Sophie Zielhuis
Student number – 11036656 University of Amsterdam Economics and Business
Specialization – Economics and Finance Bachelor Thesis
Supervisor – Shivesh Changoer 12 ECTS
June 26, 2018
Influence of the Financial Crisis in 2008 on U.S. Firms’
Capital Structure Determinants
In this study, I examine the influence of the 2008 financial crisis on firms’ capital structure determinants. My analysis is based on a sample of S&P500 firms. The results show that the crisis did not have any significant effect on profitability and tangibility as capital structure determinants. Therefore, I conclude that the financial
Statement of Originality
This document is written by Student Sophie Zielhuis who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
1. Introduction
Modigliani and Miller’s (1958) irrelevance theory is a broadly accepted theory. This theory suggests that a firm’s capital structure does not affect the firm value. However, this theory explains reality to a very limited extent, since it based on some non-realistic assumptions such as perfect capital markets and the absence of corporate taxes (Robichek & Myers, 1958).
Nevertheless, in reality the capital structure does influence a firms’ value, which led to multiple studies investigating the capital structure under non-perfect market conditions. This revealed new theories explaining financial decisions regarding the capital structure. Kraus and Litzenberger (1973) introduced the trade-off theory, whereas Myers and Majluf (1984) identified the pecking order theory. More recent research examines the influence of firm- and country-specific determinants of the capital structure (Titman & Wessels, 1988; Rajan & Zingales, 1995; Wald, 1999; De Jong, Kabir & Nguyen, 2008; Frank & Goyal, 2009). These researchers show that profitability and tangibility are important capital structure determinants.
In this paper, I examine whether the influence of profitability and tangibility as most
influential capital structure determinants has changed during the 2008 financial crisis. I choose to only analyze the effect of the financial crisis on profitability and tangibility as capital structure
determinants, since Frank and Goyal (2009) conclude these determinants are the most influential and consistent factors determining a firms’ capital structure.
My analysis is based on a sample of firms listed on the S&P500 in the United States. I focus on U.S. listed firms for multiple reasons. The first reason is that the financial crisis in 2008 started in the United States and exerted a significant influence on U.S. firms’ capital structure. For this reason, the change that I can measure the effect of the crisis is the largest within this sample with U.S. firms. The second reason is that there is a large amount of data available for U.S. firms. Finally, my results can be compared with the results in previous research (see, e.g., Rajan & Zingales, 1995; Akhtar, 2012; Fosberg, 2012; Alves & Francisco, 2013) on firm-specific capital structure determinants since this is often based on sample of U.S. listed firms.
The sample period covers the years 2004 until 2011. I focus on this period based on prior studies investigating the financial crisis in 2008 (Alves & Francisco, 2013; Harrison & Widjaja, 2014; Iqbal & Kume, 2014). The fall of the Lehman Brothers on 15 September 2008 is identified as the official start of the global financial crisis (Mishkin, 2011). For this reason, the prior crisis period covers the period of 2004 until 2007 and I have chosen the period of 2008 until 2011 as the crisis and after-crisis period.
My results suggest that while the 2008 financial crisis did not have significant impact on tangibility and profitability as capital structure determinants, it increased the importance of financial decisions since the crisis shows a positive effect on the long-term leverage of U.S. firms.
profitability of firms becomes a less strong capital determinant and tangibility becomes a stronger capital determinant.
My study builds on the work of Akhtar (2012) who examines the effects of various business cycle phases on capital structure decisions, and extends the work of Deesomsak, Paudyal and Pescetto (2004) and Iqbal and Kume (2014). Deesomsak et al. (2004) examine the effect of a financial crisis in 1997 on the leverage ratios for countries in the Asia Pacific region, whereas Iqbal and Kume (2014) conduct a similar analysis for the 2008 financial crisis in the UK, France and Germany. Comparable research is performed by Alves and Francisco (2013), who examine the effect of several crises on financing decisions and the leverage ratio for 43 different countries worldwide with a great difference in financial environment. However, limited research has been performed on the influence of the financial crisis in 2008 on the capital structure determinants in the United States. The United States has a well-developed capital market and is market-based, and thus firms have easy access to external financing sources (Alves & Francisco, 2013). For this reason, the influence of the financial crisis may differ from countries with other legal, financial and institutional environments investigated in prior research (Deesomsak et al., 2004; Alves & Francisco, 2013; Iqbal & Kume, 2014). Some countries in prior research have a special regulation regarding investors protection, such as Australia, and some have banking-orientated economies, such as France and Germany (Deesomsak et al., 2004; Alves & Francisco, 2013). Moreover, the United States offers many financing sources with relatively low adaption costs. This flexibility causes that firms can easily shift between financing sources (Myers, 2001). Therefore, the major contribution of this study is to examine the influence of the 2008 financial crisis on capital structure determinants in the well-developed and flexible capital markets of the United States.
The structure of this paper is as follows: In section 2, I provide information by a literature review, in which I describe three capital structure theories and give an overview of prior research. In the third section I present my hypothesis based on existing literature. Subsequently, I explain the research method and data description in the fourth section. In the fifth section I show the unified analysis and the results of the fixed effect regression. To continue with section 6, in which I perform a sensitivity analysis to check whether the results in this research are sensitive. Finally, in section 7 I state the conclusion and limitations of this research as well as recommendations for further research.
2. Literature Review
The capital structure of a firm refers to how a company is financed (Myers, 1984). Specifically, whether firms choose to finance their investments externally with debt or equity. Debt refers to an obligation to pay money in an agreement, which makes debt a liability for the firm (Lee & Lee, 2006). According to Downes and Goodman (2014) equity represents the stake of shareholders in a firm. With the issuance of shares, you raise money for investments (Downes & Goodman, 2014). The opposing characteristic of debt to equity is that for debt a corporation agrees to repay a specific amount at a specific date and for equity not (Lee & Lee, 2006). There are different theories explaining the debt-equity choice of a firm. The three best known are the capital structure irrelevance theory, the trade-off theory and pecking order theory, which I discuss in this section. Subsequently, I present an overview of existing literature on the capital structure and their main conclusions.
