Master’s Thesis Finance The Stability of the Interest Coverage Ratio and its Relation with Enterprise Value

(1)

1

Master’s Thesis Finance

The Stability of the Interest Coverage Ratio and its Relation

with Enterprise Value

Elma Oltmans1

University of Groningen

June 2013

Abstract

This study conduct several tests in determining the stability of the interest coverage ratio of large and small firms and the influence of the enterprise multiple on the interest coverage ratio. Both variables are calculated by using two operational incomes; EBITDA and EBIT. I find statistical instability of the interest coverage ratio for small and large firms. However, the interest coverage ratio in the telecommunication, health care, technology and consumer goods is stable. The relationship and the correlation between the enterprise value and the interest coverage ratio are significant, but weak. The economic effects of the interest coverage ratio on the enterprise multiple is limited.

JEL classification: D22, E32, G32

Keywords: interest coverage ratio, enterprise value, determining stability, fluctuation interest coverage ratio, firm behavior

(2)

2

The Stability of the Interest Coverage Ratio and its Relation

with Enterprise Value

University of Groningen Faculty of Economics and Business

(3)

3

1. Introduction

“Careful design and management of a company’s capital structure do more to prevent value destruction than to boost value creation” (Koller, Goedhart and Wessels, 2010). Good care of the capital structure of a firm can save for bankruptcy in crisis periods, like the high-tech bubble in 2001 and the credit crisis which we are experiencing since 2007.

The risk of bankruptcy can be measured by a credit rating. Ratings are a useful summary indicator of capital structure health, meaning that the lower the rating, the higher the probability of default (Koller et al., 2010). The interest coverage ratio play an important role in the credit rating. Following Pettit et al. (2004) the size, the financial leverage and the interest coverage ratio of a firm have a significant effect on the credit rating. They find that these factors represent two third of the variation in the credit ratings across all industries. Yet, their model works best for ratings between A and B, however it also maintains reasonable predictive power for the other ratings. Damodaran (1999) also find a relation between credit ratings and financial ratios. He explained that rating agencies assess financial ratios, such as debt ratios and coverage ratios, to assign credit ratings. Koller et al. (2010) also study the influence of credit ratios on the credit rating of firms listed in the S&P with a market capitalization of more than one billion. They find that the credit ratios explained the credit rating “fairly well”, but the interest coverage ratio alone explains about 45%. This makes the interest coverage ratio an informative variable in determining the default risk, while the ratio measures the companies’ ability in meeting the short-term interest obligations.

(5)

5 this is only true when value is hardly related to the interest coverage ratios between five and 11 (Koller et al., 2010).

This research focuses on two topics. First, it will provide empirical evidence in the stability of the interest coverage ratio for large firms (> $1 billion) and small firms (< $1 billion). Second, the relation of the interest coverage ratio and the enterprise value of the firm will be researched. Based on these topics there are three hypotheses. The first hypothesis states that the interest coverage ratio for the large firms is stable, the second hypothesis state a stable interest coverage ratio for small firms, and the third hypothesis states that there is no relation between the size of the firm and the interest coverage ratio. The sample contains 2130 non-financial firms that are included in the Russell3000 with a data set from 1992 to 2012.

To tests the first two hypotheses, this study starts with an analysis that is comparative to the analysis of Koller et al. (2010). After explaining this analysis, this study will focus on statistical evidence. To control for stability, the ANOVA F-test, the Chi-Square tests (based on the median) and the Kruskal-Wallis test are performed. Three tests are used, to cover differences in the distribution. The third hypothesis is analyzed by the Spearman-rank correlation and will be graphically analysed with a scatterplot. Finally, a panel regression is conducted.

By analyzing the interest coverage ratio in stability three ranges are set up. The three ranges represent an interest coverage ratio smaller than five, interest coverage ratio between five and 11 (effective level) and the interest coverage ratio bigger than 11. Based on these ranges the distributions over time are calculated. This table shows that the distribution over the years are somewhat equal. Another test tries to find evidence how often firms go in and out of the effective level. This is presented with a transition probability matrix. This matrix shows that the majority of the firms with an interest coverage ratio in the effective range stays in the effective range for the following year.

(6)

6 interest coverage ratio when applying a significance level of five percent. By testing the stability of the interest coverage ratio into the effective level, only three industries have a statistical stable interest coverage ratio, which are the Telecommunication, Technology and the Consumer Goods industry.

The relationship between enterprise value and the interest coverage ratio shows that the interest coverage ratio does explain some part of the enterprise multiple, nevertheless it is a small part. The Spearman correlation coefficients, which are significant, also show a small negative relationship. These results are more or less consistent with the analysis of Koller et al. (2010). However they state that the relationship within the effective level is almost nihil, while the empirical evidence in this study shows a small significant negative relationship.

In conclusion, there is no statistical stability of the interest coverage ratio for the large and the small firms, and the interest coverage ratio has little economic value in explaining the enterprise multiple.

(7)

7

2. Discussion existing literature

This chapter provides a discussion of the existing literature which tries to answer the hypotheses of this study. The first part describes the meaning of the interest coverage ratio. The second part will focus on the underlying theory of the fluctuations in the interest coverage ratio. This part is focused on the capital structure, while the capital structure determines more or less the costs of debt, which influences the interest coverage ratio. These are important variables in determining the interest coverage ratios. The third part describes the stability and tries to answer how existing literature assess stability, and if the interest coverage ratio is expected to be stable for small and large firms. In the last part, the existing literature between the relation of the enterprise value and the interest coverage ratio will be outlined.

2.1. Interest coverage ratio

Simple financial ratios are used to judge the stability and solvency of a firm. Financial ratios allow managers to compare apples with apples, no matter what kind of company they are involved in and no matter the industry: they all tell the financial health of a firm (Magoon, 2008). More than half a century ago Merwin (1942) study 939 US firms and concluded that firms in difficulties experience different ratio measurements than continuing firms. He concluded that ratio analysis can be useful in the detection of operational and financial difficulties of a firm. Horigan (1965) study a similar topic and find that the long-term solvency ratios are important ratios in determining bankruptcy. One of these long-term solvency ratios is the ‘times-interest-earned-ratio’, also known as the interest coverage ratio.

(8)

8 while they are already overleveraged. Milgram et al. (1999) argued that when the interest coverage ratio is low or falls, investors are warned of “possible cash flow problems”. A decrease in the interest coverage ratio may result in non-payment of e.g. dividends, because the firm has to meet its legal obligations first (Dheeriya, 1999). When the interest coverage ratio increases, there are disadvantages for the firm as well. These are discussed in the next session where decisions about the capital structure of the firm are outlined.

