• No results found

The effect of corporate taxes on leverage and investment : evidence from a tax reform in Germany

N/A
N/A
Protected

Academic year: 2021

Share "The effect of corporate taxes on leverage and investment : evidence from a tax reform in Germany"

Copied!
39
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

1

The effect of corporate taxes on leverage and

investment: Evidence from a tax reform in Germany

Abstract

This thesis examines the effects of corporate taxes and financial constraint on the leverage ratio and investments of German firms, a question that is far from settled in the existing literature on the subject. This thesis contributes the ongoing debate over whether taxes are a determinant of the capital structure of firms by conducting differences-in-differences regressions. This thesis also contributes to the ongoing debate whether the cost of capital is a determinant of corporate investment. The research sample consists of 6,159 firm-years for 1,088 firms and includes all German and Dutch non-financial and non-utility companies that were trading publicly during the 2008-2015 period. Taxation appears to have a positive but insignificant effect on leverage, including for financially unconstrained firms, contrary to expectations. It appears to have a significantly negative effect on investment, including for unconstrained firms, confirming expectations. In addition, this research finds that return on assets, firm size, tangibility, and market-to-book ratio have had a significant effect on leverage and investment.

Nava Bossink 10552499 June 2017

MSc Finance – Corporate Finance University of Amsterdam

Master Thesis

(2)

2

Statement of Originality

This document is written by Nava Bossink who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Table of Contents

Title and Abstract ...1

Statement of Originality ... 2 1. Introduction ... 4 1.1 Introduction ... 4 1.2 Problem Discussion ... 6 1.3 Research Question ... 6 1.4 Overview ... 6 2. Literature Review ... 7

2.1 Effects of Taxation on Leverage ... 7

2.2 Effects of Taxation on Investment ... 9

2.3 The Effects of Financial Constraint ... 10

2.4 Hypotheses ... 12

3. Methodology ... 13

3.1 Main Variables ... 13

3.2 Control Variables ... 14

3.3 The Regression Models ... 15

4 Data and Descriptive Statistics ... 18

4.1 Sample Selection ... 18

4.2 Distribution and Descriptive Statistics ... 18

5. Results ... 21 5.1 Empirical Results ... 21 5.2 Explanation ... 26 6. Robustness Checks ... 28 6.1 Empirical Results ... 28 6.2 Explanation ... 30 7. Conclusion ... 32 7.1 Conclusion ... 32

7.2 Limitations and Suggestions ... 33

References ... 35

Appendix A: Variable Definitions... 38

(4)

4

1. Introduction

1.1 Introduction

When interest payments are tax deductible, debt confers a tax benefit (also known as a tax shield). At least since Modigliani and Miller (1958), this tax benefit has been a cornerstone of corporate finance. The effect of the cost of capital on corporate investment is also an important aspect of corporate finance. However, despite being strongly motivated by theory, an important paradox has been the inability of researchers to find a significant effect of the cost of capital on corporate investment. The impact of financial constraint on leverage and corporate investment is a core question not only in this field, but also for asset pricing field and monetary policy. Financially constrained firms are subjected to a market imperfection in the sense that external capital is not a perfect substitute for their internal capital, particularly in the short run. These firms have difficulty in raising external capital and are sometimes forced to forego profitable investment opportunities.

Moreover, little is known about the impact of these issues in Europe, since almost all studies focus on the US. European countries and the US differ in many ways; the US is a unique country with different regulations in each state. Therefore, it would be incorrect to conclude that results obtained for the US also apply in European countries. Germany is one of the most prosperous countries in Europe and is effectively overseeing the European Union. Therefore, decisions taken in Germany have a large impact on other European countries. This makes Germany an interesting country to study. As can be ascertained from Figure 1, Germany has experienced several increases in the corporate income-tax rate during the 2008-2015 period. On the other hand, as indicated in Figure 2, in the Netherlands the corporate tax rate remained unchanged during the 2008-2015 period. This provides a useful opportunity to investigate the effect of increases in the corporate-income tax rate on the leverage ratio and investments of German firms in a natural experiment. German firms are treated firms, while firms listed in the Netherlands are control firms.

