Determinants of Capital Structure across Industries

(1)

1

Determinants of Capital Structure across

Industries

An Empirical Analysis of Capital Structure Determinants in the

S&P 500 for 2000-2010

Thesis

Master of Science in Business Administration Specialisation: Finance

University of Groningen

Faculty of Economics and Business

Author: Maarten Weijermars Student number: 1948334 Supervisor: dr. P.P.M. Smid

(2)

2

Abstract

Previous research on capital structure has focused primarily on the many different determinants. However, there is little research about the influence of the determinants on capital structure across industries. The goal of this research is to test various hypotheses about the determinants of capital structure and to establish whether and how the relationship of these determinants varies across industries. Leverage was found to be positively related to non-debt tax shields and negatively to profitability, tangibility, size, growth opportunities and volatility. Significant differences across industries were found in all determinants of capital structure.

JEL classification

G32, C23

Keywords

Capital structure determinants, Leverage and Industry differences.

(3)

3

Preface

The last six months I have been writing my master thesis as a final assignment of completing my Master of Finance at the University of Groningen. Recent years the stock markets have experienced tremendous financial shocks. Managers of large enterprises in all industries have been criticized of taking on unnecessary high levels of debt.

This has led me to the question how the determinants of capital structure actually influence capital structure and if the effects of these determinants differ across industries. In this master thesis I will try to answer these questions and test if these results hold for 500 companies listed on the S&P 500 in the United States.

I would like to thank my supervisor dr. P.P.M. Smid for his support and advice during the entire process. In addition I would like to thank my parents and brother for their support during my studies.

(4)

4

Table of content

1. Introduction ... 5

2. Literature review ... 7

2.1 Fundamental theories ... 7

2.2 Determinants of capital structure ... 10

2.3 Effects of determinants of capital structure across industries ... 17

(5)

5

1.

Since the propositions of Modigliani and Miller (1958; 1963) on capital structure there has been an important and on-going debate in the corporate finance literature about the optimal capital structure. Modigliani and Miller (1958) showed that in a perfect market, capital structure would be irrelevant to the value of the firm. In a world with corporate taxes and deductibility of interest payments, the firm value would be maximised by using as much debt as possible (Modigliani and Miller (1963)). An optimal capital structure allows a firm to achieve lower financing costs and consequently maximize shareholder value (Sheel (1994)). Much of the existing research about capital structure has advanced theoretical models to explain the determinants of capital structure and has provided empirical evidence whether these theoretical models have explanatory power when applied to the real world. Given that a company’s choice of capital structure can significantly affect its cost of capital, an enhanced understanding of the determinants of capital structure should enable managers to maximize shareholder wealth. Two important alternative theories about capital structure have been developed. These theories are the static trade-off and the pecking order theory. The trade-off theory specifies that firms within the same industry should have similar capital structures since they have comparable types of assets, business risks and profitability. The pecking order theory specifies that the various industries experience different business environments and consequently, such circumstances can cause differences in the capital structure. Therefore, the purpose of this research is to provide further empirical evidence on the effect of industries on capital structure.

(6)

6 The main research questions in this study are as follows:

1. How do the main determinants dictated by literature influence capital structure? 2. Do the effects of these determinants of capital structure vary across industries?

To test the research questions, multiple regressions are employed. An OLS regression is used to analyse the effect of the selected determinants on capital structure, an unrestricted pooled OLS regression is employed to analyse whether the determinants differ across industries and a F-test is employed to compare the unrestricted OLS regression with seven restricted OLS regressions to test if the differences of the variables vary significant across industries.

Financial panel data from companies included in the S&P 500 index is collected. The S&P 500 is of particular interest for three reasons. First, the American companies in de S&P 500 cover about 75% of the U.S. equity market by capitalization1. Second, most existing studies on capital structure have focused on Anglo Saxon countries making the results of this study more comparable with former research. Third, looking only at quoted companies included in the S&P 500 allows for a relative clean test and avoids the problem of having to control for possible institutional differences between countries.

The presented research contributes to the literature in several ways. This paper shows several mixed results provided by former research. The added value of this study to the existing literature in the field of capital structure is that this study adds new knowledge about the effect and influence of the determinants across industries.

The remainder of this paper is organized as followed. Section 2 provides an outline of the theoretical framework and relevant literature and serves as a background for this research. Section 3 explains the chosen methodology; section 4 discusses the data collected. Section 5 will discuss the results and finally section 6 will provide a conclusion.

(7)

7

2.

In this section the relevant theoretical background is presented. It consists of three sections. Section 2.1 will discuss the two fundamental theories underlying capital structure. Section 2.2 will discuss the determinants of capital structure and section 2.3 will end up with the effect of determinants of capital structure across industries.

2.1 Fundamental theories

Research on the modern theory of capital structure was initiated by the important work of Modigliani and Miller (1958;1963). Modigliani and Miller demonstrated that if a company’s investment policy is taken as given, then in a perfect world, a world without taxes, perfect and credible disclosure of all information and no transaction costs associated with raising money or going bankrupt, the extent of debt in a company’s capital structure does not affect the ﬁrm’s value. After the pioneering work of Modigliani and Miller (1958) on capital structure, two important alternative theories about capital structure have been developed. These theories are the static trade-off and the pecking order theory.

Static trade-off theory

(8)

8 agency costs that can arise when firms are financed with debt. The first agency cost is referred to as the “asset substitution problem”. Debt provides an incentive for managers acting on behalf of equity holders to take on unnecessary risk, substituting low risk investments for high risk investments. Equity holders can realize an unlimited upside, but in the event of an unfavourable outcome, they can do no worse than lose their entire investment because they have limited liability. The payoff to the equity holders will be either the firm’s cash flows less its debt obligation (when cash flows from assets exceed the debt obligation), or zero (when the debt obligation exceeds the cash flows) (Hillier, Grinblatt and Titman (2008)). The second agency cost is referred to as the “debt overhang problem”; this problem arises when the company’s debt becomes riskier. The debt overhang phenomenon leads to sub-optimal investments by equity holders in the form of investing in risky projects even when they are value decreasing for the firm. Conversely the effect works the same, when a company is likely to go bankrupt; there is no incentive for equity holders to invest in value increasing projects, because the largest part of the earned profit will go to the debt holders (Meyers (1977)). The third agency cost is called “milking the property”. Shareholders could distribute high ordinary dividends and even extraordinary dividends, by selling part of the assets in place (Vernimmen et al. (2005)).

(9)

9 The optimal capital structure balances the costs and benefits of debt and maximizes the value of the firm. Many researchers have tested the actual explanatory power of the trade-off theory (Huang and Song (2006); Booth et al. (2001); Pandey (2001)).