2.2. Theory of capital structure

The capital structure, and more specific ‘the amount of debt’, plays a huge role in calculating the interest coverage ratio. The fluctuations in debt will influence the stability of the interest coverage ratio. This part explains the reason of determining the amount of debt and the empirical evidence in finding the capital structure. Although academic researchers search for an optimal capital structure for years, they still did not develop a final model which can help in deciding a firm’s capital structure. However, existing literature does describe theories in determining an optimal capital structure. Barclay and Smith (1999) defined this optimal capital structure as a ratio that creates the most value for shareholders. Barclay and Smith (1999) noticed that there is evidence that leverage would bring key benefits in the form of reductions in taxes and avoidance of overinvestment. However, it is also associated with costs arising from business erosion and conflicts of interest among investors. This is also discussed by Titman and Wessels (1988), who noticed that the capital structure depend on factors that determine various costs and benefits that are associated with the financing of the firm.

(9)

9 their earlier article. In this paper Modigliani and Miller (1963) state that taxes do influence the capital structure. They assumed that the interest (cost of debt) is tax deductible while the cost of equity is not. Managers should now have a preference to finance with debt, because the tax advantages create value for the firm. The value of the firm is now equal to the sum of the total operating assets and the tax shield.

According to the corrected theory of Modigliani and Miller, firms should finance with debt to profit from the tax shield. But the trade-off theory shows that there are limits of financing with debt. The classical version of this theory is from Kraus and Litzenberger (1973). They argue the perfect balance of capital structure, while relatively much debt creates benefits of tax savings, but (relatively) too much debt causes costs of bankruptcy. In the trade off theory there is some optimal level of leverage. When the determinants of this optimal level of leverage (like the expected profitability, volatility of profitability, assets specificity, and so on) are more or less constant, you might expect that leverage ratios and hence coverage ratios are more or less stable.

(10)

10 believe the equity is overpriced. This assumption of the investors will cause a drop in the share price. Barcley and Litzenberger (1988), Jung, Kim and Stulz (1996) and many others find empirical evidence of this share price drop after the announcement of issuing new equity. Shangguan and Vasudevan (2008) find the same effect twenty years later. They concluded that the stock price reaction to the announcement of equity offering is negative.

Important empirical evidence is the preference of internal financing of firms, even when the firms are not in the target capital structure. Byoun (2008) studied when and how firms adjust their capital structure towards their targets. Byoun (2008) finds that the most adjustments occurs when the firms have a leverage ratio that is higher than the target capital structure plus having a financial surplus, or when firms have a leverage that is below the target capital structure with a financial deficit. These empirical studies suggested that firms move toward the target capital structure when they face financial deficit of a financial surplus.

2.3. Determining stability

An important question is why financial ratios are tested for stability. According to Bushman (2007) an analysis of financial ratios can have valuable information to investors and external users, who must determine the financial stability of an organization. This is important because the stability represents the soundness, dependability and efficiency of the business. The interest coverage ratio is one of the important ratios, because this ratio is used in determining the ‘margin of safety’ of a firm in the ability to repay interest payments in a certain period.

Existing literature researched the stability of financial ratios. One of the researchers is Pinches, et al. (1973). They study long-term stability of 48 financial ratios of 221 US industrial firms for the years 1951-1969. They find that one group of financial ratios, financial leverage ratios, is the most stable group. The instable ratios were the short-term liquidity ratios. These results are consistent with Johnson (1979), who studied 61 financial ratios over 465 firms, representing retailers and manufactures over the period 1972 to 1974.

(11)

11 analysis is employed to isolate independent patterns of financial ratios. The analysis employs financial ratios as variables and industrial firms as the cases, which in turn produces factor patterns of the financial ratios in terms of industrial firms. The similarity of each variable in the reduced space with the factors is measured by its factor loading; which is simply the correlation of an original variable with a factor. Johnson (1979) also conducts the R-factor analysis to find the similarity of each ratio, which is indicated by the factor loading (correlation). This analysis result in several financial ratio groups. Each group contains financial ratios that correlate with each other. Further Johnson (1979) tests the stability statistical by using t-tests. These are performed to assess the existence and significance of differences among the averages of the ratios between 1972 and 1974.

While the financial ratios are studied in their stability, the interest coverage ratio specifically has not been studied in stability so far. Herefore, this study also searches for existing literature in the stability of the credit rating, since the interest coverage ratio represents around 45% of the credit rating (Koller et al., 2010). For example Pettit (2004), who study continuously S&P-rated and US-listed companies between 1993 and 2003. He finds that non-financials have relatively stable levels of credit ratings, Debt/EBITDA and Debt/Enterprise Value. Pettit (2004) determine the stability by using t-statistics between the years 1993 and 2003. Altman (2004) explains the stability of credit ratings. He states that the rating agencies have chosen to focus on the long term perspective. This in turn lowers the sensitivity of agency ratings to short-term fluctuations in credit quality. The objective of agencies is to provide “an accurate relative and ordinal ranking of credit risk without reference to an explicit time horizon”. These findings and arguments lead to the conclusion that there are only adjustments in the credit ratings when there are significant changes in credit quality (Cantor and Mann, 2003). This is confirmed by the Standard and Poor’s corporate ratings criteria document. They explain that the value of the credit rating is greatest when their rating focus is on the long term and does not fluctuate with near term performance. This is also preferred by the investors, who want to keep their portfolio rebalancing as low as possible (Ellis, 1998).

(12)

12 interest coverage ratio, which is based on the leverage and the operational income, instability should, according to existing theory and empirical evidence in the stability of the leverage, be caused by fluctuations in the operational income, because existing literature state a stable capital structure.

This study hypotheses first that the interest coverage ratios for large firms are stable in the effective level and second that the interest coverage ratio is stable for small firms in the effective level. These hypotheses are motivated by Koller et al. (2010) by their claim that the enterprise value is hardly affected by the coverage ratio when the coverage ratio is in the effective level. This suggests that size does not influence the stability of the interest coverage ratio.

2.4. Relation enterprise value with interest coverage ratio

This part use existing literature in determining the relation between the enterprise value and the interest coverage ratio. In explaining this relation the interest coverage ratio is sometimes adopted as a substitute for leverage, while leverage measures the ability to cover its interest payments on the long-term and the interest coverage ratio measures the ability to cover its interest payments on the short term (Koller et al., 2010).

The already above mentioned Modigliani and Miller (1963) study the optimal balance of debt to equity. They attempted to measure the effect of leverage on value of a firm which explains that up to a certain point the tax advantages of debt add value, but after a point the costs associated with bankruptcy reduce the value of a firm. Modigliani and Miller presented this in a graph with on the x-axis the leverage and on the y-axis the enterprise value. This study is interested in the interest coverage ratio, so I interpret the leverage as the interest coverage ratio. Figure 1 presents the relation between the interest coverage ratio and the enterprise value, based on theory. In the middle of the graph the optimal interest coverage ratio is presented based on the enterprise value.