2.1 Capital Structure Irrelevance Theory
The capital structure irrelevance theory is developed by Modigliani and Miller (1958). This theory suggests that the choice of the capital structure and therefore, financial policy, is irrelevant for the value of a firm. Their theory has some strict assumptions, namely perfect capital markets, no transactions cost and the absence of corporate tax (Modigliani & Miller, 1958).
2.2 Trade-off Theory
The trade-off theory origins from the work of Kraus and Litzenberger (1973). Their research builds on the irrelevance theory of Modigliani and Miller (1958) by including the taxation on corporate profits and the costs of bankruptcy in a complete market. By adding these market imperfections, Kraus and Litzenberger (1973) show that the choice of debt and equity is influenced by the amount of which the firm benefits from tax savings and suffers from possible bankruptcy costs. If there were no
bankruptcy costs, firms would be completely debt financed since this maximizes their interest tax shield (Bradley, Jarrell & Kim, 1984). However, Bradley, Jarrell & Kim (1984) state that in reality there are significant costs related to bankruptcy for which firms are not fully debt financed. Therefore, a firm chooses a debt-equity mix at which the firm value is maximized. This is at the point where the present value of the tax advantage is offset by the present value of related distress costs, which is shown in the static trade-off framework of Myers (1984). According to Myers (1984) a firm targets a debt-to-equity ratio. If the leverage ratio of a firms does not equal its target the firm will move back towards its target leverage ratio (Sing & Kumar, 2012). 1
1
Regarding to the static framework, the dynamic trade-off theory considers next periods profits and dividends and allows for small deviations from the target debt-to-equity ratio. The constant adjustments towards one target leverage ratio in the static trade-off theory leads to high transaction cost considering multiple periods, therefore the dynamic theory sets an upper and lower bound to a firms’ target leverage ratio in which a firm let its leverage move (Frank & Goyal, 2005).
2.3 Pecking Order Theory
The Pecking Order Theory was first introduced by Myers and Majluf (1984). According to the pecking order theory firms prefer internal financing to external financing. Furthermore, firms prefer raising debt rather than equity if external financing is required (Myers, 1984). Consequently, this theory suggests in contrary to the trade-off theory that there is no target leverage ratio towards firms tend to move. The leverage ratio of firms can be interpreted as the result of ordered financing. The reason behind this preferred order of finance is explained by Myers and Majluf (1984). The range of financing sources clarifies why a firm prefers internal financing, rather than debt financing, rather than equity financing (Myers & Majluf, 1984). According to Myers (1984) external financing comes at a cost, which can be explained by the information asymmetry between insiders and outsiders of the firm and signaling problems associated with external financing. To avoid those costs related to external financing, first internal financing will be used. Secondly, raising debt is preferred to raising equity, because the issuance of equity may signal that the stock is overvalued, since managers wish to issue equity only when equity is overvalued (Myers & Majluf, 1984). For this reason, the share price decreases at the equity issuance (Myers & Majluf, 1984). To avoid the costs associated with the decline in share price, managers prefer to raise debt (Myers, 1984).
However, previous studies often cannot conclude which of these theories have the most explanatory power, but investigate the firms-specific determinates which influence the choice of the capital structure of firms.
2.4.1 Prior Research on the Capital Structure
Rajan and Zingales (1995) analyze firm-specific determinants of the capital structure for listed companies in several industrialized countries, including the United States. Their sample period covers the years 1987-1990. Their main findings are that tangibility and size are positively related to
leverage. Furthermore, the market-to-book ratio, also referred as investment opportunities, and profitability show are negative relationship with the leverage ratio.
Wald (1999) contributes to the work of Rajan and Zingales (1995) by also making an international comparison and including several more firm-specific determinants. One of his main findings is that profitability, the most influential factor in this model, is negatively related to the leverage ratio. This result is in line with the pecking order theory of Myers (1984), as it predicts that unprofitable firms are forced to seek for external finance sources because they have less retained earnings to use as internal finance sources than profitable firms. Wald (1999) also finds that a firms’ growth is negative related to the leverage ratio. This finding is also in line with the pecking order theory of Myers and Majluf (1984), because it predicts that firms with high growth opportunities will make less use of debt since then their profits return to bondholders instead of stockholders.
De Jong, Kabir and Nguyen (2008) extend the work of Rajan and Zingales (1995) by examining the impact of firm-specific variables and country-specific variables on leverage for 42
different countries in the period of 1997-2001, including the United States. They regress several firm specific characteristics on the leverage ratio, and find a positive and significant impact of tangibility and firm size on corporate borrowing. Furthermore, they also find that the leverage ratio is negative related to growth opportunities, liquidity, profitability and business risk.
The fact that the recent and earlier research show similar findings is not surprising, since Lemmon, Roberts and Zender (2008), who examine a sample of non-financial firms in the United States covering a broad sample period between 1965 and 2003, show that the firm-specific factors influencing the capital structure are stable in the long-run.
2.4.2 Prior Research on Capital Structure Developments
More recent research focuses on how the capital structure changes over the years and how fluctuations in the economic environment influence financing decisions.
For example, Akhtar (2012), who examines the effect of the four stages in the business cycle (namely, peak, contraction, trough and expansion) on the capital structure. His sample contains all U.S. listed firms and covers a sample period of 1950-2010. He shows that the different business cycles play a significant role in financing decisions. Moreover, Akhtar’s (2012) findings also show that during a recession, firms have more debt relative to good economic times, since they have insufficient profits during a recession to finance investment internally.
Motivated by the latter finding, Alves and Francisco (2013) study the impact of various recent crises on the capital structure of firms in the United States. They include several environmental variables in their regression model, such as capital market development and GDP growth. To estimate the effect of different crises they include four crisis dummy variables, which respectively describe the dot.com bubble, subprime crisis bubble, European sovereign debt crisis and the 2008 financial crisis. They find that the leverage ratio increases during a crisis.
Iqbal and Kume (2014) extend the research of Alves and Francisco (2013) by examining the influence of the 2008 financial crisis on the capital structure in the UK, France, and Germany. Their study uses a sample of non-financial and non-utility firms in a period of 2006 until 2011. They find that leverage ratios, on average, increase during the crisis period and return to pre-crisis levels in the post-crisis periods. However, this result is only shown for firms in the UK, but there is no significant evidence for firms in France and Germany.