Pecking order theory

Gordon Donaldson (1961) found that managers prefer financing investments first with retained earnings, second, after the supply of retained earnings has been exhausted, with debt, and finally, when it is impossible for the firm to borrow additional amounts, outside equity is issued. This financing hierarchy is known as the pecking order of financing choices (Hillier, Grinblatt and Titman (2008)). Huang & Song (2006) state that the basic difference with the trade-off theory is that the pecking order theory suggests that there is no optimal capital structure and that possible capital structure differences arise when there are not enough internal funds available for investments. The pecking order theory uses different dependent variables that describe the flow of funds and investment opportunities, including free cash flow, capital expenditures, dividend pay-out and profitability (Huang & Song (2006)). There are multiple explanations for the pecking order of financing decisions, including the following ones:

(10)

10 the company’s stock is overvalued, which in turn implies that the stock price will fall when the company announces that it will issue new shares. Ross (1977) explains the information asymmetry as following: lower quality firms have higher marginal bankruptcy risk and consequently lower debt levels. In contrast, the high quality firms can hold higher debt levels due to the lower bankruptcy risk. Because these higher debt levels cannot been imitated by low quality firms, a strong signal is given to the stakeholders when issuing debt. The fourth and last explanation of the pecking order is that financial claimants may disagree about the attractiveness of issuing equity. If bankruptcy costs are borne primarily by the firm’s debt holders, the equity holders benefit little from an infusion of new equity. Share prices will decline when firms replace debt with equity because decreasing the firm’s leverage increases the value of existing debt and transfers wealth from the equity holders to the debt holders. An exception to this general rule occurs when reducing leverage significantly cuts the costs of financial distress and thus significantly increases the total value of the firm (Hillier, Grinblatt and Titman (2008)).

2.2 Determinants of capital structure

(11)

11 Table I

Overview of the results of previous empirical studies

Authors Sample Research

period Methodology Dependent variable(s) Total leverage Long-term leverage Short-term leverage Independent variable Kayo and Kimura

(2011) 17,061 companies from 40 different countries 1997-2007 HLM with a maximum likelihood Long-term debt ratio -*** -*** + +*** +*** Growth opportunities, Profitability, Bankruptcy distance, Size, Tangibility. Jong, Kabir and

Nguyen (2008)

11,845 companies from 42 different countries

1997-2001 OLS analysis Long-term debt ratio +*** -*** +*** +***/-*** -*** -*** +***/-*** Tangibility, Risk, Size, Tax, Growth opportunities, Profitability, Liquidity. Tang and Jang

(2007) 27 lodging companies and 27 software companies 1997-2003 OLS analysis, GLS analysis. Long-term debt ratio +*** +/-*** +/-*** +*** +*** -*** Profitability, Tangibility, Growth opportunities, Volatility, Size, Agency costs. Huang and Song

(2006)

1086 Chinese-listed companies

1994-2003 OLS analysis Tobit model

Total debt ratio, Long-term debt ratio. -*** +** -* + -*** -*** + -*** + -*** +*** -*** --*** --* + Profitability, Tangibility, Tax, Size,

Non-debt tax shields, Growth opportunities, Volatility, Managerial ownership, Ownership structure. Burferna, Bangassa and Hodgkinson (2005) 55 Libyan companies, public and private

1995-1999 OLS analysis Total debt ratio, Short-term debt ratio, Long-term debt ratio. +*** + -*** +*** -+ + + +*** + -*** +*** Profitability, Tangibility, Growth, Size. Chen (2004) 88 Chinese public-listed companies 1995-2000 Pooled OLS, Fixed effects, Random effects.

Total debt ratio, Long-term debt ratio. -*** +*** +** + --*** -* -+ +*** -+ Profitability, Size, Growth opportunities, Tangibility,

Cost of financial distress, Tax shields effects. Bevan and Danbolt (2002) 822 UK companies 1991 Cross-sectional regression analysis

Total debt ratio, Short-term debt ratio, Long-term debt ratio. -*** +** +*** --*** +*** +*** + -*** + +*** -Profitability, Tangibility, Size, Growth opportunities. Fama and French

(2002) 2844 World wide companies 1965-1997 Panel regression -*** -*** -*** -*** +*** Profitability, Tangibility, Growth opportunities, Non-debt tax shields, Size.

Huang and Song (2002)

1000 Chinese listed companies

1994-2000 OLS analysis Tobit model

Total debt ratio, Long-term debt ratio. -*** -+ -*** +*** +* +*** -*** +*** + -* +* + -Profitability, Tangibility, Size, Non-debt tax-shields, Growth opportunities, Volatility, Ownership structure. Sayilgan, Karabacak, kucukkocaoglu. (2002) 123 listed Turkish manufacturing firms

(12)

12 Table 1 shows a summary of previous empirical studies that relate to capital structure. Most of the studies used data from multiple years and performed an OLS analysis on the data. Previous studies used multiple ways to calculate the dependent variables. The most used dependent variables are total and long-term debt ratio. The results found also differ

Table I (continued)

Authors Sample Research period Methodology Dependent variable(s) Total leverage Long-term leverage Short-term leverage Independent variable Prasad and Green

(2001)

165 Malaysian companies and 174 Thai companies

Average period of about 5.5 years

OLS analysis Long-term debt ratio, Short-term debt ratio. -+**/-** -*** +/-** +** +*** -+*** +*** -** Profitability, Growth opportunities, Volatility, Size, Non-debt tax-shields. Pandey (2001) 106 Malaysian companies

1984-1999 Pooled OLS Total debt ratio, Short-term debt ratio, Long-term debt ratio. +*** --*** + +*** -*** +*** --*** -+** -+** --*** + +** -*** Growth opportunities, Investment opportunities, Profitability, Risk, Size, Tangibility. Booth et al. (2001) 727 companies from 10 developing countries 1980-1990 Cross-sectional regression

Total debt ratio, Long-term debt ratio. -*** -*** +*** -*** -*** --*** -*** +*** +*** -*** -Profitability, Growth opportunities, Size, Tangibility, Tax, Volatility. Mocnik (2001) 251 Slovene Manufacturing companies 1991-1996 Questionnaire, OLS analysis

Total debt ratio. -*** -** -+*** Profitability, Specific assets, Volatility, Number of employees. Hall, Hutchinson and Michaelas (2000)

3500 UK companies 1995 OLS analysis Long-term debt ratio, Short-term debt ratio. + + +*** +*** -*** -*** +*** -*** -*** -*** Profitability, Growth opportunities, Tangibility, Size, Age. Wiwattanakantang (1999)

270 Thai companies 1996 OLS analysis Total debt ratio. -*** + -*** -+*** -* Profitability, Tangibility, Non-debt tax-shields, Risk, Size, Growth opportunities. Shyam-Sunder and Myers (1999)

157 U.S. companies 1971-1989 OLS analysis Total debt ratio. -*** -+*** + Profitability, R&D, Tangibility, Tax. Balakrishnan and Fox (1993)

295 U.S companies 1978-1987 OLS analysis Total debt ratio. --* -* +*** +*** Volatility, R&D, Growth opportunities, Depreciation, Advertising. Titman and Wessels (1988)

469 U.S companies 1974-1982 OLS analysis Long-term debt ratio, Short-term debt ratio. -*** +*** -** +*** -*** -*** -*** +*** -** +*** -*** -*** Profitability, Tangibility, Uniqueness, Non-debt tax-shields, Growth opportunities, Volatility.

Bradley, Jarrell and Kim

(1984)

851 U.S companies, 25 industries.

1962-1981 OLS analysis Total debt ratio. +*** -*** -*** -*** Non-debt tax-shields, Volatility, R&D, Advertising. This table reports the main results of former previous empirical studies on capital structure. A + means that a positive coefficient is found and a - means

that a negative coefficient is found. *** indicates a significance level of 1%

(13)

13 much from each research. The summary of previous empirical studies serves as a motivation for the choice of variables used in this study. The most used significant determinants of capital structure in table I are: Profitability, tangibility, size, non-debt tax shield, growth opportunities and volatility. The effects of each variable on capital structure are described below.