(13)

13 In theory, the optimal pattern of leverage will differ between companies, depending on their characteristics (Koller et al., 2010). He argued that the higher a firms return, the lower its growth and business risk, and the more fungible its assets and capabilities, the more highly it should be leveraged. These firms have more tax benefits, while their income is stable. This is also the case for the interest coverage ratio, because the interest coverage ratio already includes the operational income, it should be, comparing with leverage, a more volatile ratio.

Figure 1 Relation Enterprise Value and Interest Coverage Ratio

Figure 1 presents the determination of enterprise value, based on the interest coverage ratio. The lower the interest coverage ratio, the lower the enterprise value, because of e.g. business erosion. The higher the interest coverage ratio, the less tax advantages, the lower the enterprise value. The original versions of Modigliani and Miller (1963 and 1958) have instead of the interest coverage ratio the leverage on the x-axis.

Firms with relatively high fluctuations in the operational income should have a relatively high interest coverage ratio, to cover the fluctuations and thereby the risk of overleverage. Binsbergen, et al. studied 126,611 firm-year observations and found that the costs of being overleveraged is higher than the costs of being underleveraged. Firms being overleveraged can be identified by the interest coverage ratio. When the theory of Binsbergen et al. (2010) is true, the enterprise value, presented in figure 1, should be lower at the left side than at the right side of the black solid line.

(14)

(15)

15

3. Data and methodology

This chapter discusses the data and the methodology that this study uses. First the dataset is outlined, second the variables are presented. Third, the treatment of outliers will be explained. In the fourth part, the methodology is presented.

3.1. Database

The constituents of the Russell 3000 index are used in the database of this study. Russell3000 is an index that measures the performance of 3000 U.S. companies that are approximately representing 98% of the investable U.S equity market. This index is constructed to provide a comprehensive and stable barometer of the market and is completely reconstituted annually to guarantee new and growing equities. The exact constituent list is from 19 March 2013 and included 2951 firms. Thompson DataStream is used to collect annual financial information of the Russell 3000. The most essential financial information that has been collected is the interest expense, enterprise value and operational income (the EBIT and the EBITDA). The control variables which are used in the regression, like the variables including in the ROIC and the industry, are collected as well.

Table 1 Overview selection data

This table presents the selection based on the original dataset. This study includes 2130 firms over 21 years, that is from 1992 till 2012.

Russell 3000 Selection Number of firms

Full constituent list 2951

No information about the firm 104

Less than one interest coverage ratio 117

Financials 600

Final 2130

(16)

16

3.2. Variable definition

Table 2 presents the variables, their symbols, the data source and/or the method of construction.

Table 2 Overview symbols of raw input

This table presents an overview and description of the symbols that are used. These are directly from Thompson Datastream. This raw input is used in calculating the variables defined in table 3.

Symbol Variable name

Source (All fields from Thompson DataStream) EBITit Earnings before interest and

taxes Field 18191

EBITDAit Earnings before interest, taxes,

depreciation and amortization Field 18198

EVit Enterprise Value Field 18100 (calculation contains

the market value equity + total debt + preferred shares + minority interest and minus cash) Iit Interest expense on debt Field 01251

NIit Net income Field 07250

RTit Reported taxes Field 01451

MT Marginal tax rate Stable at 35% (justified by OECD)

Dit Net Debt Field 18199 (calculation contains

long and short term debt, minus available cash)

Eit Total Equity Field 03995 (calculation contains

the value of the common stock plus preferred stock)

∑

Industry dummy Field 04070

(17)

17 Table 3 Overview Symbols of calculated variables

This table presents an overview of the calculations which are conducted in finding the results. The variables used in these calculations can be find in table 2.

Symbol Variable name Calculation

IOit Interest and other financial

income

IOit = EBIT – Net Income - Taxes

OTit Operational tax rate

ROICit Return on invested capital

Cit Interest coverage ratio based _{on EBITDA}

C'it Interest coverage ratio based _{on EBIT}

Mit Enterprise multiple based on _EBITDA

M'it Enterprise multiple based on _EBIT

3.3. Treatment of outliers

In this study an outlier exists because a firm can have a one-time negative operational income or the firm can have very low (even negative) debt. Existing literature also eliminates outliers, like Foster (1986) and Ditmann and Weiner (2005).

Only from the enterprise multiple and the interest coverage ratio outliers are deleted, because a certain number of the multiples and ratios can be expressed as an extreme which can bias the conclusions. Grubbs (1987) defined an outlier as data that appears to deviate clearly from other members of the sample in which it occurs. Table 4 presents the descriptive statistics from before deleting the outliers and after deleting the outliers. As presented in the table the median hardly changes. More descriptive statistics (after deleting the outliers) are presented in appendix A.1.

(18)

18 positive income. Rearranging the negative observations, by setting the negative observations to zero, would be an option. Yet, the distribution and statistical results would be biased. Because there are enough data observations, the negative observations are deleted. This is also done by Foster (1986). He deleted all negative EBIT’s when calculating average profit before taxes to tangible net worth ratios.

Next to deleting the negative ratios the interest coverage ratios and enterprise multiples above 30 are also eliminated from the dataset. The firms with an interest coverage ratio and enterprise multiple in the range of 0 to 30 represent the sample best. Removing multiples that are higher than 30 is in line with rules stated on the Damodaran website. This will provide the dataset with a possible upward bias. But following Nenide, Pricer and Camp (2003) it must be remembered that each financial statement deletion weakens the representativeness of the sample, although it is authorized when the adjustments need to be made to make the dataset more meaningful.

3.4. Methodology

This part describes the procedure for testing the hypothesis and will explain the methodology that this study will use in performing the empirical tests. This section first discusses the methodology in determining stability. The second part focuses on the relationship between the enterprise multiple and the interest coverage ratio. While there are no previous studies that study the same research questions, comparative studies are used to remain consistent in measuring stability and the relationship.

3.4.1. Determining stability of the interest coverage ratio

(19)

19 rating of A+ till BBB- is stable, the interest coverage ratio in the effective level should be stable as well.

In analyzing the stability of the interest coverage ratio, this study first will provide a histogram and a distribution of the dataset. The distribution is presented in a table that provides an overview per segment (that is per industry, size and year). These distributions are based on three groups, ranged from a group which contains the number of observation of the interest coverage ratio smaller than five, from five till 11 and higher than 11.