3. Hypothesis
Various studies try to identify the influence of a crisis on the capital structure and on capital structure determinants. One of these studies is the research of Harrison and Widjaja (2014). They focus on the effect of the financial crisis in 2008, and use data firms in the United States within a sample period covering 2004 until 2011. Harrison and Widjaja (2014) regress their model on the period of 2004-2007 and 2008-2011 separately and compare the coefficient estimates for all the capital determinants of these two regressions. However, they do not test whether the change in the coefficient of these variables is significant before and after the crisis. For this reason, they cannot conclude whether the crisis had a significant effect on the firm-specific factors. Therefore, I focus in this research on the currently unknown effect of the 2008 financial crisis on two firm-specific capital determinants. I focus on profitability and tangibility because Frank and Goyal (2009) show that only profitability and tangibility are significant and consistent firm-specific factors explaining a firm capital structure decisions.
3.1 Profitability
Rajan and Zingales (1995) find a highly significant negative relation of profitability on the leverage ratio. Also, De Jong et al. (2008) and Fama and French (2002) find a significant negative correlation between profitability and the leverage ratio for the United States. These findings are as expected because the pecking order theory suggests that profitable firms can use their retained earnings to finance investments instead of making use of external finance sources (De Jong et al., 2008). However, these findings contradict the trade-off theory, which assumes that profitable firms face lower costs of financial distress (Frank & Goyal, 2005). As a result, this theory predicts higher
leverage ratios for profitable firms (Frank & Goyal, 2005). Supporting the pecking order theory, Thies and Klock (1988) also find a significant negative relationship between profitability and long-term debt. I expect that the capital determinant profitability becomes less influential during the crisis. As shown in Appendix B, profits and consequently internal financing capacity decreased during the 2008 financial crisis (Yardeni & Abbott, 2018). However, firms still need to finance their operations. Therefore, as expected by the pecking order theory firms increase their leverage to finance their investments, for which the negative influence of profitability on the leverage ratio becomes less negative. Finding such an effect would support Akhtar (2012), who shows a negative interaction term between profitable and leverage during the peak and expansion phases of the business cycle and during the contraction and recession. However, the influence of profitability on leverage during contraction and recession is less negative than for the peak and expansion. These findings would also be consistent with Iqbal and Kume (2014) who find a positive interaction in the UK between crisis and profitability on the capital structure, for which profitability becomes a less strong determinant. For this reason, my first hypothesis states:
Hypothesis 1: The impact of profitability on leverage is negative and becomes less negative during
the financial crisis.
3.2 Tangibility
Frank and Goyal (2005) define tangible assets as inventory, net property plant and equipment, also Titman and Wessels (1988) use this definition. As the pecking order theory suggests firms with more tangible assets have higher leverage ratios, since these fixed assets can serve as collateral for debt (Frank & Goyal, 2005). This positive relation between tangibility and leverage is shown by Titman and Wessels (1988), Rajan and Zingales (1995) and by De Jong et al. (2008). According to Frank and Goyal (2005) these findings are also in line with trade-off framework. A high level of tangible assets reduces the agency costs of debt and thus the default probability and bankruptcy costs, which explains a higher level of debt (Frank & Goyal, 2005). Mishkin (2011) suggests that fixed assets on the
balance sheet mitigate asymmetric information and thus reduce adverse selection since they serve as collateral, therefore loans are more easily provided to firms with a high level of tangible assets (Mishkin, 1990). I expect that the capital determinant tangibility becomes stronger during the financial crisis. The reason for this is that during the crisis uncertainty in the financial market increased for which the adverse selection problem increased (Mishkin, 1990). To compensate this problem, tangible assets become more valuable as collateral and therefore have a stronger impact on the leverage ratio. Finding such an effect, would support with the findings of Akhtar (2012). He shows that during the business cycles contraction and trough tangibility becomes a stronger determinant for the capital structure. For this reason, my second hypothesis states:
Hypothesis 2: The impact of tangibility on leverage is positive and becomes more positive during the
4. Methodology
To test my hypothesis, I run the following regression analysis using a fixed effect model:
LEVi,t = β0 + β1PROFITi,t + β2TANGi,t + β3CRISIS + β4CRISIS*PROFITi,t+ β5CRISIS*TANGi,t + β 6SIZEi,t + β7MTBi,t + β8LIQUIDi,t + μi + εi,t;
In this model: i denotes a single firm. t represents the different years. μi is the unobservable
time-invariant which is unique for each firm and time and εi,t is the error term. LEVi,t represents the
long-term debt ratio at the end of the fiscal year. This ratio is defined as the book value of long long-term debt divided by the book value of total assets (De Jong et al., 2008; Akhtar, 2012; Alves & Francisco, 2013). PROFITi,t refers to the profitability of a firm. I use earnings before interest, taxes, depreciation
and amortization over the book value of total assets to measure profitability (Deesomsak et al., 2004; Frank & Goyal, 2009; Fosberg, 2012). TANGi,t is the tangibility of a firm. To measure the tangibility,
I use the net property, plant and equipment over the book value of total assets (Titman & Wessels, 1988; Rajan & Zingales, 1994; Akhtar, 2012). SIZEi,t represents the size of a firm, which I measure as
the natural logarithm of total assets (Lemmon et al., 2008; Akhtar, 2012; Iqbal & Kume, 2014). MTBi,t
refers to the market-to-book ratio of a firm, which is measured as the market value over the book value of total assets (Rajan & Zingales, 1995; Harrison & Widjaja, 2014). LIQUIDi,t represents the
liquidity of a firm. This ratio is defined as the total current assets over the book value of total assets (Deesomsak et al., 2004; De Jong et al., 2008). CRISIS equals 0 for the period prior to the financial crisis, which covers a period of 2004 until 2007, and 1 for the period during and after the financial crisis, which is from 2008 until 2011 (Alves & Francisco, 2013; Iqbal & Kume, 2014). The data on the financial accounts is data from the fiscal year end in the financial statements.