This thesis investigates the effects of corporate taxes and financial constraint on the leverage ratio and investments of German firms. The sample for this study comprises 6,159 firm-years for 1,088 firms and includes all German and Dutch non-financial and non-utility companies that were publicly trading during the 2008-2015 period. The sample is constructed with the Compustat Global Fundamentals Annual database. On the basis of the previous

(5)

5

studies described in the literature review, it is expected that an increase in the corporate-income tax rate has a positive effect on the leverage ratio and a negative effect on the investments of firms. This effect is expected to be more pronounced for unconstrained firms than for constrained firms.

The results obtained in this thesis do not provide evidence of a significant effect of taxation on the leverage ratio of German firms. Moreover, in response to a tax increase, unconstrained firms do not behave differently compared to constrained firms. This differs from expectations. However, the results provide evidence of a significant effect of taxation on investments of German firms. Moreover, the results show a significant effect for unconstrained firms. This confirms expectations.

Figure 1: German Corporate-Tax Rate

The corporate tax rate in Germany for the period 2011-2016. As indicated, Germany faced several increases in the corporate tax rate during this period.

Figure 2: Dutch Corporate-Tax Rate

The corporate tax rate in the Netherlands in the period 2011-2017. As indicated, the corporate tax rate remained constant in the Netherlands during this period.

(6)

6

1.2 Problem Discussion

Despite plentiful research investigating the effect of taxation on leverage, the debate over whether taxes are a determinant of the capital structure of firms is far from settled. Firms with higher profits may borrow more than low-profit firms in order to exploit the tax advantage provided by debt. However, it is also possible that these firms borrow more because they have a lower default risk than low-profit firms (Heider & Ljungqvist, 2015). This thesis contributes the ongoing debate over whether taxes are a determinant of the capital structure of firms by conducting differences-in-differences regressions. Compared to previous research, the differences-in-differences estimator used in this study has a unique definition. This variable is further explained in the methodology chapter. This thesis also contributes to the ongoing debate whether the cost of capital is a determinant of corporate investment. The existing literature has not found a final answer to this question; there are papers that find an insignificant effect of taxation on investment, but also papers that found a significantly positive or negative effect. Moreover, while much of the literature on the effect of taxation on investment is outdated, this thesis investigates a recent period (2008-2015); consequently, the results are more contemporary than previous studies.

This thesis provides a distinct perspective on the effects of taxation in Europe. While previous studies have investigated either the effect of financial constraint, or the effect of taxation in the form of tax increases, on leverage or investment, this study investigates these effects simultaneously

1.3 Research Question

The research question in this study is: What are the effects of corporate taxes and financial constraint on the leverage ratio and investments of German firms?

1.4 Overview

Chapter 2 provides a review of the relevant literature. The methodology used in this thesis is described in Chapter 3, and the data sources and descriptive statistics discussed in Chapter 4. Chapter 5 presents the results, and Chapter 6 the robustness tests. Finally, Chapter 7 presents the main conclusions.

(7)

7

2. Literature Review

This chapter provides an overview of the main theories of the effects of taxation in the existing literature. The first section discusses theories of the effects of taxation on leverage, the second outlines the effects of taxation on investments, and the third describes the effects of financial constraints. Finally, in the fourth section, the hypotheses are generated.

2.1 Effects of Taxation on Leverage

As outlined in the introduction, this thesis contributes to the ongoing debate about whether taxes are a determinant of the capital structure of firms. Despite the depth of the existing literature on this topic, there is as yet no a satisfying answer to the question.