Other tests to determine stability of the interest coverage ratio are calculated by the ANOVA F-test, the Kruskal-Wallis test and the Chi-Square tests based on medians. Existing literature studies financial ratios by using the factor or component analysis. Similarity between the years is tested by using the two sample t-test (i.e. Pinches et al., 1973). Another way of determining stability is by calculating the congruency coefficient (Devine and Seaton, 1995). The congruency coefficient is, according to Devine and Seaton (1995), a measure of stability related to financial ratios. They argued that congruency coefficients greater than .95 indicate considerable agreement between factors. However, this study cannot complete this calculation, while I am not in the possession of the software. And while the t-test only focus in testing the equality of two variables, this study will focus on the ANOVA F-test. The ANOVA F-test tests under the null hypothesis the equality over the means per year, in where each year will be compared to the year that follows. If the test statistic is significant, the means of the years differ, which suggest instability. Following Keller (2008) the ANOVA F-test is calculated based on the mean square for treatments and the mean square for errors2_{. Furthermore,}

the ANOVA F-test is based on the assumption of a normal distribution of the residuals. As can be found in appendix A.1, the distribution of the interest coverage ratio per year is presented. These distributions are not normal. The descriptive statistics shows per year the amount of firms included, the mean, the first quartile, the median, the third quartile, the standard deviation, the minimum, the maximum, the Jarque-Bera test statistic and the corresponding probability of normality. The test of normality of Jarque-Bera shows that for every year there is a non-normal distribution. This influences the test results of the ANOVA F-test. However, problems might arise when the distribution of the coverage ratios for each industry (in each year) is non-normal, although there might be other relevant

(20)

20 classifications. Therefore I use a parametric as well as a non-parametric test, to avoid stressing the normality issue. For this reason, the Kruskal-Wallis test is conducted, which is applied to make sure that possible non-normality does not bias the results (Brooks, 2008). Keller (2008) argued that the data can be either ordinal or interval to perform the Kruskal-Wallis test, which tests the similarity of the population locations. The null and alternative hypotheses for this test are similar to the hypothesis specified earlier. If the null hypothesis is true, the ranks should be evenly distributed among the groups.

The Chi-Square test for the median is used as a third test in testing for differences of the interest coverage ratios across years. This test is used to determine whether there is enough evidence to infer stability and tests if differences exist between the expected frequencies and the observed frequencies in one or more categories (Keller, 2008). In fact, the Chi-Square test that is based on a median is a rank-based ANOVA F-test based on the comparison of the number of observations above and below the overall median in each subgroup. Conover (1980) described this test as the median test. The Chi-Squared distribution is a special case of the gamma distribution. This is the reason that this tests is also included in this study, because the sample shows a gamma distribution, which is a positively skewed distribution. This study also tests the stability of the firms in the effective interest coverage ratio. First the same statistic tests as pronounced above are conducted to test the stability of the firms with a median interest coverage ratio between five and 11, the effective level. Another test that is performed tries to contain information about the times a firm opts out the effective level. First the firms are set in a range, that is equal to the ranged used earlier. This contains three ranges, the first range has an interest coverage ratio lower than five, the second range is the effective level and the third range contains the firms with an interest coverage ratio higher than 11. Based on these ranges the transition probability matrix is calculated.

3.4.2. Relationship between interest coverage ratio and enterprise value

(21)

21 Koller et al. (2010) also try to explain the relation between the interest coverage ratio and the enterprise multiple, however they based this only on the tax shield. They found a relation that was relatively stable within the range five till 11. But when the interest coverage ratio was lower than five there was a negative relation with the enterprise multiple. To show this relation this study will provide a scatter plot. Also a correlation factor will be calculated by doing the Spearman’s rank correlation test. In statistics, this coefficient assess how strong the relationship is between the interest coverage ratio and the enterprise multiple. A perfect Spearman correlation lies between -1 or +1. This occurs when each of the variables is a perfect monotone function of the other.

The above analysis is incomplete to determine whether there is a relation or not. To explain whether the interest coverage ratio has significant influence on the enterprise multiple a dated panel regression will be performed. A dated panel comprises of both time series elements and cross-sectional elements. An important feature of a dated panel regression is that it keeps the same individuals or objects and it measures some quantity about them over time (Brooks, 2008). Following Brooks there are important advantages in conducting a panel data analysis, in comparison with a simple OLS. It is easier to tackle more complex problems, and because of the cross section and time series data together the power of the test increases, because the number of degrees of freedom increases. Third, by structuring the model in an appropriate way, it is possible to remove the impact of certain forms of omitted variable bias in regression results.

(22)

22 suggests that the Random Effects model is inconsistent. This is when the Chi-Square statistic is large. When this is the case, the fixed effects model is used (Hausman (1978), Brooks (2008). In this study the Hausman model shows that the random fixed effect model is not sufficient under the five percent significance level. For this reason the fixed effect model is used in this study.

This regression use several control variables are added to increase the explanatory power and in order to prevent that this study is relating enterprise value to coverage, whereas the underlying explanation for the variation in enterprise value is due to some other variable. The control variables are the return of invested capital (ROIC) and nine industry dummies. The ROIC is included because this is a variable of the key value driver formula (Koller et al., 2010), which explains a part in the valuation of a firm. He find that firms with a higher ROIC and a higher growth are valued more highly in the stock market; these firms have a higher firm value. In this study the growth variable is also considered to act as a control variable, however this was correlated with the enterprise multiple, while both variables included the market value of the firm. The industry dummies are inserted to see which industries have a significant value in explaining the enterprise multiple, while different industries can present different intercepts and different slopes. This should increase the explaining power of the variables in the regression. The dummy variable trap will be kept into mind, to avoid perfect multicollinearity. This is done by excluding one industry dummy. To perform the panel dated regression formula three and four are used. In this formula the subscript i and t stands for respectively the corresponding firm and time measured in years.

∑ (3)

∑ (4)

Where,

Mit = The enterprise multiple, Mit is based on the EBITDA and M’it is

based on the EBIT

Cit = The interest coverage ratio, Cit is based on EBITDA and C’it is

based on EBIT

(23)

23

3.5. Descriptive statistics

The descriptive statistics of the enterprise multiple and the interest coverage ratio over the whole period are presented in table five. In appendix A.1 the descriptives per year of the interest coverage ratio can be found. The descriptive statistics show, with the Jarque-Bera test, that the variables are not normally distributed.

Table 4 Descriptive Statistics

This table presents the descriptive statistics of the enterprise multiple and the interest coverage ratio of both variants, the EBITDA and EBIT. Panel A presents the basic descriptive statistics per variable, and panel B presents the average of the variable per industry.