LEV is chosen as the dependent variable in my research to indicate a firms’ capital structure. I focus on the long-term debt ratio since this is often used as a measure for leverage in prior research (Akhtar, 2012; Alves & Francisco, 2013; Cheng & Jiang, 2001). I do not include the short-term debt ratio, since De Jong et al. (2008) criticized the use of total debt as a measure for leverage. They state that the level of short-term debt issuances is influenced by different determinants and therefore these results are not relevant.
PROFIT is a performance measure for a firm. I include this variable because prior research shows that the profitability of a firm is a significant and consistent determinant for its capital structure (Frank & Goyal, 2009). This variable shows a negative relation with leverage in previous studies (Thies & Klock, 1988; Rajan & Zingales, 1995; Fama & French, 2002; De Jong et al., 2008). This negative relationship is explained by the pecking order, suggesting that profitable firms can finance more of their investments internally before its required to make use of external finance sources (De Jong et al., 2008).
include this variable because prior research shows that the tangibility of a firm is a significant and consistent and influential determinant for its capital structure (Frank & Goyal, 2009). Previous studies find a positive relationship between tangibility and leverage, since tangible assets reduce the costs of financial distress and can serve as collateral for debt (Frank & Goyal, 2005).
I include an interaction between PROFITand CRISIS to examine the effect of the financial crisis on profitability as a capital determinant. I expect that the coefficient of this interaction term is positive, because the impact of profitability on leverage becomes less negative during the crisis. As I state in the hypothesis, throughout the crisis profits decreased and to finance operations firms make more use of external finance sources.
I also include an interaction between TANG and CRISIS to examine the effect of the financial crisis on tangibility as a capital determinant. I expect the coefficient of this interaction term to be positive, because the impact of tangibility on leverage becomes stronger during the crisis. As I state in the hypothesis, in times of a crisis uncertainty rises and to mitigate this, therefore tangible assets become more valuable as collateral.
SIZE is a measure for the firm size. I include this measure as a control variable because De Jong et al. (2008), Degryse, Goeij and Kappert (2012) and Rajan and Zingales (1995) find a positive relationship between size and leverage. They all conclude that larger firms are more diversified and have more stable cash flows. Therefore, they face lower costs of financial distress and issue more debt as expected by the trade-off theory (De Jong et al., 2008; Rajan & Zingales, 1995). Similarly, the pecking order theory expects a positive relation between size and leverage because size decreases the volatility of earnings and therefore mitigate asymmetric information problems (Fama & French, 2002; Degryse et al., 2012). Based on prior research, I expect the relationship between size and leverage to be positive.
MTB is a proxy for the growth opportunities of a firm. I include this as a control variable because Deesomsak et al. (2004), Rajan and Zingales (1995) and De Jong et al. (2008) find a negative relationship between growth opportunities and leverage. Deesomsak et al. (2004), as well as Rajan and Zingales (1995) state that large firms tend to invest in risky projects, which result in an increase in costs of financial distress. Therefore, firms with a high market-to-book value make more use of internal finance or issue equity because of the high share price, which is suggested by the trade-off theory (Deesomsak et al., 2004). On the other hand, this contradicts the pecking order theory since firms with growth opportunities have insufficient internal finance sources and seek for external finance, for which they issue more debt (Michaeles, Chittenden & Poutziouris, 1999). However, De Jong et al. (2008) state that firms with large growth opportunities do not prefer to issue debt, since returns from profitable investments would transfer wealth to bondholders instead of stockholders. Based on findings in prior research, I expect the relationship between the market-to-book value and leverage to be negative.
control variable because De Jong et al. (2008), Deesomsak et al. (2004), and Harrison and Widjaja (2014) find a negative relationship between liquidity and leverage. De Jong et al. (2008) state that this is in line with the pecking order theory, since firms use cash and other liquid assets for internal finance, before debt is issued to finance investments. On the other hand, the trade-off theory predicts a positive relationship since highly liquid firms face lower bankruptcy costs (Degryse et al., 2012). Based on findings in previous studies, I expect the relationship between liquidity and leverage to be negative.
I include CRISIS, a crisis dummy, to study the effect of the financial crisis on the leverage ratio. Iqbal and Kume (2014) and Alves and Francisco (2013) also include a crisis dummy in their research. Both state that the leverage ratio increased significantly during the financial crisis.
Additionally, Fosberg (2012) presents an increase in leverage for U.S. firms’ in debt during the crisis because of the problems in money and capital markets. Based on previous studies, I expect the relationship between crisis and leverage to be positive.
I include a time-invariant error term which is unique for every firm. Torres-Reyna (2017) states that “this allows to control for variables you cannot observe or measure like cultural factors or difference in business practices across companies; or variables that change over time but not across entities. This is, it accounts for individual heterogeneity” (pp. 3). The unobservable time-invariant accounts for the firms’ individual characteristics, also named, time-invariant characteristics, that may impact or bias the outcomes on the relation between the firms-specific variables and leverage. Therefore, this model only captures the net effects of the capital determinants on leverage.
I estimate my model with a fixed effect regression, because most studies examining the effect of a crisis on the capital structure use a fixed-effect regression (Iqbal & Kume, 2014; Alves &
Francisco, 2013; Akhtar, 2012). Lemmon et al. (2008) find that 60% of the variation in leverage is explained by firm fixed effects and conclude that a fixed effect regression is an appropriate model to measure the effect capital structure determinants on leverage. Considering this research, which examines the effect of the financial crisis on different firms, a fixed effects model accounts for firms-specific effects and firm-invariant effects over time and captures the net effects of the determinants on leverage.
4.2 Data
To study the influence of the financial crisis on the determinants of the capital structure I focus on all listed firms on the S&P500 in the United States. I have chosen this sample for several reasons. First, most prior research on the capital structure uses data from the United States. Therefore, this sample improves the comparability to previous studies on the U.S. (Lemmon et al., 2008; Akhtar, 2012). This is also the reason that I choose S&P500 firms since previous studies investigating the effect of a crisis use listed firms (Iqbal & Kume, 2014; Harrison & Widjaja, 2014). Secondly, as stated by Harrison and Widjaja (2014) given that global financial crisis started in the United States and exerted a
significant influence on the U.S. firms’ capital structure, the change that I measure the effect of the crisis on capital structure determinants is the largest in this sample with U.S. firms. Finally, financial data on U.S. listed firms can be easily obtained and most relevant data is complete.