The first theory of capital structure was offered by Modigliani and Miller (1958). They stated that the value of a firm is unaffected by the way an investment is financed if the firm is in a perfect capital market. A perfect capital market refers to an ideal world with rational economic agents, but with no information asymmetries, no taxes, and no other transaction costs. This theory was reformed by Kraus and Litzenberger (1973). They added some relaxations of the original, unrealistic assumptions, recognising the existence of bankruptcy penalties and the taxation of corporate profits as market imperfections. This theory is called the tradeoff theory. They argued that the risk of default increases with a higher debt level, which leads to a higher interest rate to compensate for this risk. Conversely, a higher debt level also results in a higher tax advantage since interest expenses are tax deductible. When the leverage ratio increases, the marginal cost of financial distress increases, and the marginal benefit of the tax advantage declines. This continues until the value of a firm is again independent of the capital structure, as in that case these two forces perfectly offset each other. According to this theory, firms do not increase their debt level in response to a tax increase. They are already at their optimal level of debt, so a further increase in debt will reduce firm value. In that case, the marginal costs of a debt increase are higher than the marginal benefits.

In 1984, Myers introduced the pecking order theory, which states that no perfect market exists. This is explained by the fact of asymmetric information in capital markets. Managers know more about a company’s values and risks than do investors. This asymmetric availability of information affects the financing decisions of firms. According to Myers (1984), new investments can be financed with retained earnings, debt or equity. Of these, for Myers,

(8)

8

the first is the most preferable, as firms are then not dependent on external financing. Issuance of debt is the second most preferable choice. A debt issuance signals that that the current stock is undervalued and that the management has confidence in the profitability of an investment. An equity issuance is seen as a lack of confidence and signals that the stock is overvalued. This theory has no clear answer to the question of whether taxation affects the leverage ratio of firms. In 1984, Myers asserted: ‘I know of no study clearly demonstrating that a firm’s tax status has predictable, material effects on its debt policy. I think the wait for such a study will be protracted.’

There are several studies that find a significant positive relationship between taxation and leverage. Graham (1996) investigated whether the incremental use of debt is positively related to firm-specific tax rates. He analysed a sample of more than 10,000 firms for the period 1980-1992 and found that high tax-rate firms have higher debt levels than their low tax-rate counterparts. MacKie-Mason (1990) studied incremental financing decisions using discrete choice analysis. He examined a large dataset to study the effects of two tax shields on leverage, and also found a positive relationship between tax and leverage. However, Fama and French (1998) have argued that these approaches are vulnerable to endogeneity biases as the marginal tax rate of firms may correlate with omitted variables.

Rajan and Zingales (1995) investigated the determinants of the capital structure of firms by analysing the financing decisions of public firms in the G7 countries. They demonstrated that firms in countries with high corporate-tax rates use more leverage than countries with lower tax rates. Heider and Ljungqvist (2015) analysed the effect of staggered corporate income-tax changes on the leverage of firms. They found an asymmetric effect: firms increase their leverage by around 40 percentage points for every percentage-point tax increase, but leverage does not respond to tax cuts.

In light of the findings of these paper, it can be concluded that firmsincrease their debt ratio in response to an increase in the corporate income tax rate, despite the increase in bankruptcy risk. This means they value the tax benefit and the positive signal they deliver to the market more highly than they fear the costs of the increase in bankruptcy risk.

(9)

9

2.2 Effects of Taxation on Investment

As outlined in the introduction, this thesis contributes to the ongoing debate over whether the cost of capital is a determinant of corporate investment. Most papers do not find a significant relationship between cost of capital and investment, despite the strong theoretical motivations.

Modigliani and Miller (1958) state that the advantage of using debt over equity is relatively small. The only difference in cost is simply the tax on the ‘grossed up’ interest payment. Even in the extreme case where a firm could borrow at a rate close to zero, the advantage of debt would be close to zero.

After the Tax Reform Act of 1986 (TRA86) in the US, equipment investment no longer qualified for an investment tax credit. Therefore, investments were expected to decline after the reform (Cummins & Hassett, 1992). However, there was instead a substantial increase in equipment investment after 1986. According to these authors, this circumstance confirms that the cost of capital is an unimportant factor for predicting investment. However, Auerbach and Hassett (1991) show that this conclusion had been drawn too soon. They developed an econometric technique to estimate the effect of changes in tax policy on investments. This technique relates forecast errors in investment to forecast errors in the cost of capital. The authors found a large, significant effect for changes in tax policy on equipment investment: the actual investment in equipment was significantly lower than what had been predicted.