Leverage

Bevan and Danbolt (2002) point out that capital structure studies examining the determinants of leverage based on total debt may disguise the significant differences between long-term and short-term debt. Therefore, in line with Bevan and Danbolt (2002), Burferna et al. (2005) and Pandey (2001), this study does not only research the total debt ratio but also decomposes the book value of debt into long- and short-term.

Profitability

(14)

14 Tangibility

The tangible assets of a firm can be considered as the representatives of the real guarantees to its creditors in the form of collateral. In their pioneering paper on agency costs, Jensen and Meckling (1976) point out that the agency cost of debt exists as the firm may shift to riskier investments after the issuance of debt, and transfer wealth from creditors to shareholders to exploit the option nature of equity (Huang and Song (2006)). If a firm’s tangible assets are high, then these assets can be used as collateral, diminishing the lender’s risk of suffering from agency costs of debt. Hence, a high fraction of tangible assets is expected to be associated with high leverage. The agency costs of equity leads to underinvestment problems. The information asymmetry is the reason that the new equity is under-priced. Issuing debt with tangible assets as collateral reduces these agency costs (Chen (2004)). A positive relationship between leverage and tangibility is consistent with both the trade-off model in terms of financial distress and bankruptcy costs and the pecking order theory in terms of asset mispricing (Chen (2004)). Some empirical studies that confirm the positive relation are: Kayo and Kimura (2011), Huang and Song (2006), Bevan and Danbolt (2002) and Shyam-Sunder and Myers (1999). In consideration of the theoretical arguments and the majority of the prior empirical evidence, the relationship between leverage and tangibility is expected to be positive.

Size

(15)

15 (Wiwattanakantang (1999); Panday (2001); Chen (2004)). This means that large firms should release more information to the public than smaller firms. Larger firms with less asymmetric information problems should tend to have more equity than debt and thus have a lower leverage (Huang and Song (2006)). However, larger firms are often more diversified and have more stable cash flow. The probability of bankruptcy for large firms is

smaller compared with smaller firms (Prasad and Green (2001)). Thus according to the

pecking order theory, leverage is negatively related to size. Booth et al. (2001) found evidence to support the negative relation between total leverage and size. In consideration of the theoretical arguments and the majority of the prior empirical evidence, the relationship between leverage and size is expected to be positive.

Non-debt tax shields

(16)

16 Growth opportunities

There is no consensus regarding the influence of growth opportunities on capital structure. The market-to-book ratio is adopted to capture the relative value of the growth opportunities viewed by the market. Market value reflects the expectation of the market on the current net worth of the company and the company’s future earnings. Since future earnings serve as the proxy for growth opportunities, the market-to-book ratio presents the current expectation of the company’s future growth opportunities to the book value (Tang and Jang (2007)). According to the static trade-off theory, firms holding future growth opportunities, which are a form of intangible assets, tend to borrow less than firms holding more tangible assets because growth opportunities cannot be collateralised. Kayo and Kimura (2011), De Jong, Kabir and Nguyen (2008), Huang and Song (2006) and Burferna et al. (2005) all found a negative relationship between growth opportunities and leverage. According to the pecking order theory, firms prefer debt before equity if the retained earnings have been exhausted. Rapidly growing firms would therefore have relatively more debt. According to the agency costs theory, firms with high growth opportunities are more likely to have high agency costs of debt due to the increased risk associated with high growth opportunities. Therefore, firms with higher growth opportunities are more likely to have more equity than debt financing. Tang and Jang (2007), Sayilgan, Karabacak, kucukkocaoglu (2006) and Chen (2004) all found a positive relationship between growth opportunities and leverage. So to conclude, a negative relationship would confirm the agency- and trade-off theory and a positive relationship would confirm the pecking order theory. In consideration of the theoretical arguments and the majority of the prior empirical evidence, the relationship between leverage and growth opportunities is expected to be negative.

Volatility

(17)

17 compensation for undertaking extra risk. In addition, creditors tend to avoid firms with high earnings volatility to reduce their risk exposure in loans. Therefore, the static trade-off theory expects a negative relationship between leverage and volatility. De Jong et al. (2008), Huang and Song (2006), Booth et al. (2001), Titman and Wessels (1988) and Bradley et al. (1984) all found a negative relationship between leverage and volatility. However,Hsia (1981) combines the option pricing model, the capital asset pricing model, and the Modigliani-Miller theorems to show that as the variance of the value of the firm’s assets increases, the systematic risk of equity decreases. Huang and Song (2002) also found a positive relationship between leverage and volatility of profits. Huang and Song (2006), Booth et al. (2001), Balakrishnan and Fox (1993) and Titman and Wessels (1984) found no significant relation between leverage and volatility. In consideration of the theoretical arguments and the majority of the prior empirical evidence, the relationship between leverage and volatility is expected to be negative.

2.3 Effects of determinants of capital structure across industries

(18)

18

Table II Hypotheses

H1,0: There is no relationship between leverage and profitability. H1,A: The relationship between leverage and profitability is negative. H2,0: There is no relationship between leverage and tangibility. H2,A: The relationship between leverage and tangibility is positive. H3,0: There is no relationship between leverage and size.

H3,A: The relationship between leverage and size is positive

H4,0: There is no relationship between leverage and non-debt tax shield. H4,A: The relationship between leverage and non-debt tax shield is negative. H5,0: There is no relationship between leverage and growth opportunities. H5,A: The relationship between leverage and growth opportunities is negative. H6,0: There is no relationship between leverage and volatility.

H6,A: The relationship between leverage and volatility is negative.

H7,0: The nature of the hypothesised relationships does not vary across industries. H7,A: The nature of the hypothesised relationships varies across industries.

The table summarizes the hypotheses for the dependent variables of capital structure and the hypothesis for the equal coefficient test.

(19)

19

3.

To measure whether the selected determinants have a significant effect on capital structure, a regression analysis is applied. In order to research the relationship between the six selected determinants and capital structures across industries, three regressions are performed similar to the work of De Jong, Kabir and Nguyen (2008), Talberg et al. (2008) and Hall et al. (2000). Namely, the restricted model, which analyses the effect of the selected determinants on capital structure, an unrestricted model, which analyses whether the determinants differ across industries and a F-test where the unrestricted model is compared to seven restricted models.

3.1 Restricted model

A regression model will be calculated to gain insight into the relationship between the six selected determinants and capital structure. The first question is whether a linear regression model is appropriate. This is done by looking at scatter diagrams and calculating a correlation matrix (Table V) of every independent and dependent variable. These results combined with the capital structure literature provided the basis to select the ordinary least squares (OLS) method. This resulted in regression 1.

(1)

Equation 1 is used to test H1-H6, where, TLEV_i_,_t stands for the dependent variable total leverage of firm i at the end of year t. Total leverage is defined as total debt scaled by total assets. There are six independent variables in the regression which are defined as follows: Profitability (PROFIT ) is defined as earnings before interest and taxes (EBIT) scaled by total assets. Tangibility (TANG) is defined as fixed assets scaled by total assets. Size is defined as the natural logarithm of sales. Non-debt tax shield (NDTS) is defined as the ratio of annual depreciation expense scaled by total assets. Growth opportunities (GROWTH ) is defined by the market to book value. Volatility (VOL) is defined as the standard deviation of earnings before interests and income taxes (EBIT) during the 3-year

(20)

20 period (12 quarters) prior to time t scaled by total assets. The symbol ₀ is a constant and

t i ,

 is the error term. The selection and definition of the dependent and independent variables is based on prior research on capital structure (see table I and the accompanying explanation). A complete overview of all variables, their definitions and sources are provided in Appendix A.