Descriptive Statistics Enterprise multiple (EBITDA) Enterprise multiple (EBIT) Interest coverage ratio (EBITDA) Interest coverage ratio (EBIT) Panel A: descriptive statistics

Average 9.68 12.68 8.87 7.33 Median 8.54 11.71 6.68 4.9 Maximum 29.98 29.99 30 30 Minimum 0.02 0.01 0.00 0.00 Standard deviation 4.94 5.51 6.84 6.66 Skewness 1.39 0.78 1.13 1.4 Kurtosis 5.3 3.48 3.55 4.34 Jarque-Bera 13690 2492 4707 8324 Probability JB 0.00 0.00 0.00 0.00 Observations 25217 22562 20836 20642

Panel B: average of the variables per industry

Oil & Gas 8.69 12.48 9.25 7.16

Basic Materials _8.95 _12.16 _8.64 _6.75 Industrials 9.25 12.36 9.37 7.69 Consumer Goods 9.25 11.74 8.80 7.49 Health Care _11.50 _13.63 _10.71 _9.52 Consumer Services 9.58 12.85 8.44 6.92 Telecommunication 8.45 13.25 6.03 4.08 Utilities _8.67 _12.78 _4.63 _3.14 Technology 11.86 14.10 11.24 10.13

(24)

(25)

25

4. Results

In this chapter the results will be discussed. In the first part the data is presented and analyzed and the statistical tests will be discussed. In the second part the relation between the interest coverage ratio and the enterprise multiple will be discussed. The third part describes sensitivity tests.

4.1. Analyzing stability of the interest coverage ratio

Figure 2 presents the fluctuation in the interest coverage ratio of the whole sample, based on the first quartile, the median and the third quartile. In this figure there are shocks visible, which are apparently sudden changes in the interest coverage ratio. These changes correlate with economic downturns; the internet bubble around 2001 and the financial crisis around 2008. The first quartile seems to be less sensitive than the third quartile. This could be caused by an already high cost of debt in firms of the first quartile that was included before the economic downturn a high risk premium. The development of the interest coverage ratio based on EBIT is presented in appendix A.2. Between these two graphs there is high similarity. The differences between these figures are small.

Figure 2 The distribution of the interest coverage ratio (EBITDA)

This figure presents the median, the first quartile and the third quartile of the interest coverage ratio for all the firms in the database. The upper line represents the third quartile, the middle line the median and the lowest line the first quartile.

Figure 2 show that the median interest coverage ratio over the most years ranged around the five till seven, which is also shown by the descriptive statistics.

The distribution of the interest coverage ratios is presented in figure 2. This figure shows that the number of interest coverage ratios is highest between the three and

(26)

26 five, if the interest coverage ratio is based on EBITDA. When the interest coverage ratio is based on EBIT the amount of interest coverage ratio’s between three and four is highest. In theory this is normal because the operational income is divided by the interest expenses. Meaning that the higher the operational income, the higher the interest coverage ratio.

There needs to be noticed that the number of interest coverage ratios between these two histograms in figure 3 differs. The histogram based on EBITDA includes 20836 interest coverage ratios and the histogram based on EBIT includes 20642 interest coverage ratios. In appendix A.3 the distribution of the enterprise multiple is presented.

Figure 3 Distribution of the interest coverage ratio

This figure presents the distribution of both variants of the interest coverage ratio, EBIT and EBITDA. The Black bold line is based on EBIT and the grey bold line is based on EBITDA.

Koller et al. (2010) present that 72% of the credit ratings are in the rating categories A+ to BBB-. While the majority was in this group he states and assumes, also based on existing literature, that credit ratings are stable over time. He concludes that firms probably do not move in and out of the range A+ to BBB-. Further he translated this range to an interest coverage ratio between five and 11. This suggests that the majority of the interest coverage ratio should be in this range as well. But figure 3 shows that this is not true in this study. Table 6 present will present more about the amount of firms in this range.

(27)

27 Table 5 Overview distribution interest coverage ratio per segment This table presents the distribution of the interest coverage ratio per segment. Three groups are presented, which are the interest coverage ratios lower than five, between five

and 11 and higher than 11. The middle group represents the effective level; firms should not move out of this level, while there are disadvantages in the outside groups. The large firms represents firms with a market capitalization of more than one billion, and the small firms represent firms with a market capitalization smaller than one billion.

Percentage of observations

per range per subgroup Total

Interest coverage EBITDA ranges in percentage

< 5 Effective _level > 11

All firms 20836 37.19 33.42 29.39

Large firms 12668 33.14 35.14 31.73

Small firms 8168 43.46 30.77 25.77

Oil & Gas 1658 33.47 35.22 31.30

(28)

28 Table 6 presents the percentage firms which have an interest coverage ratio in a certain range per size (small/large), per industry and per year of the dataset. The distribution of the interest coverage ratio is based on three different ranges, the interest coverage ratio till five, from five till 11 and from 11. The middle range represents the effective level. The table present that over the year-observations of the large firms only 35% represent the effective level. This is lower than the rate of 72% which Koller et al. (2010) presented. Small firms show a lower percentage of firms in the effective level, namely 30.8%. Over the years the interest coverage ratio-percentage of firms in the effective level shows that there are small downturns visible in the middle of the period, but overall, the share of interest coverage ratio in the effective level seems to shuffle around the one third, with an average of 33.8% and a standard deviation of 2.27. From this point of view, it could be said that it is fairly stable. In appendix A.4 the distribution of the interest coverage ratio per segment in percentages for the interest coverage ratio based on EBIT is presented.

Further tests are conducted to find stability by statistical tests. Table 6 presents the results of the stability of the interest coverage ratio of the whole sample. This table shows the test results of three different tests. Based on the test statistics the null hypothesis is rejected. Every test statistic for all three segments shows significant differences across the years. By rejecting the null hypothesis, I find that the interest coverage ratio is not stable, for the large and small firms, or when all firms are tested together.

Table 6 Statistical Results Stability

This table presents the test results about the stability of the interest coverage ratio. Observations are tested for stability by the ANOVA F-test, Kruskal-Wallis test and Chi-Square test, by using the 1% significance (***), 5% significance (**) and 10% significance (*). The large firms are firms with a market capitalization of one billion and small firms have a market capitalization of smaller than one billion.

Segment _observationsNumber of

Interest coverage ratio ANOVA

F-test Wallis test Kruskal- Chi-Square test All firms EBIT

2130 8.70*** 205.25*** 163.46***

EBITDA 8.59*** 200.31*** 135.01***

Large Firms EBIT

1137 8.07*** 181.50*** 164.93***

EBITDA 6.64*** 152.32*** 114.03***

Small Firms EBIT

997 2.65*** 80.73*** 53.63***

(29)

29 Appendix A.5 shows the results of the interest coverage’s stability per industry. The tests are most of the times significant at the 5% significant level, saying that there is no stability. The industries that show statistical stability are the Technology-, Telecommunications-, Health Care- and the Consumer Goods- industry.