My sample period covers a period of 2004 until 2011. This period is chosen based on prior research to measure the impact of financial crisis (Harrison & Widjaja, 2014; Iqdal & Kume, 2014). Iqdal & Kume (2014) define the years 2006 and 2007 as the pre-crisis period, the years 2008-2009 as crisis period and the years 2010 and later as the after-crisis period. As adopted by Harrison and Widjaja (2014), the prior crisis period refers to data in the period of 2004 until 2007, and the crisis and after crisis period refers to data in the period of 2008 until 2011.
I use panel data since it consists of multiple observations across time on different individuals (Hsiao, 1986). Furthermore, Degryse et al. (2012) make use of a panel data, because their data set contains yearly observations of various firms. To gather financial data for my sample I use the COMPUSTAT Database North America – Fundamental Annuals accessed via Wharton Research Database System. I use this database since it is often used in previous studies on the capital structure (Frank & Goyal, 2009; De Jong et al., 2008; Lemmon et al., 2008; Akhtar, 2012). Furthermore, this database contains all relevant data over the chosen period. The data on the financial accounts is data from the fiscal year end.
I omit firms operating in the financial sector (Lemmon et al., 2008). The reason is that these firms have unique capital structure characteristics which cannot be compared to non-financial firms and have different balance sheets (Lemmon et al., 2008; Akhtar, 2012). As Lemmon et al. (2008) and Akhtar (2012), I eliminate firms with missing data or negative values of book assets on financial accounts, since these data is essential for the calculation of several ratios. After I applied all criteria, I obtain a total sample consisting of 2,733 observations including 382 active S&P500 firms.
4.2.1 Descriptive statistics Table I
Descriptive statistics
This table presents the descriptive statistics for the sample in the period of 2004 to 2011. The sample consists of 382 firms in the United States listed on the S&P500. The data is retrieved from Compustat Database North America – Fundamental Annuals. For all variables in the regression model, the mean, number of observations, standard deviation, median, first and third quartile are shown. Variable definitions are explained in Appendix A.
Variable Mean N Std. Dev Q1 Median Q3
LEV 0.220 2,733 0.146 0.123 0.200 0.296
PROFIT 0.156 2,733 0.041 0.102 0.147 0.196
TANG 0.323 2,733 0.236 0.126 0.249 0.505
MTB 1.314 2,733 1.115 0.606 1.029 1.690
LIQUID 0.355 2,733 0.185 0.197 0.348 0.475
CRISIS 0.490 2,733 0.500 0.000 0.000 1.000
CRISIS*PROFIT 0.075 2,733 0.092 0.000 0.000 0.141
CRISIS*TANG 0.157 2,733 0.234 0.000 0.000 0.223
Table I summarizes the descriptive statistics for the sample. The mean of LEV, representing the long-term debt ratio, is equal to 0.220 indicating that on average, U. S. firms’ finance less than a third of their total assets with debt. This is consistent with the findings presented by Akhtar (2012) and Lemmon et al. (2008), who show a mean for the long-term debt ratio of 0.26 and 0.27 respectively. PROFIT has an average value of 0.156 which is higher than presented in previous studies of Lemmon et al. (2008) and De Jong et al. (2008). However, their samples include all U.S. firms and the S&P500 firms in my sample have on average high profits. The average value of TANG is 0.323 which is corresponding with the values presented by Akhtar (2012), Lemmon et al. (2008) and De Jong et al. (2008). However, the descriptive statistics show a large difference in the first and third quartile for TANG which indicates a large variation in tangible assets between either firms or over time. The average value of MTB is 1.314, which suggests that firms in the U.S. are overvalued since they have a higher value in the market than their actual book value. Similar results on U.S. firms in this period are shown in previous studies (Lemmon et al., 2008; De Jong et al., 2008; Akhtar, 2012). There is a large difference between the first and third quartile for the value of MTB, which indicates a large difference in market-to-book values between either firms or over time. LIQUID shows an average value of 0.355 which is consistent with findings of Harrison and Widjaja (2014).
5. Results
5.1 Unified Analysis Table II
Pearson correlations in the full sample
This table presents the correlation coefficients for the sample in the period of 2004 to 2011. The sample consists of 382 firms in the United States listed on the S&P500. The data is retrieved from Compustat Database North America – Fundamental Annuals. Variable definitions are explained in Appendix A.
To get a better understanding on how the variables in the regression are related to each other, and related to the leverage ratio, and more important to test for autocorrelation I show the Pearson correlations in table II. The coefficient of the correlation is always between -1 and +1 and according to Harrison and Widjaja (2014) the maximum Pearson correlation coefficient accepted equals 0.8. From table II, I conclude that there is no multicollinearity within the sample, since there is no correlation above 0.8. Consequently, I do not check for any multicollinearity with the variance inflation factor (VIF) test. Only CRISIS and CRISISPROFIT, CRISISTANG show a high correlation, but this makes sense since the interaction variables consist of a multiplication with the dummy.
Table II shows that PROFIT is negatively correlated with LEV, which is line with the pecking order theory, but contradicting the trade-off theory. These findings are consistent with previous studies (Fama & French, 2002; Frank & Goyal, 2005). TANG is positively related to LEV, which is expected based on the trade-off theory and the pecking order theory. This is also consistent with the positive relationship found by Rajan and Zingales (1995). MTB and LIQUID are negatively related to LEV which is also expected based on previous studies (De Jong et al., 2008; Deesomsak et al., 2004; Harrison & Widjaja, 2014). SIZE is positively related to LEV as expected by both the trade-off and pecking order theory and consistent with prior research (Fama & French, 2002; De Jong et al., 2008;
LEV PROFIT TANG SIZE MTB LIQUID CRISIS CRISIS
PROFIT CRISIS TANG LEV 1.0000 PROFIT -0.1239 1.0000 TANG 0.1783 -0.0216 1.0000 SIZE 0.0240 -0.1922 0.2023 1.0000 MTB -0.2604 -0.6090 -0.2243 -0.3174 1.0000 LIQUID -0.3784 0.1907 -0.5560 -0.3437 0.3551 1.0000 CRISIS 0.0913 -0.0260 -0.0099 0.1086 -0.1795 -0.0213 1.0000 CRISISPROFIT 0.0515 0.3556 -0.0374 0.0245 0.0475 0.0499 0.8345 1.0000 CRISISTANG 0.1499 -0.0544 0.5205 0.1850 -0.2182 -0.3126 0.6837 0.5412 1.0000
Rajan & Zingales, 1995). CRISIS shows a positive relation with LEV, which is in line with expectations based on findings in prior research (Fosberg, 2012; Iqbal & Kume, 2014).
