods than reducing the leverage ratio. Leverage could therefore be considered as a bad measurement of risk-aversion, hence future research may examine the most reliable measurement.
B. External Validity
The sample used in this study consists of large US nonfinancial firms during the period of 2002 to 2014. For instance, the average amount of the firm’s total assets in the sample is 36,326.57 million dollar, what can be deducted from the logarithms of the average firm’s total assets shown in table I. Therefore the conclusions of my analysis do not necessarily have to apply to smaller firms. But, according to Berger et al. (1997), large US companies represent a substantial fraction of the market capitalization of all US public companies. The empirical results should therefore be valid for all US publicly traded companies. Although this study should be applicable to US firms, future research is needed to examine if these empirical results are also valid for firms in other countries. The U-shaped relationship between leverage and management ownership, found by Bruslerie and Latrous (2012) for French companies, is, for instance, not applicable to US firms.
VI. Conclusion
This paper introduced a new method to analyze the impact of management ownership on the firm’s capital structure. While contradictory theories show mixed results, this study finds new evidence that is partly consistent with the ownership theory of Jensen and Meckling (1976). The empirical results presented in this paper support the negative relationship between leverage management ownership, but also reject one possible explanation for this negative relation that is stated by Jensen and Meckling (1976). In particular, this study examines the impact of the financial crisis on the management ownership factor, and finds evidence against the risk aversion theory of equity owning CEOs.
A panel analysis of US nonfinancial firms is used to examine the relationship between leverage and management ownership. This analysis extends the “core model of leverage”, described in Frank and Goyal‘s (2009) paper, by adding an additional, significant management ownership factor into the model. The results of this study provide support for the absence of one of the real important determinants of capital structure in the “core model of leverage”, complemented by a negative effect on leverage arising from equity ownership of CEOs. This evidence is consistent with the findings of Friend and Lang (1988) and Firth (1995) that less debt is needed as management ownership increases because the preferences of the CEOs and the external shareholders become more aligned. In addition, the estimates presented in this paper imply a big economic effect of CEO equity ownership on the leverage ratio of a firm. After controlling for other factors, an increase of 1 standard deviation in management ownership, results in a reduction of the leverage ratio by 11.3% of a standard deviation.
The results also indicate that the findings of Bruslerie and Latrous (2012) on French firms are not applicable to US firms. For US firms, the estimates do not suggest the existence of a U-shaped relationship between leverage and management ownership. At low levels of management ownership, leverage seems not to increase due to the CEO’s entrenchment. However, the non-linear analysis show that the relation between leverage and management ownership could be better interpreted as a quadratic function.
Since literature show contradictory findings about the influence of management ownership on leverage, this study also examines the interaction between management ownership and the financial crisis. The estimates imply that the management ownership factor is not affected by the financial crisis at all. This outcome provides new evidence for the management ownership discussion because the explanation of Jensen and Meckling (1976)
for a negative relationship is now partly rejected. According to their study, the leverage ratio of a firm decreases for an increase in management ownership, which is partly caused by the risk aversion of CEOs. Under the assumption that default risk of firms increased during the financial crisis, these insignificant estimates provide evidence to reject the risk aversion channel of Jensen and Meckling’s (1976) theory.
Some questions remain, however, unsolved. Section V already describes some limitations of this study, which open the way to future analysis. Moreover, this paper focuses mainly on the examination of the risk aversion channel of Jensen and Meckling (1976). Since several other studies find a positive relationship between leverage and management ownership due to managerial entrenchment, there is still no clear theory to explain the relation. Future research could therefore investigate these contradictory results by using the same method of analysis as this paper. Testing the interaction between management ownership and certain circumstances could exclude some management ownership theories.
References
Barclay, M. J., Marx, L.M., and Smith, C.W. (2003). The joint determination of leverage and maturity. Journal of Corporate Finance, 9, pp. 149–167.
Berger, G. P., Ofek, E., and Yermack, D. L. (1997). Managerial entrenchment and capital structure decisions. Journal of Finance, 52, pp. 1411-1438.
Billett, M.T., King, T.D., and Mauer, D.C. (2007). Growth opportunities and the choice of leverage, debt maturity, and covenants. Journal of Finance, 62, pp. 697–730.
Bruslerie, H., and Latrous, I. (2012). Ownership structure and debt leverage: Empirical test of a trade-off hypothesis on French firms. Journal of Multinational Financial
Management, 22(4), pp. 111-130.
Danis, A., Rattl, D., and Whited, T. (2014). Refinancing, profitability and capital structure. Journal of Financial Economics, 114(3), pp. 424-443.
Duchin, R., Ozbas, O., Sensoy, B. (2010). Costly external finance, corporate investment, and the subprime mortgage credit crisis. Journal of Financial Economics, 97, pp. 418–435. Elsas, R., Flannery, J.M., and Garfinkel, J.A. (2013). Financing major investments:
Information about capital structure decisions. Review of Finance, 18(4), pp. 1341-1386.
Fama, E.F., and French, K.R. (2002). Testing trade-off and pecking order predictions about dividends and debt. Review of Financial Studies, 15, pp. 1–33.
Firth, M. (1995), The Impact of Institutional Stockholders and Managerial Interests on the Capital Structures of Firms. Managerial and Decision Economics, 16(2), pp. 167–75. Frank, M., and Goyal, V. (2009). Capital structure decisions: Which factors are reliably
important?. Financial Management, 38, pp. 1–37.
Friend, I., and Lang, L.H.P. (1988). An Empirical Test of the Impact of Managerial Self Interest on Corporate Capital Structure. Journal of Finance, 423(2), pp. 271–81. Huang, R., and Ritter, J.R. (2009). Testing Theories of Capital Structure and Estimating the
Speed of Adjustment. Journal of Financial and Quantitative Analysis, 44, pp. 237-271.
Jensen, M.C., and Meckling, W.H. (1976). Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure. Journal of Financial Economics, 3, pp. 305-360. Korajczyk, R.A., and Levy, A. (2003). Capital Structure Choice: Macroeconomic Conditions
and Financial Constraints. Journal of Financial Economics, 68, pp. 75-109. Myers, S.C. (1984). The Capital Structure Puzzle, Journal of Finance, 39, pp. 575-592.
Palepu, K.G. (1986). Predicting takeover targets: A methodological and empirical analysis. Journal of Accounting and Economics, 8(1), pp. 3-37.
Rajan, R. and Zingales, L. (1995). What Do We Know about Capital Structure: Some Evidence from International Data. Journal of Finance, 50, pp. 1421-1460. Shyam-Sunder, L., and Myers, S.C. (1999). Testing static tradeoff against pecking order
models of capital structure. Journal of Financial Economics, 51, pp. 219–244. Stock, J.H., and Watson, M.W. (2011). Introduction to Econometrics 3th edition. Boston,
MA: Pearson Education, Inc.
Stulz, R. (1988). Managerial control of voting rights: Financing policies and the market for corporate control. Journal of Financial Economics, 20, pp. 25-54.
Taggart, R.A., Jr. (1985). Secular Patterns in the Financing of US Corporations. In B.M. Friedman, Ed., Corporate Capital Structures in the United States, Chicago, University of Chicago Press.
Titman, S., and Wessels, R. (1988). The determinants of capital structure choice. Journal of Finance, 43, pp. 1–18.