In this study a few tests are prepared around the effective level, which is an interest coverage ratio from five till 11. The first test focuses on the stability of only the firms who have a median interest coverage ratio in the effective level. The test-statistics are presented in Appendix A.6 and are calculated for the small, large and for all firms. The results show that each group there is instability. However, for the small firms the test statistics is smaller, suggesting that this group might have a ‘more stable’ interest coverage ratio’. This test is also conducted per industry. However, only three industries show some stability in the interest coverage ratio. These industries are the Consumer Good industry (accounts for only the interest coverage ratio that is based on EBITDA), the Telecommunications industry and the Technology industry (accounts only for the interest coverage ratio based on EBIT). Summarized, the results show that firms in the effective level show no more stability comparative to the results presented for all firms.

(30)

30 Table 7 Distribution matrix based on percentages

This table presents the flow of interest coverage ratios between years. The interest coverage ratios are distributed between three ranges; a range with interest coverage ratio below five, a range with the effective level and a range with interest coverage ratio higher than 11.

Distribution matrix in

percentages < 5 Effective level > 11

Panel A: based on EBITDA

< 5 32.29 50.75 16.96

Effective level 5.01 62.03 32.96

> 11 1.47 25.51 73.02

Panel B: Based on EBIT

< 5 45.96 45.38 8.65

Effective level 8.75 75.20 16.05

> 11 3.55 39.41 57.04

4.2. Relationship with enterprise value

In figure 4 a scatter plot is presented with on the horizontal axis the interest coverage ratio and on the vertical axis the enterprise multiple, based on EBITDA. At first glance there seems to be no real relationship, while the dots are widely distributed. However, this scatter plot also present a trend line, which is based on logarithms to express a solid fluid line. The trend line show a similar result as Koller et al. (2010) present, while the figure that they present was based on only the tax shields. The trend line is not what was expected. What was expected is a similar line that is presented in figure 1. The trend line presented in figure 4 can be clarified by explaining that this relation is driven by the fluctuation in the operational income (EBITDA). When the EBITDA increases, the interest coverage ratio increases as well (assuming a stable cost of debt), but the enterprise multiple decreases (assuming a stable enterprise value). The expected trend line was presented in figure one. However, this figure assumes a stable operational income, while that relation is more focused on debt issues, like the costs of business erosion and the tax advantages.

(31)

31 Figure 4 Scatter plot Enterprise Multiple and Interest Coverage Ratio

The scatter plot presents the relationship between the interest coverage ratio and the enterprise multiple based on the EBITDA. The trend line in the middle is based on logarithms.

Based on the scatter plot there is no real relationship in the range of five till 30. There is a negative relation between zero and five, however this relation stabilizes after the interest coverage ratio five has been reached. Other statistics, like the Spearman’s correlation coefficient, show no strong relationship as well. Both of these Spearman’s correlation coefficients, the EBIT and the EBITDA’s variant, are significant under the 1% significance level; significantly different from zero. The correlation between the interest coverage ratio and enterprise multiple based on EBITDA is -0.092, and for the EBIT variant the correlation is -0.155. Based on these correlations there seems to be a small negative relation, which is equivalent to the results presented in the scatter plot.

(32)

32 included in the panel regression. The ROIC also is of significant value in determining the enterprise value. In this study I assume that the effect of the ROIC and the interest coverage ratio on the enterprise multiple is the same across industries; this effect should be positive on the enterprise multiple.

Table 8 Regression results based on a Panel Data regression This table presents the regression results of a panel data regression. First the coefficients of the variables are presented. These coefficients are statistically tested if their value is significant different than zero. The significance is presented by as the 1% significance (***), 5% significance (**) and 10% significance (*). The numbers between brackets represent the test statistic. The table present three panels, panel A, B and C. Panel A includes only the interest coverage ratio, panel B includes the interest coverage ratio and the ROIC, panel C includes the interest coverage ratio with the dummy variables.

Dependent var.: Mit EBITDA EBIT A B C A B C Cit -0.07*** -0.07*** -0.09 -0.13*** -0.13*** -0.15 (-13.27) (-13.62) -17.94 (-21.4) (-19.94) -23.75 Constant 10.05*** -0.28*** 12.25 13.79*** -1.00*** 15.48 -178.37 (-3.69) 88.77 -225.04 (-9.43) 91.48 ROIC 10.21*** 14.05*** -165.07 -206.1

Oil and Gas -2.32*** -1.31***

(33)

33 In panel C the ROIC is excluded. Appendix A.8 presents the panel regression inclusive the ROIC, however the ROIC has little economic meaning when it is included in panel C, while the coefficient becomes negative and significant. Panel C presents the panel regression of the interest coverage ratio and the industry dummies. Almost all industries seems to be of significant value, however all the coefficients are negative. The only industry that has a non-significant coefficient is the health care industry, suggesting that the coefficient of this industry is statistically equal to zero.

Based on the R-square and the adjusted R-square the explaining power of the regression is low. Although the R-square is increasing by adding more variables, for the EBITDA variant the final adjusted R-square is 3.6%. This suggests that the assumption of Koller et al. (2010) can be supported. In conclusion, there is no real relation between the interest coverage ratio and the enterprise multiple. The explaining power is low and the coefficient of the interest coverage ratio is low in every panel of table 8. Also the intercept is relatively high.

4.3. Sensitivity tests

(34)

34

5. Conclusion

In this section a summary of this paper and the conclusions will be discussed. These will be followed by limitations of this study and gives suggestions of further research.

5.1. Summary and main findings

This study focuses on the stability of the interest coverage ratio for small (< $ one billion) and large firms (> $ one billion). Following on this, the first hypothesis states that the interest coverage ratio is stable for the large firms and the second hypothesis states that the interest coverage ratio is stable for the small firms. This study also researches the relation between the enterprise value and the interest coverage ratio, which can be seen as a cross-check in the finding of the first hypothesis. This leads to a third hypothesis which state that there is no relation between the enterprise value and the interest coverage ratio.

The interest coverage ratio is defined as the EBITA divided by the interest expenses (Koller et al., 2010). There needs to be remarked that the EBITA is not always correctly specified in financial reports because of confusion with the depreciation. The EBITDA will replace the EBITA. In earlier existing research the EBIT is often used as the nominator in the calculation of the interest coverage ratio. For this reason this study uses the interest coverage ratio based on EBIT and EBITDA.

The sample contains 2130 non-financial firms that are included in the Russell3000 with a data set from 1992 to 2012. The financial firms, like the insurance firms, the bank and the intermediary sector are excluded.