5.2 Fixed effects regression results Table III
Fixed effect regression
This table presents the results from the fixed effect regression on leverage for three model in the period 2004-2011. The first column of the table shows the results for model 1: LEVi,t = β0 + β1PROFITi,t + β2TANGi,t + β3SIZEi,t + β4MTBi,t +
β5LIQUIDi,t + μi + εi,t; The second column shows the results for model 2: LEVi,t = β0 + β1PROFITi,t + β2TANGi,t + β3
CRISIS + β4SIZEi,t + β5MTBi,t + β6LIQUIDi,t + μi + εi,t; This model is similar to model 1 only the dummy variable CRISIS is
added; (CRISIS = 1 if YEAR>2007, otherwise CRISIS = 0). The third column shows the results for the model 3: LEVi,t = β0
+ β1PROFITi,t + β2TANGi,t + β3CRISIS + β4CRISIS*PROFITi,t+ β5CRISIS*TANGi,t + β 6SIZEi,t + β7MTBi,t + β8LIQUIDi,t +
μi + εi,t; This model includes the interaction terms between CRISIS and PROFIT, TANG. The coefficients of the parameters
are presented where *, **, *** means significant at 10%, 5% or 1% respectively. The standard errors are presented in parentheses. The definitions of the variables are explained in Appendix A.
Model Fixed effects regression
Equation 1 Equation 2 Equation 3
Dependent variable LEV Intercept 0.3871*** (0.0573) 0.5775*** (0.0604) 0.5913*** (0.0621) Independent variable PROFIT -0.1610*** (0.0380) -0.1894*** (0.0375) -0.2175*** (0.0434) TANG 0.0255 (0.0368) -0.0174 (0.0366) -0.0185 (0.0393) SIZE -0.0098* (0.0052) -0.0295*** (0.0056) -0.0306*** (0.0056) MTB -0.0166*** (0.0024) -0.0102*** (0.0025) -0.0095*** (0.0026) LIQUID -0.1060*** (0.0260) -0.1400*** (0.0259) -0.1404*** (0.0259) Dummy variable CRISIS - 0.0276*** (0.0031) 0.0182** (0.0076) CRISISPROFIT - - 0.0543 (0.0400) CRISISTANG - - 0.0050 (0.0118)
N observations 2,733 2,733 2,733
N groups 382 382 382
R2 overall 0.1633 0.1160 0.1131
R2adj 0.1414 0.1144 0.1150
F-test 27.99*** 37.01*** 28.01***
Table III shows the results of the fixed effect regression on three equations. In the first column, I show the results on the regression for model 1: LEVi,t = β0 + β1PROFITi,t + β2TANGi,t + β3SIZEi,t +
β4MTBi,t + β5LIQUIDi,t + μi + εi,t; In the second column, I show the results for the regression on
model 2: LEVi,t = β0 + β1PROFITi,t + β2TANGi,t + β3 CRISIS + β4SIZEi,t + β5MTBi,t + β6LIQUIDi,t + μi
+ εi,t; I included the dummy CRISIS in this model to estimate the effect of the financial crisis on LEV.
In the third column, I show the results for the regression on model 3: LEVi,t = β0 + β1PROFITi,t +
β2TANGi,t + β3CRISIS + β4CRISIS*PROFITi,t+ β5CRISIS*TANGi,t + β 6SIZEi,t + β7MTBi,t +
β8LIQUIDi,t + μi + εi,t; I include the interaction variables between CRISIS and PROFIT, TANG in this
model to estimate the effect of the crisis on PROFIT and TANG.
The coefficient of PROFIT is significant and negative in all the models, which is consistent with the findings in previous research (Thies & Klock, 1988; Rajan & Zingales, 1995; Fama & French, 2002; De Jong et al., 2008). This result in line with the pecking order theory, since profitable firms use their retained earnings as internal finance source for investments before they issue debt (De Jong et al., 2008).
I find an insignificant result for the coefficient of TANG in each of the models. Ignoring the significance, only the first column shows the expected result. The positive relationship between TANG and LEV is consistent with the trade-off and pecking order theory since tangible assets reduce the costs of financial distress, and the physical assets can serve as collateral for debt (Frank & Goyal, 2005). However, column 2 and 3 show a negative coefficient which is not consistent with prior research (Titman & Wessels, 1988; Mishkin, 1990; Rajan & Zingales, 1995, De Jong et al., 2008).
The coefficient of SIZE is negative and significant in all models. This is not in line with the results of previous studies (Rajan & Zingales, 1995; Fama & French, 2002; De Jong et al, 2008; Akhtar, 2012). This result also contradicts the expectations based on the trade-off and pecking order theory, since it is expected that the greater diversity and smaller volatility in earnings in large firms reduce the costs of financial distress and mitigate the asymmetric information problem, which increases the issuance of debt (Fama & French, 2002; De Jong et al., 2008).
I find a significant negative coefficient for MTB in all models, which is consistent with the trade-off theory and previous studies (Rajan & Zingales, 1995; Deesomsak et al., 2004; De Jong et al., 2008). As expected this result shows that firms with a high market-to-book value tend to invest in risky projects and therefore face higher costs of financial distress, thus use less debt. Additionally,
their share price is high for which these firms prefer to issue equity instead of debt (Rajan & Zingales, 1995).
LIQUID shows a significant and negative coefficient in each of the models, which is in line with the results in prior research (Deesomsak et al., 2004; De Jong et al., 2008). As expected by the pecking order theory firms can use their liquid assets as an internal finance source, before debt or equity is issued (De Jong et al., 2008).
I find a positive coefficient on CRISIS in model 2 and 3. This results is also shown in the research of Alves and Francisco (2013), Fosberg (2012) and Iqbal and Kume (2014). They show an increase in the leverage ratio of firms during a crisis.