3.2 Unrestricted model

In order to gain insight into the relationship between the six selected determinants and capital structure across eight different industries classified by two-digit Standard Industrial Classification (SIC) codes, pooled regression 2 is employed.

(2)

in which TLEVi,t,PROFITi,t,TANGi,t,SIZEi,t,NDTSi,t,GROWTHi,t,VOLi,t, respectively

are the total leverage and firm-specific characteristics of firm i at the end of year t. D_jare the industry dummies. The unrestricted model is similar to the restricted model, but now the coefficients of the determinants of capital structure are freely to differ across eight different industries. The constant term is excluded from the regression to avoid perfect multicollinearity or the so called “dummy variable trap” (Brooks (2008)).

3.3 F-test

In order to determine whether there is a significant industry effect, an F-test is performed. Under the F-test, seven regressions are compared, one unrestricted with seven restricted regressions. The unrestricted regression is the one in which the coefficients are freely to differ across different industries. Six restricted regressions are the regressions in which the coefficients of one independent variable are restricted in every regression (Brooks

(21)

21 (2008)) and one restricted regression in which all variables are restricted. Establishing for which variables their impact on capital structure varies between industries is achieved through comparison of the residual sum of squares (RSS) of the unrestricted model with that of the seven restricted models. Six models omit all industry interaction variables of one independent variable and one model omits all industry interactions of all independent variables. The F-test is given by

Where T is the number of observations, mis the number of regressors omitted in the restricted models, k is the number of regressors in the unrestricted regression including

the intercept, and RRS and URSS are the residual sum of squares from the restricted and unrestricted models. The F-test is performed to test H7.

3.4 Robustness tests

To test the robustness of the results, three additional robustness checks are performed. The first robustness check tests the restricted model, unrestricted model and F-test with two different measures for the dependent variable leverage. The dependent variable leverage will be measured as long- and short-term debt scaled by total assets. The second robustness check tests the unrestricted model and F-test with a different industry classification. The firms will be classified with the general industry classification (GIC) which reports whether a company is an industrial, utility, transportation, banking, insurance or other financial company. The third and last robustness check tests regression 1 with time fixed effects.

(22)

22

4. 4.1 Sample selection

The dataset consists of financial information from firms listed on the S&P 500 index from 2000 to 2010 extracted from the DataStream database. The time period is selected because it covers the credit crisis (2008/2009) and still provides enough information about the long term mean of the capital structure of the different companies. The industry definitions from Standard Industrial Classification (SIC) are used to classify each company into an industry. SIC codes are four-digit codes used to classify businesses by the activity they are involved in, and the type of products and services they offer. By looking only at the first two digits of this code, the firms can be divided in eight different industries as presented in table III.

This results in 5500 yearly observations of 500 firms over 8 industries. The dataset consists of three dependent variables and six independent variables. The dependent variables are total leverage, long-term leverage and short-term leverage and the independent variables are profitability, tangibility, size, non-debt tax shield, growth opportunities and volatility. Book values are used to calculate the variables. A survey of Graham and Harvey (2001), reports that managers use book values as a basis to select the capital structure of the firm. The measure of all the variables is based on previous studies on the determinants of capital structure (table I). A complete overview of all variables,

Table III

Descriptive Statistics: Industry distribution S&P 500

SIC Industry No. of firms % of firms

in sample 10-14, 29 6% 15-17 5 1% 20-39 191 38% 40-49 71 14% 50-51 14 3% 52-59 42 8% 60-67 90 18% 70-89 58 12% Total 500 100% Wholesale Trade Retail Trade

Finance, Insurance, And Real Estate

Services

This table reports the number of observations per industry of companies listed on the S&P 500. Data sample contains all 500 firms over the period 2000 - 2010, classified by 2 digit SIC codes.

Mining Construction Manufacturing Transportation,

(23)

23 their definitions and sources is provided in Appendix A. The dataset is subjected to an outlier test. Basically, there are two different types of probable outlier: data errors and outliers which are legitimate. The first type should always be removed or corrected, but with the second type, removal is not always the best option. These outliers are corrected by the “mean plus two standard deviations method” (Field (2009)). This method is simply multiplying the standard deviation by two and adding it to the mean of the sample.

4.2 Descriptive statistics

(24)

23

Table IV

Descriptive Statistics: mean per industry

Nr. Industry TLEV LLEV SLEV PROFIT TANG SIZE NDTS GROWTH VOL

1 0.314 0.286 0.021 0.102 0.563 14.86 0.075 2.711 0.056 2 0.361 0.296 0.065 0.067 0.519 15.63 0.016 1.824 0.060 3 0.262 0.215 0.043 0.107 0.517 15.58 0.037 3.938 0.043 4 0.394 0.334 0.050 0.070 0.340 15.79 0.042 2.164 0.025 5 0.192 0.172 0.020 0.131 0.566 15.91 0.029 3.163 0.024 6 0.222 0.183 0.033 0.144 0.558 16.09 0.051 3.720 0.036 7 0.346 0.239 0.091 0.062 0.303 15.21 0.015 2.514 0.019 8 0.222 0.184 0.028 0.114 0.595 14.63 0.032 4.706 0.051 Mining Wholesale Trade Retail Trade Construction Manufacturing Transportation, Communications, Electric, Gas, And Sanitary Services

Finance, Insurance, And Real Estate

Services

This table reports the mean statistics for the following dependent and independent variables per industry. Total leverage, TLEV, is defined as total debt scaled by total assets. Long-term leverage, LLEV, is defined as long-term debt scaled by total assets. Short-term leverage, SLEV, is

defined as short-term debt scaled by total assets. Profitability, PROFIT, is defined as earnings before interest, taxes scaled by total assets. Tangibility, TANG, is defined as fixed assets scaled by total assets. Firm size, SIZE, is defined as the natural logarithm of total assets. Net debt

tax-shield, NDTS, is defined as depreciation scaled by total assets. Growth opportunities, GROWTH, is defined as the mark et to book value and volatility, Vol, defined by the standard deviation of earnings before interests and income taxes (EBIT) during the 3-year period (12 quarters) prior to time t. Data sample contains all 500 firms listed on the S&P500 in the period 2000 - 2010 and is classified by 2 digit SIC

(25)

24 In order to understand and describe the main features of the data-set in quantitative terms, descriptive statistics and the normality test is analyzed. Table V shows the descriptive statistics of the dependent and independent variables.

Table V includes the mean, median, standard deviation, minimum, maximum and Jarque-Bera of the variables. It is easily seen how the variables are represented in the data set. The variable GROWTH in the sample varies considerable with an average of 3.42 and a maximum of 21.24. All variables in the sample exhibit strong variations as shown by the standard deviations which are with the exception of SIZE close to the mean. Another noteworthy characteristic is the average SLEV of 4.4% indicating that the firms in the sample do not rely strongly on short-term debt financing. Table V also shows that the Jarque-Bera test is significant at the 1% significant level for all variables, indicating that the distribution of the sample is significant different from a normal distribution. Brooks (2008) mentions that for a large sample size, violation of the normality assumption is virtually inconsequential because of the central limit theorem2. Without outliers, the sample size consists out of 4818 observations, which is enough reason to believe that a deviation from normality in this study is not enough to bias the results of the regression analysis. In a model where the goal is to understand how the different independent

2_{The central limit theorem states conditions under which the mean of a sufficiently large number of} independent random variables, will be approximately normally distributed.