By analyzing the interest coverage ratio in stability, the distribution over the three ranges are somehow equal. The three ranges represent an interest coverage ratio smaller than five, interest coverage ratio between five and 11 (effective level) and the interest coverage ratio bigger than 11. Also by conducting a transition probability matrix the majority of the firms in the effective level also stays in the effective level the next year. This is important, while staying in the effective level has advantages.

(35)

35 study. This distribution is similar to a positive skewed distribution which is comparative with the distribution in this study. Another test that is conducted is the Kruskal-Wallis test. This statistic is created for non-normal distributions and is based on medians. Because these last two statistics are based on medians these statistics are useful for interval data.

The first results show that there is hardly any stability in interest coverage ratio of the whole sample. With a significance level of one percent the test statistics show that there is a clear significant difference across the years. Also the size of the firm did not matter; both, the small and the large firms show significant instability. The stability of the interest coverage ratio per industry is statistically tested as well. It seems that the telecommunication, technology, health care and consumer goods industry show stability for the most tests under the five percent significance level. The last check in testing stability of the interest coverage ratio is by calculating the effective interest coverage ratio. This is an interest coverage ratio between five and 11. The test results of this effective interest coverage ratio show that there is no more stability in the interest coverage ratio. This is an unexpected sign, because according to the theory the benefits and advantages of setting the right capital structure is somewhere in between.

This study also performed a panel regression to show the relationship between the two variables, the enterprise multiple and the interest coverage ratio. The results of this regressions show that the interest coverage ratio does influence the enterprise multiple; the coefficient is significant different than zero, however, the relationship is negative and small. Also the R-square is fairly low. So the explaining factor of the interest coverage ratio is not that high.

(36)

36

5.2. Limitations and suggestions for future research

This study is based on the period 1992 and 2012. In this period two economic downturns took place, which is the internet bubble during 1999 and 2002 and the financial crisis during 2007 and 2009 (for the U.S.). These economic downturns might play a significant role in the outcomes of this study. Although this study did try to filter these economic downturns (which is unreported), there was no stability in the interest coverage ratio as well. It is hard to filter this economic downturn, because in the leading up to the crisis, the firms are affected, as well as after the crisis when everything needs to be build up again. In these periods the management might change their target capital structure; influencing the stability of this structure.

(37)

37

6. References

Altman. E., Rijken. H., 2004. How rating agencies achieve rating stability. Barclay, M., Smith, S., 1999. The capital structure puzzle: another look at the evidence. Journal of Applied Corporate Finance 12, 8-20.

Barcley, M.J., Litzenberger, R.H., 1988. Announcement effects of new equity issues and the use of intraday price data. Journal of Financial Economics 21, 71-99. Binsbergen, J., Graham, J., Yang, J., 2010. The cost of debt. Journal of Finance 65: 2089-2136

Bhojraj, S. and Lee, C.M.C., 2002. Who Is my peer? a valuation-based approach to the selection of comparable firms, Journal of Accounting Research 40, 407 – 439. Brooks, C., 2008. Introductory econometrics for finance, 2nd_{edition. Cambridge}

University Press, Cambridge.

Byoun, S., 2008. How and when do firms adjust their capital structures toward targets?. The Journal of Finance 63, 3069-3096.

Damodaran, A., 1999. Financing innovations and capital structure choices. Stern School of Business.

Damodaran, A., 2006. Damodaran on valuation: security analysis for investment and corporate finance. Journal of Finance 50, 751-754.

Devine, K., Seaton, L.,1995. An examination of quarterly financial ratio stability: implications for financial decision making. Journal of Applied Business Research 11, 81-98.

Dittmann, I. and Weiner, C., 2005, Selecting Comparables for the Valuation of European Firms, working paper, Humboldt University, Berlin, Available online at: http://sfb649.wiwi.hu-berlin.de/papers/pdf/SFB649DP2005-002.pdf.

Dheeriya, P.L., 1993. Financial analysis of the US mineral and mining industry. Recourses Policy 19, 106-123.

Greene, W., 2011. Fixed effects vector decomposition: a magical solution to the problem of time-invariant variables in fixed effects models?. Political Analysis 19, 135–146.

Hausman, J. 1978. Specification tests in econometrics. Econometrica 46, 1251– 1271.

Horigan. J., 1965. Some empirical bases of financial ratio analysis. The Accounting Review. 558-568

(38)

38 Hovakimian, A., Opler, T.C., and Titman, S., 2001. The debt-equity choice: an

analysis of issuing firms. Journal of Financial and Quantitative Analysis 36, 1–24. Johnson. W., 1979. The cross-sectional stability of financial ratio patterns. The Journal of Financial and Quantitative Analysis 14, 1035-1048.

Jung, K., Kim, Y., and Stulz, R., 1996. Timing, investment opportunities, managerial discretion and the security issue decision. Journal of Financial Economics 42, 159-186.

Kemp. M., 1963. Speculation, profitability, and price stability. The Review of Economics and Statistics 45, 185-189.

Keller, G., 2008. Statistics for management and economics. 8th_edition.

South-Western College Pub.

Koller, T., Goedhart, M., Wessels, D., 2010. Corporate valuation: measuring and managing the value of companies, 5th_{Edition. John Wiley & Sons, New Jersey.}

Kraus, A., Litzenberger, R.H., 1973. A state-preference model of optimal financial leverage. Journal of Finance 28, 911-922.

Lie, E. and Lie, H.J., 2002. Multiples used to estimate corporate value, Financial Analysts Journal 58, 44 – 54.

Majluf, N.S., Myers, S.C., 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13, 187-221.

Merwin, C., 1942. Financial small corporations. Bureau of Economic Research. Milgram, L., Spector, A., Treger, M., 1999. Managing smart, 1st_{edition. Gulf}

Publishing Company, Houston.

Modigliani, F., Miller, M.H., 1958. The cost of capital, corporation finance and the theory of investment. The American Economic Review 48, 1-38.

Modigliani, F., Miller, M.H., 1963. Corporate income taxes and the cost of capital: a correction. The American Economic Review 53, 433-444.

Myers, S.C., 2001. Capital structure. Journal of Economic Perspectives 15, 81-102. Nenide, B., Pricer, B., Camp, S.M., 2003. The use of financial ratios for research: problems associated with and recommendations for using large databases. Presented at the Fourth Annual Conference of the National Business and Economics Society, St. Thomas, Virgin Islands.