The coefficient of the interaction variables, CRISISPROFIT and CRISISTANG, are both insignificant. This result means that the 2008 financial crisis did not exerted any influence on PROFIT or TANG as capital determinants. Ignoring for sake the insignificant result on CRISISPROFIT, the coefficient is positive. This finding suggests that during the financial crisis the negative effect of a firms’ profitability on debt weakens, which is in line with Akhtar (2012) and Iqbal and Kume (2014). As expected, during the crisis the availability of internal finance sources decreased, because of
decreased profitability. However, firms still need to finance their operations during the crisis and issue debt or equity. This indicates the decrease in the influence of profitability on the capital structure choice. If I ignore the insignificance of CRISISTANG, the coefficient is positive. These findings suggest that during the crisis the influence of tangible assets on debt increased. During the financial crisis, the adverse selection problem increases. To mitigate this problem tangible assets become more valuable since they can serve as collateral for debt (Mishkin, 1990; Mishkin, 2011). This finding is in line with the stronger impact of tangibility in the crisis period found by Deesomsak et al. (2004), Iqbal and Kume (2014) and Akhtar (2012).
6. Sensitivity Analysis
To estimate the influence of firm-specific determinants on the capital structure, various measures for leverage can be used. In this research, I choose to define the leverage ratio as the long-term debt ratio since this is also used in previous studies (Cheng & Jiang, 2001; Akhtar, 2012; Alves & Francisco, 2013). On the contrary, the total debt ratio is used by Deesomsak et al. (2004) to estimate the effect of a crisis as well as by Iqbal and Kume (2014). I did choose to not include short-term debt, since De Jong et al. (2008) state the determinants of short-term debt issuance differ from those for long-term debt and therefore results are difficult to interpret.
However, to guarantee that my results are driven by the used definition of LEV, I define LEV as the total debt over the total book value of assets and perform a regression on the model. The results for LEV measured as the total debt ratio are corresponding to results where LEV is defined as the long-term debt ratio. Additionally, I define LEV as the market value of long-term debt and find insignificant results for TANG. Furthermore, CRISIS shows a negative impact on the leverage ratio, which is not in line with previous studies (Fosberg, 2012; Iqbal & Kume, 2014).
Secondly, I investigate whether my results are similar if I use different measures for the independent variables SIZE and TANG. To check for sensitive results, I regress the model where SIZE is defined as the natural logarithm of total sales, since this measure is often used to indicate firm size (Rajan & Zingales, 1995; Alves & Francisco, 2013; Harrison & Widjaja, 2014). However, the results are corresponding the results where SIZE is measured by the natural logarithm of total assets. I also regress the model where TANG is defined as the net fixed assets, which is used by Iqbal and Kume (2014) and De Jong et al. (2008). This regression shows a multicollinearity problem. For this reason, net fixed assets are not an appropriate measure to estimate TANG.
Finally, I investigate whether the random effects regression model shows similar results. This regression model is used in the research of Harrison and Widjaja (2014). The results are
corresponding to the results using a fixed effects regression for my model. In addition, I also investigate the random effects model regressing my model on the period 2004-2007 and the period 2008-2011 separately as in the study of Harrison and Widjaja (2014). I exclude the dummy and interaction variables in the model. However, the results show an insignificant coefficient on PROF, TANG and SIZE and I cannot conclude whether the crisis had a significant effect on PROF or TANG.
In conclusion, the measures I use in this regression for leverage, profitability, tangibility, size, market-to-book value and liquidity are appropriate measures, since other computations of these measures used in similar studies give similar or less significant results. This also applies to the fixed effect regression I use to estimate my model.
7. Conclusion
In this research, I investigate the impact of the financial crisis of 2008 on the capital structure determinants of S&P500 listed firms in the United States. Specifically, I aim to identify whether two firm-specific capital determinants, profitability and tangibility, are affected by the financial crisis and what their contribution is to the leverage ratio.
To answer this question, I use a fixed effect regression on panel data over the period 2004 to 2011. The panel data includes data in the sample period for 382 active firms listed on the S&P500. I use the leverage ratio to estimate a firms’ capital structure and I include five firm-specific
determinants in the regression model: profitability, tangibility, firm size, market-to-book ratio and liquidity. The financial crisis and after crisis period refers to the period of 2008 until 2011, therefore I include a dummy variable CRISIS which investigates the overall influence of the financial crisis on the leverage ratio. Additionally, I include two interaction variables for profitability and tangibility with the crisis dummy variable to estimate the influence of the crisis on both parameters.
The results of this research show a negative relation between profitability and leverage, supporting the pecking order theory, but contradicting the trade-off theory. The findings show that the 2008 financial crisis did not influenced the impact of profitability as a capital determinant. The impact of profitability on leverage is negative, but there is no significant evidence that this impact becomes less negative during the financial crisis. This is not as expected, since I expected that the overall decrease in profits of firms during the crisis caused firms to issue more debt and thus the negative impact of profitability to become less negative. For this reason, I conclude that the results for
profitability are only partly corresponding to the first formulated hypothesis. There are no significant results for tangibility as capital structure determinant or for the influence of crisis on tangibility. For this reason, I conclude that the results are not corresponding to the second formulated hypothesis. In conclusion, this study did not find significant evidence for the influence of the financial crisis in 2008 on profitability and tangibility as capital structure determinants for listed firms in the United States. Further research is essential to investigate the capital structure determinants during the financial crisis. The results of this research are sensitive, since I tested the sensitivity of my results for other measures of leverage, size and tangibility and for another regression model. The measures and regression model used in comparable studies give similar or less significant results.
As in previous research, there are some limitations and recommendations for further research. The first limitation of this research is the number of independent variables and the relatively low R-squared values. To explain more of the variance in the long-term leverage ratio I recommend to include additional independent variables, such as business risk, corporate taxation, median of leverage, or income volatility. Furthermore, I recommend to account for the supply side factors of debt. Secondly, I only examine the effects of the financial crisis on profitability, tangibility and leverage. Consequently, also the effect of the financial crisis on the other firms-specific capital determinants should be analyzed. Thirdly, to study the effect of the financial crisis more accurately,
also a post- or pre-crisis dummy should be included. Therefore, the effect of the financial crisis can be studied over the before, after and during crisis period. Lastly, I only investigate one country, one crisis and a small sample period. To investigate whether a crisis in general affects capital structure
determinants I recommend to study multiple crisis periods and various countries and compare these results.