Table V

Descriptive statistics of the collected sample

Variables Mean value Median Standard deviation Minimum Maximum TLEV 0.265 0.250 0.182 0.000 1.149 6377 *** LLEV 0.221 0.199 0.169 0.000 1.047 1355 *** SLEV 0.044 0.022 0.060 0.000 0.360 7366 *** PROFIT 0.101 0.093 0.082 -0.258 0.386 6856 *** TANG 0.474 0.431 0.275 -0.068 1.789 21963 *** SIZE 15.493 15.514 1.327 6.735 18.634 242 *** NDTS _0.036 _0.033 _0.024 _0.000 _0.138 _{1488 ***} GROWTH 3.420 2.670 2.770 -3.160 21.240 288225 *** VOL 0.033 0.022 0.034 0.000 0.259 30359 ***

This table reports descriptive statistics for the dependent and independent variables. The descriptive statistics are corrected for data errors and legitimate outliers (N=4818).

*** indicates a significance level of 1% ** indicates a significance level of 5% * indicates a significance level of 10%

(26)

25 variables impact the dependent variables, multicollinearity is also a problem. Multicollinearity gives high coefficient standard errors and thus cause problems with statistical interpretations and makes hypothesis testing less conclusive (Brooks (2008)). Table VI reports the correlations between the dependent and independent variables. The highest observed correlations between the explanatory variables are found between SIZE and TANG (0.326) and between VOL and TANG (0.258). These correlations are sufficiently small that the presence of multicollinearity can be reasonably ignored (Brooks (2008)).

Table VI

Correlation data

Variables

TLEV LLEV SLEV TANG SIZE NDTS VOL

TLEV LLEV 0.906*** SLEV 0.368*** 0.067*** PROFIT -0.098*** -0.095*** -0.043*** TANG -0.265*** -0.242*** -0.241*** 0.219*** SIZE -0.058 -0.095*** 0.178*** 0.046*** -0.326*** NDTS 0.122*** 0.175*** -0.162*** 0.080*** 0.182*** -0.085*** GROWTH -0.102*** -0.092*** -0.010 0.093*** 0.093*** -0.016 -0.012 VOL -0.083*** -0.077*** -0.173*** -0.148 0.258*** -0.218*** 0.191*** -0.001 The sample consists of 4818 observations from the period 2000 - 2010. All variables in the sample show moderate correlation effect.

With a negative sign in front of the number meaning the variables are negatively correlated and a positive sign in front of the number means the variables are positively correlated.

(27)

26

5.

In this section, the results will be discussed. Every subsection will interpret the results of a different test. In section 5.1, the results of the restricted model which tested H1-H6, will be explained. Section 5.2 will provide the results of the unrestricted model; section 5.3 shows the results of the F-test which tested H7 and finally 5.4 will report the robustness checks of the results.

5.1 Restricted model

Table VII shows the result from regression 1. The restricted model provides insight into the relationship between the six selected determinants and total leverage. It shows that all independent variables are significant related to leverage. The adjusted R-squared value shows that 20.3% of the variable total leverage in the restricted model is explained by the variables in the sample.

The coefficient of PROFIT is negative at the 1% significant level. H1,0, is therefore

rejected, the relation between total leverage and profitability is negative. This result is

Table VII

Restricted model total leverage

Variable Coefficient t -Statistic

PROFIT _-0.097 _-2.589 _*** TANG _-0.272 _-24.085*** SIZE _-0.021 _-10.294_*** NDTS _1.317 _12.602_*** GROWTH _-0.003 _-3.060 _*** VOL _-0.626 _-7.911 _*** Intercept _0.713 _21.062_*** R-squared 0.203 Adjusted R-squared 0.202

Sum squared residuals 115.809

(1)

The sample consists of 4818 observations from the period 2000 - 2010.

(28)

27 supported by most of the prior research (see table I) and confirms the pecking order theory. The pecking order theory suggests that firms will use retained earnings first as investment funds and then move to bonds and new equity only if necessary. In this sense, profitability is an important capital structure determinant since it reflects the amount of earnings that may be available to invest. TANG also proves to be negatively related to TLEV at the 1% significant level. This effect of tangibility is odd, because it implies that firms tend to use less leverage as their fixed asset value increases. The negative relationship does not reject H2,0 and that is similar to the results of Pandey (2001), Booth

et al. (2001) and Sayilgan et al. (2002) who also expected a positive relation, but found a negative one between total leverage and tangibility. SIZE shows a small negative coefficient significant at the 1% significant level. The negative relationship does not reject H3,0 and is similar to the results of Booth et al. (2001) who also found a negative

relation between total leverage and size. The negative relation confirms the pecking order theory that explains the negative relation by lower informational asymmetries for large firms between insiders within a firm and capital markets. Larger firms with less asymmetric information problems should tend to have more equity than debt and thus have a lower leverage. NDTS shows a positive coefficient significant at the 1% significant level. The negative relationship does not reject H4,0 and is similar to the

results of Chen (2004), Titman and Wessels (1988) and Bradley et al. (1984) who also found a positive relationship between leverage and NDTS. This result confirms the theory of Graham (2005) who notes that firms with substantial NDTS often have substantial collateral assets which can be used to secure debt which is less risky than that which is unsecured. These firms will therefore tend to have higher leverage levels. If profitable firms invest heavily and also borrow to fund this investment, this can induce a positive relation between leverage and NDTS. GROWTH shows a small negative coefficient significant at the 1% significant level. The negative relationship rejects H5,0,

(29)

28 collateralized. VOL shows a negative coefficient significant at the 1% level. The negative relationship rejects H6,0, the relation between total leverage and volatility is negative.

This result is also found by De Jong et al. (2008), Huang and Song (2006), Booth et al. (2001), Titman and Wessels (1988) and Bradley et al. (1984) and confirms the trade-off theory. Higher earnings volatility causes greater uncertainty in the market and is a sign for the probability of financial distress, which signifies a higher risk to creditors. The creditors, therefore, will ask for higher compensation for undertaking extra risk. In addition, creditors tend to avoid firms with high earnings volatility to reduce their risk exposure in loans.

5.2 Unrestricted model

(30)

29

Table VIII

Unrestricted model total leverage

Variable Coefficient t -Statistic Variable Coefficient t -Statistic PROFIT Ind. 1 -0.194 -1.508 GROWTH Ind. 1 -0.005 -0.962

PROFIT Ind. 2 0.486 1.687* GROWTH Ind. 2 -0.022 -0.856

PROFIT Ind. 3 -0.033 -0.646 GROWTH Ind. 3 0.002 1.284

PROFIT Ind. 4 -0.529 -3.781*** GROWTH Ind. 4 -0.004 -0.981

PROFIT Ind. 6 -0.510 -3.949*** GROWTH Ind. 6 0.007 2.349**

PROFIT Ind. 8 0.117 1.361 GROWTH Ind. 8 -0.007 -2.828***

TANG Ind. 1 -0.076 -1.364 VOL Ind. 1 0.064 0.250

TANG Ind. 2 -0.009 -0.091 VOL Ind. 2 0.167 0.230

TANG Ind. 3 -0.285 -17.461*** VOL Ind. 3 -0.591 -5.628***

TANG Ind. 4 -0.270 -8.129*** VOL Ind. 4 -1.214 -5.515***

TANG Ind. 5 -0.155 -2.038** VOL Ind. 5 -0.413 -0.493

TANG Ind. 6 -0.244 -6.842*** VOL Ind. 6 -0.172 -0.577

TANG Ind. 7 -0.123 -3.223*** VOL Ind. 7 -0.818 -2.617***

TANG Ind. 8 -0.293 -10.247*** VOL Ind. 8 -0.061 -0.273

SIZE Ind. 1 -0.012 -1.605 Intercept Ind. 1 0.371 2.976***

SIZE Ind. 2 -0.166 -4.326*** Intercept Ind. 2 3.167 5.311***

SIZE Ind. 6 0.019 2.613*** Intercept Ind. 6 0.039 0.285

SIZE Ind. 7 -0.009 -1.730* Intercept Ind. 7 0.417 5.192***

NDTS Ind. 1 2.222 5.758*** NDTS Ind. 2 -12.652 -5.456*** NDTS Ind. 3 0.265 1.471 NDTS Ind. 4 1.953 6.839*** NDTS Ind. 5 2.659 3.320*** NDTS Ind. 6 0.748 2.077** NDTS Ind. 7 8.418 16.010*** NDTS Ind. 8 0.986 3.278*** R-squared 0.340 Adjusted R-squared 0.331 Sum squared residuals 95.991