Pettit, J., Fitt, C., Orlov, S., 2004. The new world of credit ratings, UBS investment bank. online available at:

(39)

39 Pinches, G., Mingo, K., Caruthers, J., 1973. The stability of the financial patterns in industrial organizations. The Journal of Finance: 389-396

Shangguan, Z., Vasudevan, G.. Do firms knowingly issue overvalued equity? New evidence from withdrawn and equity offerings. Journal of Business and Economic Studies 14, 81 – 92.

Spearman, C., 1904. The proof and measurement of association between two things. The American Journal of Psychology 15, 72–101.

Titman, S., Wessels, R., 1988. The determinants of capital structure choice. The Journal of Finance 43, 1-19.

Websites

Russell 3000 index

http://www.russell.com/indexes/data/fact_sheets/us/russell_3000_index.asp

Information ratio analysis by Bushman (2007)

Website: http://voices.yahoo.com/using-ratio-analysis-assess-financial-stability-151999.html?cat=3

Website of Damodaran

(40)

40

7. Appendices

A.2. Distribution interest coverage ratio over the years based on EBIT A.3. Distribution enterprise multiple, exclusive the outliers

A.4. Distribution of the Interest Coverage Ratio (EBIT) in percentages over three ranges A.5. Industry effects

(41)

41

(42)

42

A.2. Distribution interest coverage ratio over the years based on EBIT

This graph shows the distribution of the interest coverage ratio over the period 1992 – 2012. The interest coverage ratio is based on the EBIT.

(43)

43

A.3. Distribution enterprise multiple, exclusive the outliers

This figure presents the distribution of the enterprise multiple based on year-observations over 21 years.

0

500 1000

1500

2000

2500

3000

3500

0

2

4

6

8 10 12 14 16 18 20 22 24 26 28 30

(44)

44

A.4. Distribution of the Interest Coverage Ratio (EBIT) in percentages over three ranges

This table presents statistical the distribution of the interest coverage ratio across three different ranges. The first range represents an interest coverage ratio lower than 5, the middle range represents the effective interest coverage ratio and the 3th range represents the interest coverage ratio larger than 11.

Number of observations

per range per subgroup Total

Interest coverage EBIT ranges in percentages

< 5 5-11 > 11

All firms 39141 50.64 27.07 22.29

Large firms 20054 46.90 29.68 23.42

Small firms 19087 56.50 22.98 20.51

Oil & Gas 3074 51.83 27.27 20.91

(45)

45

A.5. Industry effects

This table presents the results of the stability tests. Three tests are conducted. Observations are tested for stability by the ANOVA F-test, Kruskal-Wallis test and Chi-Square test, by using the 1% significance (***), 5% significance (**) and 10% significance (*).

Interest coverage ratio

Industry _IncomeOp. _observationsNumber of ANOVA _F-test

Chi-Square test Kruskal-Wallis test All firms EBIT 2130 8.70*** 163.46*** 205.25***

EBITDA 8.59*** 135.01*** 200.31***

Oil & Gas EBIT 167 5.82*** 96.97*** 122.63***

EBITDA 4.18*** 65.07*** 78.84***

Basic Materials EBIT 130 2.39*** 63.95*** 65.42***

EBITDA 1.89** 41.06*** 50.18***

Industrials EBIT 493 4.40*** 95.71*** 130.08***

EBITDA 4.67*** 71.61*** 109.98***

Consumer Goods EBIT 242 1.77** 19.59 34.58**

EBITDA 1.63** 20.1 32.12**

Health Care EBIT 295 1.2 31.96** 27.82

EBITDA 1.55* 25.58 29.78*

Consumer Services EBIT 359 2.77*** 38.46*** 44.70***

(46)

46

This table presents the results of the stability tests of the firms in the effective level. Three tests are conducted. Observations are tested for stability by the ANOVA F-test, Kruskal-Wallis test and Chi-Square test, by using the 1% significance (***), 5% significance (**) and 10% significance (*).

Interest coverage ratio

Industry _IncomeOp. _observationsNumber of ANOVA _F-test

Chi-Square test Kruskal-Wallis test

All firms EBIT 639 11.58*** 208.99*** 282.33***

EBITDA 708 13.16*** 218.20*** 310.19*** Large firms EBIT 377 12.10*** 197.25*** 282.54*** EBITDA 400 12.03*** 223.48*** 291.23***

Small firms EBIT 262 1.75** 39.04*** 42.31***

EBITDA 308 4.07*** 50.37*** 77.39***

Oil & Gas EBIT 56 5.34*** 71.66*** 95.21***

EBITDA 61 6.18*** 107.08*** 127.51*** Basic Materials EBIT 38 1.84** 40.96*** 44.84***

EBITDA 55 3.48*** 50.99*** 66.09*** Industrials EBIT 179 5.40*** 139.19*** 154.32***

EBITDA 204 6.56*** 94.63*** 147.56***

Consumer Goods EBIT 92 2.16*** 31.12* 37.83***

EBITDA 87 1.56* 23.33 28.26

Health Care EBIT 70 3.03*** 65.93*** 68.72***

EBITDA 66 2.88*** 46.55*** 65.82*** Consumer Services EBIT 112 5.87*** 101.52*** 124.56***

EBITDA 103 4.84*** 75.99*** 107.65***

Telecommunications EBIT 7 0.76 29.00* 26.91

EBITDA 13 1.39 22.9 25.58

Utilities _EBIT ₁ Not

avail. Not avail. Not avail.

EBITDA 20 1.70** 26.04 30.25*

Technology EBIT 83 0.9 15.63 20.75

(47)

47

A.7. Scatter plot Enterprise Multiple and Interest Coverage Ratio (EBIT)

(48)

48

A.8. Dated panel regression analysis inclusive ROIC

This table presents the regression results of a panel data regression. First the coefficients of the variables are presented. These coefficients are statistically tested if their value is significant different than zero. The significance is presented by as the 1% significance (***), 5% significance (**) and 10% significance (*). The numbers between brackets represent the test statistic. The table present three panels, panel A, B and C. Panel A includes only the interest coverage ratio, panel B includes the interest coverage ratio and the ROIC, panel C includes the interest coverage ratio with the dummy variables.

Dependent var.: Mit EBITDA EBIT A B C A B C Cit -0.07*** -0.07*** -0.10*** -0.13*** -0.13*** -0.15*** (-13.27) (-13.62) (-17.67) (-21.4) (-19.94) (-22.08) Constant 10.05*** -0.28*** -0.29*** 13.79*** -1.00*** -0.98*** (178.37) (-3.69) (-3.75) (225.04) (-9.43) (-9.26) ROIC 10.21*** -2.02*** 14.05*** -1.23*** (165.07) (-10.81) (206.1) (-5.28)

Oil and Gas -2.25*** -1.81***

Master’s Thesis Finance The Stability of the Interest Coverage Ratio and its Relation with Enterprise Value