References
Alves, P., & Francisco, P. (2013). The impact of the institutional environment in firms’ capital structure during the recent financial crises. Retrieved on May 30, 2018 from
https://mpra.ub.uni-muenchen.de/51300/
Akhtar, S., (2012). Capital structure and business cycles. Accounting and Finance, 52(1), 25-48.
Bradley, M., Jarrell, G., & Kim, E. (1984). On the existence of an optimal capital structure: Theory and evidence. The journal of Finance, 39(3), 857-878.
De Jong, A., Kabir, R., & Nguyen, T. T. (2008). Capital structure around the world: The roles of firm- and country-specific determinants. Journal of Banking & Finance, 32(9), 1954-1969.
Deesomsak, R., Paudyal, K., & Pescetto, G. (2004). The determinants of capital structure: evidence from the Asia Pacific region. Journal of multinational financial management, 14(4), 387-405.
Degryse, H., De Goeij, P., & Kappert, P. (2012). The impact of firm and industry characteristics on small firms' capital structure. Small business economics, 38(4), 431-447
Downes, J., & Goodman, J. (2014). Dictionary of finance and investment terms (9). New York, United States of America: Barron’s Educational Series.
Fama, E., & French, K. (2002). Testing trade-off and pecking order predictions about dividends and debt. Review of financial studies, 15(1), 1-33.
Fosberg, R. (2012). Capital structure and the financial crisis. Journal of Finance and Accountancy, 11, 1.
Frank, M., & Goyal, V. (2005). Trade-off and pecking order theories of debt. Handbook of empirical corporate finance, 2, 135-202.
Frank, M. & Goyal, V. (2009). Capital structure decisions: which factors are reliably important?. Financial management, 38(1), 1-37
Harrison, B., & Widjaja, T. (2014). The determinants of capital structure: Comparison between before and after financial crisis. Economic Issues, 14(2), 55-82.
Hsiao, C. (2014). Analysis of panel data (3). New York, United States of America: Cambridge University Press.
Iqbal, A., & Kume, O. (2014). Impact of Financial Crisis on Firms’ Capital Structure in UK, France, and Germany. Multinational Finance Journal, 18(3/4), 249-280.
Kraus, A., & Litzenberger, R. (1973). A state-preference model of optimal financial leverage. The journal of finance, 28(4), 911-922.
Lee, C., & Lee, A. (2006). Encyclopedia of finance (1). New York, United States of America: Springer.
Lemmon, M., Roberts, M., & Zender, J. (2008). Back to the beginning: persistence and the cross-section of corporate capital structure. The Journal of Finance, 63(4), 1575-1608.
Michaelas, N., Chittenden, F., & Poutziouris, P. (1999). Financial Policy and Capital Structure Choice in U.K. SMEs: Empirical Evidence from Company Panel Data. Small business economics, 12(2),113-130.
Mishkin, F. (1990). Asymmetric information and financial crises: a historical perspective. Nber working paper series, 3400, 1-38.
Mishkin, F. (2011). Over the cliff: from the subprime to the global financial crisis. Journal of economic perspectives, 25(1), 49-70.
Modigliani, F., & Miller, M. (1958). The cost of capital, corporation finance and the theory of investment. The American economic review, 48(3), 261-297.
Myers, S. (1984). The capital structure puzzle. The journal of finance, 39(3), 574-592.
Myers, S. (2001). Capital structure. Journal of economic perspectives, 15(2), 81-102.
Myers, S., & Majluf, N. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of financial economics, 13(2), 187-221.
Rajan, R., & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. The journal of Finance, 50(5), 1421-1460.
Robichek, A., & Myers, S. (1966). Problems in the theory of optimal capital structure. The Journal of Financial and Quantitative Analysis, 1(2), 1-35.
Singh, P., & Kumar, B., (2012). Trade-off theory vs pecking order theory revisited. Journal of Emerging Market Finance, 11(2), 145-159.
Thies, C., & Klock, M., (1992), Determinants of Capital Structure. Review of Financial Economics, 1(2), 40-52.
Titman, S., & Wessels, R. (1988). The determinants of capital structure choice. The Journal of finance, 43(1), 1- 19.
Torres-Reyna, O. (2007, December). Panel data analysis fixed and random effects using stata. Retrieved on June 4, 2018, from https://www.princeton.edu/~otorres/Panel101.pdf
Wald, J. (1999). How firm characteristics affect capital structure: an international comparison. Journal of Financial research, 22(2), 161-187.
Yardeni, E., & Abbot, J. (2018, June 19). Stock market briefing: S&P 500 sectors & industries profit margins. Retrieved June 22, 2018, from https://www.yardeni.com/pub/sp500margin.pdf
Appendix A;
The appendix contains the definitions on the variables used in my analysis. Furthermore, the calculations for the variables and the abbreviations in the Compustat Database are included.
Formula Compustat
Dependent variable
LEV: long-term book leverage Total long-term debt / book value of total assets
DLTT/AT
Independent variables
PROFIT: profitability Earnings before interest, tax, depreciation and amortization / book value of total assets
EBITDA/AT
TANG: tangibility Plant, property and equipment /
book value of total assets
PPENT/AT
Control variables
SIZE: size Ln (Total Assets) Ln (AT)
MTB: Market-to-book value Market value of equity / book value of assets
MKVALT/AT
LIQUID: Liquidity Current assets / book value of
total assets
ACT/AT
Dummy variable
No crisis (CRISIS = 0) STATA: Year ≤ 2007 Data 2004 - 2007
Crisis (CRISIS = 1) STATA: Year > 2007 Data 2008 - 2011
Interaction variables
CRISISPROFIT STATA: Crisis * Profitability
Appendix B;
S&P 500 Profit Margin
Reprinted from “Stock market briefing: S&P 500 sectors & industries profit margins”, by Yardeni, E., & Abbott, J. (2018, June 19). Retrieved from https://www.yardeni.com/pub/sp500margin.pdf