(2)

The sample consists of 4818 observations from the period 2000 - 2010. The order of the industries is provided in Table IV.

(31)

30

5.3 F-test

Table IX shows the results of the F-test. The F-test determines whether the difference in magnitude and significance of the coefficients across industries is actually significant by comparing the residual sum of squares (RSS) of the unrestricted regression with seven restricted regressions.

The difference of the RSS between the unrestricted and the restricted model shows an F-statistic of 140.45, which is significant at the 1% level. The independent variables measured by the restricted model all differ significant at the 1% level from the unrestricted model. These significant results reject H7,0, the determinants of capital

structure vary across industries.

5.4 Robustness tests

The results of the robustness check on the measure of the dependent variable leverage are given in appendix B. It shows the results of the restricted, unrestricted and F-test with two different measures for the dependent variable leverage. The dependent variable leverage will be measured as long- and short-term debt scaled by total assets. The restricted models (Table I) show lower adjusted R-squared values (17.3% and 15.4%) than with the variable TLEV. This means that the robustness regression model has a lower explanatory power in explaining the dependent variables LLEV and SLEV. The

Table IX

F-test: comparing unrestricted model with seven restricted models

Total leverage model F Result

Dropping PROFIT dummies 7.553*** Profitability effect varies Dropping TANG dummies 5.502*** Tangibility effect varies

Dropping SIZE dummies 14.815*** Size effect varies

Dropping NDTS dummies 42.277*** Non-debt tax shield varies Dropping GROWTH dummies 3.144*** Growth opportunities varies Dropping VOL dummies 3.611*** Volatility effect varies

Dropping all dummies 140.446*** Industry effects vary

(totally restricted model)

Where m = 7; T = 4818; k = 56. The F-statistic follows the F distribution with m = 7, (T-k ) degrees of freedom: F (0.01)(7,4776) = 2.64

(32)

(33)

32

6.

(34)

33 research and is robust with respect to the measure of leverage and industry classification. Given that a company’s choice of capital structure can signiﬁcantly affect its cost of capital, an enhanced understanding of the determinants of capital structure should enable managers to maximize shareholder wealth. This study has given new knowledge to that understanding by researching the effect and importance of the determinants of capital structure across industries.

6.1 Limitations

There are particular limitations in this study that have to be considered. One limitation of this study is the measurement of the variables. The measurement of variables is based on the majority of prior research. It would be interesting to see if the results are different when for example market values or lagged variables are used. Another limitation is the different estimation methods and different classification of industries in the existing literature. It reduces the validity of all claims presented and makes it harder to compare results. A final limitation of this study is that it did not research a trend in the importance and relation of the determinants of capital structure over time.

6.2 Extensions

There is little research about the determinants of capital structure across industries. The existing research and this study found that the effect and influence from the determinants of capital structure vary across industries. This makes it interesting for further research. Researching the effect and influence from the determinants of capital structure across industries in other countries might confirm the results found and increase understanding.

(35)

34 Balakrishnan, Srinivasan, and Isaac Fox, 1993, Asset specificity, firm heterogeneity and capital structure, Strategic Management Journal 14, 3–16.

Bevan, Alan, and Jo Danbolt, 2002, Capital structure and its determinants in the UK- a decompositional analysis, Applied Financial Economics 12, 159–170.

Booth, Laurence, Varouj Aivazian, Asli Demirguc-Kunt, and Vojislav Maksimovic, 2001, Capital structures in developing countries, Journal of Finance 56, 87–130.

Bradley, Michael, Gregg A. Jarell, and Kim E. Han, 1984, On the existence of an optimal capital structure: theory and evidence, Journal of Finance 39, 857–878.

Brooks, Chris, 2008, Introductory economics for finance, 2nd edition (Cambridge

University Press).

Buferna, Fakher, Kenbata Bangassa, and Lynn Hodgkinson, 2005, Determinants of capital structure: evidence from Libya, Research paper series 2005/2008, The University

of Liverpool.

Chen, Jean, 2004, Determinants of capital structure of Chinese-listed companies, Journal

of Business Research 57, 1341–1351.

DeAngelo, Harry, and Ronald Masulis, 1980, Optimal capital structure under corporate and personal taxation, Journal of Financial Economics 8, 3–29.

Donaldson, Gordon, 1961, Corporate Debt Capacity, A Study of Corporate debt Policy

and the Determination of Corporate Debt Capacity (Boston: Harvard University,

(36)

35 Fama, Eugene F., and Kenneth R. French, 2002, Testing trade-off and pecking order predictions about dividends and debt, The Review of Financial Studies 15, 1–33.

Field, Andy, 2009, Discovering Statistics Using SPSS, 3rd edition, (Saga Publications

Limited).

Frank, Murray Z., and Vidhan K. Goyal, 2003, Testing the pecking order theory of capital structure, Journal of Financial Economics 67, 217–248.

Graham, Jhon R., 2005, Taxes and corporate finance (Handbook of Corporate Finance).

Graham, Jhon R., and Campbell R. Harvey, 2001, The theory and practice of corporate finance: Evidence from the field, The Journal of Financial Economics 60 (2-3), 187–243.

Grinblatt, Mark, and Sheridan Titman, 2001, Financial markets and corporate strategy,

2nd edition (McGraw-Hill).

Hall, Graham, Patrick Hutchinson, and Nicos Michaelas, 2000, Industry effects on the determinants of unquoted SMEs’ capital structure, The International Journal of the

Economics of Business 7 (3), 297–312.

Hillier, David, Mark Grinblatt, and Sheridan Titman, 2008, Financial markets and

corporate strategy, European edition (McGraw-Hill).

Hsia, Chi-Cheng, 1981, Coherence of the modern theories of finance, Financial Review, 27–42.

(37)

36 Huang, Samuel, and Frank Song, 2006, The determinants of capital structure: Evidence from China, China Economic Review 17, 14–36.

Huang, Samuel, and Frank Song, 2002, The determinants of capital structure: Evidence from China, Working Paper The University of Hong Kong.

Jensen, Michael C., and William H. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs and ownership structure, Journal of Financial Economics 3, 305– 360.

Jong de, Abe, Rezaul Kabir, Thuy Thu Nguyen, 2008, Capital structure around the World: The roles of firm- and country-specific determinants, Journal of Banking & Finance 32, 1954–1969.

Kayo, Eduardo, Herbert Kimura, 2011, Hierarchical determinants of capital structure,

Journal of Banking & Finance 35, 358–371.

MacKay, Peter, and Gordon Phillips, 2005, How does industry affect firm financial structure?, The Review of Financial Studies 18 (4), 1433–66.

Marsh, Paul, 1982, The choice between equity and debt: An empirical study, Journal of

Finance 37 (1), 121– 44.

Mocnik, Dijana, 2001, Asset specificity and a firm’s borrowing ability: An empirical analysis of manufacturing firms, Journal of Economic Behaviour & Organization 45, 69– 81.

(38)

37 Modigliani, Franco, and Merton H. Miller, 1963, Taxes and the cost of capital: a correction, American Economic Review 53 (3), 433–443.

Myers, Stewart C., 1977, Determinants of corporate borrowing, Journal of Financial

Economics 5 (2), 147–175.

Myers, Stewart C., 1984. The capital structure puzzle. The Journal of Finance 39, 575– 592.

Myers, Stewart C., 1990, Still Searching for Optimal Capital Structure, Journal of

Applied Corporate Finance 6 (1), 4–14.

Myers, Stewart C., and Nicholal S. Majluf, 1984, Corporate financing and investment decisions when firms have information that investors do not have , Journal of Financial

Economics 13 (2), 187–221.

Pandey, Mohan, 2001, Capital structure and the firm characteristics: Evidence from an emerging market, Working paper, Indian Institute of Management Ahmedabad.

Prasad, Sanjiva, Christopher Green, and Victor Murinde, 2001, Corporate financial structures in developing economics: Evidence from a comparative analysis of Thai and Malay companies, Working Paper, University of Birmingham.

Ross, Stephen A, 1977, The determination of financial structure: The incentive-signalling approach, The Bell Journal of Economics 8 (1), 23–40.

Sayilgan, Guven, Hankan Karabacak, and Guray Kucukkocaoglu, 2006, The firmspecific determinants of corporate capital structure: Evidence from Turkish panel data, Investment

(39)

38 Sheel, Atul, 1994, Determinants of capital structure choice and empirics on leverage behavior: A comparative analysis of hotel and manufacturing firms. Hospitality Research

Journal 17 (3), 3–16.

Shyam-Sunder Lakshmi, and Stewart C. Myers, 1999, Testing static trade off against pecking order models of capital structure, Journal Financial Economics 51, 219-224.

Talberg, Magnus, Christian Winge, Stein Frydenberg, and Sjur Westgaard, 2008, Capital Structure Across Industries, Journal of the Economics of Business 15 (2), 181–200.

Tang, Chun-Hung, and SoCheong Jang, 2007, Revisit to the determinants of capital structure: A comparison between lodging firms and software firms, Hospitality

Management 26, 175–187.

Titman, Sheridan, and Roberto Wessels, 1988, The determinants of capital structure choice, Journal of Finance 43, 1–19.

Vernimmen, Pierre, Pascal Quiry, Yann Le Fur, Maurizio Dallacchio, and Antonio Salvi, 2005, Corporate Finance: Theory and Practice, John Wiley & Sons.

(40)

I

Appendix A: Definition of variables

Table I

Definition, symbol and source of variables

Variable Symbol Definition DataStream code

1Leverage Lev D/TA See items 4 and 7

2Long-term leverage Llev LTD/TA See items 5 and 7

3Short-term leverage Slev STD/TA See items 6 and 7

4Total debt D Book value of debt* WC03255

5Long-term debt LTD Book value of long-term debt* WC03251

6Short-term debt STD Book value of short-term debt* WC03051

7Total assets TA Book value of total assets* DWTA

8EBIT EBIT Earnings before interest and taxes* DWEB

9Fixed assets FA Book value fixed assets* NTA

10Depreciation DEP Book value depreciation* WC04049

11Market to book value MTB Market to book value*

12Profitability Profit EBIT/TA See items 8 and 7

13Tangibility Tang FA/TA See items 9 and 7

14Size Size Ln(TA) See items 7

15Non-debt tax shield NDTS DEP/TA See items 10 and 7

16Growth opportunities Growth MTB See items 11

Volatility Vol (The standard deviation of EBIT during the

3-year period [12 quarters] prior to time t)/ TA

See items 8 and 7

(41)

II

Appendix B: Robust check of dependent variable

Table I Restricted models Variable Coefficient Coefficient PROFIT -0.095 -2.680*** -0.002 -0.157 TANG -0.221 -20.705*** -0.051 -13.122*** SIZE -0.024 -12.552*** 0.003 4.608*** NDTS 1.541 15.581*** -0.223 -6.236 *** GROWTH -0.004 -3.803*** 0.001 1.568 VOL -0.451 -6.024*** -0.175 -6.453 *** Intercept 0.683 21.321*** 0.030 2.604*** R-squared 0.174 0.148 Adjusted R-squared 0.173 0.145

Sum squared resid 103.639 13.204

LLEV SLEV

The sample consists of 4818 observations from the period 2000 - 2010. *** indicates a significance level of 1%

** indicates a significance level of 5% * indicates a significance level of 10%

t -Statistic t -Statistic t i t i t i t i t i t i t i t

i PROFIT TANG SIZE NDTS GROWTH VOL

LLEV, 01( ,)2( ,)3( ,)4( ,)5( ,)6( ,), t i t i t i t i t i t i t i t

i PROFIT TANG SIZE NDTS GROWTH VOL

(42)

III

Table II

Unrestricted model long-term leverage

Variable Coefficient t -Statistic Variable Coefficient t -Statistic PROFIT Ind. 1 -0.157 -1.307 GROWTH Ind. 1 -0.010 -1.806*

PROFIT Ind. 2 0.394 1.473 GROWTH Ind. 2 -0.022 -0.942

PROFIT Ind. 3 -0.085 -1.777* GROWTH Ind. 3 -0.001 -0.419

PROFIT Ind. 6 -0.379 -3.150*** GROWTH Ind. 6 0.000 0.000***

PROFIT Ind. 7 -0.425 -3.436*** GROWTH Ind. 7 -0.007 -2.144**

PROFIT Ind. 8 0.064 0.803 GROWTH Ind. 8 -0.007 -2.872***

TANG Ind. 1 -0.078 -1.508 VOL Ind. 1 0.187 0.782

TANG Ind. 2 -0.075 -0.788 VOL Ind. 2 0.403 0.597

TANG Ind. 3 -0.244 -16.074*** VOL Ind. 3 -0.425 -4.356***

TANG Ind. 4 -0.209 -6.775*** VOL Ind. 4 -0.985 -4.809***

TANG Ind. 5 -0.141 -2.001** VOL Ind. 5 -0.174 -0.224

TANG Ind. 6 -0.221 -6.675*** VOL Ind. 6 -0.509 -1.835*

TANG Ind. 7 -0.079 -2.233** VOL Ind. 7 -0.358 -1.232

TANG Ind. 8 -0.265 -9.973*** VOL Ind. 8 -0.064 -0.309

SIZE Ind. 1 -0.015 -2.163** Intercept Ind. 1 0.390 3.366***

SIZE Ind. 6 0.018 2.584*** Intercept Ind. 6 -0.007 -0.056

NDTS Ind. 1 2.322 6.473*** NDTS Ind. 2 -9.899 -4.592*** NDTS Ind. 3 0.183 1.091 NDTS Ind. 4 1.590 5.989*** NDTS Ind. 5 2.090 2.807*** NDTS Ind. 6 1.095 3.270*** NDTS Ind. 7 9.151 18.719*** NDTS Ind. 8 0.865 3.094*** R-squared 0.339 Adjusted R-squared 0.330 Sum squared residuals 82.967

(43)

IV

Table III

Unrestricted model short-term leverage

Variable Coefficient t -Statistic Variable Coefficient t -Statistic PROFIT Ind. 1 -0.038 -0.836 GROWTH Ind. 1 0.004 2.049**

PROFIT Ind. 2 0.092 0.909 GROWTH Ind. 2 0.001 0.056

PROFIT Ind. 3 0.051 2.876*** GROWTH Ind. 3 0.002 4.783***

PROFIT Ind. 4 0.058 1.181 GROWTH Ind. 4 -0.003 -2.421**

PROFIT Ind. 5 -0.018 -0.175 GROWTH Ind. 5 -0.002 -0.576

PROFIT Ind. 6 -0.132 -2.917*** GROWTH Ind. 6 0.000 0.000***

PROFIT Ind. 7 -0.100 -2.141** GROWTH Ind. 7 0.002 1.343

PROFIT Ind. 8 0.053 1.757* GROWTH Ind. 8 0.000 -0.453

TANG Ind. 1 0.002 0.108 VOL Ind. 1 -0.122 -1.363

TANG Ind. 2 0.066 1.836* VOL Ind. 2 -0.237 -0.931

TANG Ind. 3 -0.041 -7.196*** VOL Ind. 3 -0.166 -4.511***

TANG Ind. 4 -0.061 -5.232*** VOL Ind. 4 -0.230 -2.985***

TANG Ind. 5 -0.014 -0.508 VOL Ind. 5 -0.239 -0.814

TANG Ind. 6 -0.023 -1.819* VOL Ind. 6 0.337 3.228***

TANG Ind. 7 -0.044 -3.279*** VOL Ind. 7 -0.460 -4.208***

TANG Ind. 8 -0.028 -2.786*** VOL Ind. 8 0.003 0.038

SIZE Ind. 1 0.003 1.161 Intercept Ind. 1 -0.019 -0.439

SIZE Ind. 2 -0.004 -0.317 Intercept Ind. 2 0.148 0.709

SIZE Ind. 3 0.003 2.520** Intercept Ind. 3 0.006 0.291

SIZE Ind. 5 -0.001 -0.328 Intercept Ind. 5 0.046 0.620

SIZE Ind. 6 0.002 0.601 Intercept Ind. 6 0.047 0.963

SIZE Ind. 7 0.018 10.605*** Intercept Ind. 7 -0.163 -5.818***

SIZE Ind. 8 0.002 0.855 Intercept Ind. 8 0.008 0.232

NDTS Ind. 1 -0.100 -0.744 NDTS Ind. 2 -2.753 -3.394*** NDTS Ind. 3 0.082 1.305 NDTS Ind. 4 0.363 3.636*** NDTS Ind. 5 0.569 2.030** NDTS Ind. 6 -0.347 -2.753*** NDTS Ind. 7 -0.732 -3.981*** NDTS Ind. 8 0.121 1.150 R-squared 0.242 Adjusted R-squared 0.233 Sum squared residuals 11.748

(44)

V

Appendix C: Robust check of industry classification

Table IV

F-test: comparing unrestricted model with seven restricted models

Long-term leverage model F Result

Dropping NDTS dummies 53.684*** Non-debt tax shield varies Dropping GROWTH dummies 3.202*** Growth opportunities varies

Dropping VOL dummies 2.908*** Volatility effect varies

Short-term leverage model F Result

Dropping NDTS dummies 8.631*** Non-debt tax shield varies

Dropping GROWTH dummies 3.632*** Growth opportunities varies

Dropping VOL dummies 5.481*** Volatility effect varies

Table I

Descriptive Statistics: Industry distribution S&P 500

GIC Industry No. of firms % of firms

in sample 1 368 74% 2 47 9% 3 8 2% 4 24 5% 5 26 5% 6 27 5% Total 500 100% Other Financial

This table reports the number of observations per industry of companies listed on the S&P 500. Data sample contains all 500 firms over the period 2000 -

2010, classified by GIC codes.

Industrial Utility

(45)

VI

Table II

Unrestricted model total leverage

Variable Coefficient t -Statistic Variable Coefficient t -Statistic

Profitability Ind. 1 -0.036 -0.963 Growth opportunities Ind. 1 -0.003 -2.509**

Profitability Ind. 2 -0.220 -1.061 Growth opportunities Ind. 2 0.021 3.288***

Profitability Ind. 3 -0.313 -0.702 Growth opportunities Ind. 3 -0.007 -0.568

Profitability Ind. 4 3.472 3.817*** Growth opportunities Ind. 4 -0.006 -0.525

Profitability Ind. 5 -0.392 -0.116 Growth opportunities Ind. 5 0.018 0.141

Profitability Ind. 6 -1.229 -6.166*** Growth opportunities Ind. 6 0.011 2.006**

Tangibility Ind. 1 -0.273 -23.060*** Volatility Ind. 1 -0.381 -4.813***

Tangibility Ind. 2 0.140 2.037** Volatility Ind. 2 -0.050 -0.158

Tangibility Ind. 3 -0.494 -3.068*** Volatility Ind. 3 -0.456 -0.335

Tangibility Ind. 4 0.016 0.068 Volatility Ind. 4 0.005 0.002

Tangibility Ind. 5 0.520 0.471 Volatility Ind. 5 0.114 0.040

Tangibility Ind. 6 -0.196 -3.986*** Volatility Ind. 6 0.019 0.033

Size Ind. 1 -0.015 -6.987*** Intercept Ind. 1 0.621 16.551***

Size Ind. 2 -0.054 -6.724*** Intercept Ind. 2 1.196 9.033***

Size Ind. 3 -0.016 -0.468 Intercept Ind. 3 -0.199 -1.387

Size Ind. 4 0.025 3.114*** Intercept Ind. 4 1.394 0.154

Size Ind. 5 -0.084 -0.137 Intercept Ind. 5 0.755 1.363

Size Ind. 6 -0.010 -1.046 Intercept Ind. 6 0.471 3.064***

Non-debt tax shield Ind. 1 0.745 6.854***

Non-debt tax shield Ind. 2 0.397 0.789

Non-debt tax shield Ind. 3 -0.363 -0.335 Non-debt tax shield Ind. 4 -6.949 -1.211 Non-debt tax shield Ind. 5 -2.667 -0.050

Non-debt tax shield Ind. 6 9.152 11.487***

R-squared 0.318

Adjusted R-squared 0.312

Sum squared residuals 99.161

The sample consists of 4818 observations from the period 2000 - 2010. The order of the industries is provided in Table I. *** indicates a significance level of 1%

** indicates a significance level of 5% * indicates a significance level of 10%

t i t i j j t i j j t i j j t i j j t i j j t i j j j j t i VOL D GROWTH D NDTS D SIZE D TANG D PROFIT D D TLEV , , 6 6 1 , 5 6 1 , 4 6 1 , 3 6 1 , 2 6 1 , 1 6 1 6 1 0 , ) ( ) ( ) ( ) ( ) ( ) (                               Table III

F-test: comparing unrestricted model with seven restricted models.

Total leverage model F Result

Dropping NDTS dummies 24.886*** Non-debt tax shield varies Dropping GROWTH dummies 4.174*** Growth opportunities varies

Dropping VOL dummies 0.328 Volatility effect is constant

Dropping all dummies 160.368*** Industry effect vary