Using firm-level panel data, Cummins and Hassett (1992) explored the relationship between the cost of capital and investment, and also found a statistically and economically significant relationship. The authors used TRA86 as a natural experiment to overcome measurement problems, which they argue are probably the cause of the insignificant relationships found in past research.

Abel (1982) developed a framework to analyse the dynamic effects of tax policy on corporate investment. He demonstrated that a temporary investment tax credit provides a greater encouragement to investment than a permanent investment tax credit. His analysis also indicate that the response to a temporary change in the tax rate would be wholly different if firms were allowed to write off a fraction of their capital expenditure immediately instead of in proportion to their physical life.

The findings of Summers et al. (1981) also have contradicted the statement that tax reforms have no effect on corporate investment. They found that, after a cut in the corporate

(10)

10

tax rate, q (a measure of investment opportunities) reaches a new equilibrium value at a higher level of capital than before the tax cut. In the short-run, the amount of capital in firms will decline following a corporate tax-rate increase. However, the long-term effects of a corporate income tax increase on capital are ambiguous.

In 1969, Hall and Jorgenson analysed the effect of a tax cut and a tax increase on investment. In 1964, the American corporate income-tax rate was reduced from 52% to 48%, and from October 1966 to March 1967, the investment tax credit was suspended. After the tax reduction in 1964, they found a small negative effect on net investment, gross investment and capital stock. They explained this by the fact that the rental prices of capital had increased after the tax cut. The authors also found a decrease in net investment, gross investment, and capital stock after the tax reform in 1966. The decrease in investment following a tax cut seems to contradict the results of the other papers described in this paragraph.

Based on the findings of the papers described in this paragraph, it can be concluded thatthe effect of an increase in the corporate income tax rate on investment is ambiguous: it may be positive, negative, or insignificant.

2.3 The Effects of Financial Constraint

Another important aspect to investigate is the effect of financial constraints. These have proved to have a significant impact on the leverage ratio and investments of firms (Fazzari et al., 1988; Almeida & Campello, 2007).

Fazzari et al. (1988) provided early evidence that financial constraint has an impact on investment. They divided their sample into three groups based on each firm’s dividend payout. They used dividend payouts as a measure of financial constraint, as firms that make high dividend payments instead of saving are likely not to be financially constrained. The authors tested the relationship between annual cash flows and investment across each sub-sample. Their results demonstrate that the effect of cash flows on investment is the strongest for the most constrained firms and the weakest for the least constrained firms. This is consistent with the view that firms invest more when they have more internal funds available.

Kaplan and Zingales (1997) criticised the results of Fazzari et al (1988). The former argue that investment-cashflow sensitivity does not necessarily increase with financial constraints. They refuted the methodology of Fazzari et al (1988) by re-classifying firms based on the discussion of financial constraints in their 10-K statements. Compared to a company’s

(11)

11

annual report, a 10-K statement is a more detailed summary report of a company’s performance. In these statements, firms are required to disclose their problems with raising external financing. According to Kaplan and Zingales (1997), this presents a more reliable indication of whether firms are financially constrained or not. They found that cash flows are associated with investment across all firms in the sample, regardless of whether a firm is financially constrained. The least financially constrained firms exhibited a greater investment-cashflow sensitivity than firms classified as more financially constrained. The authors therefore argue that investment-cashflow sensitivities do not provide useful measures of financial constraint.

Hennessy and Whited (2007) investigated the magnitude of financing costs. Their results imply that the costs of financing are higher for firms with low dividend payouts and firms defined as constrained according to the Cleary and Whited-Wu indexes. When financing cost parameters increase, many common financing constraint proxies decrease.

Denis and Sibilkov (2010) analysed why cash holdings are more valuable for constrained firms than for unconstrained firms. Their results imply that, for constrained firms with high hedging needs, greater cash holdings are associated with higher investment levels. Moreover, the relationship between investment and value is stronger for financially constrained firms than for unconstrained firms. Gomes et al. (2006) investigated the importance of financing constraints for the cross-section of returns. They found that financing problems provide a common factor that improves the pricing of cross-sectional returns.

Lamont et al. (2001) studied whether the impact of financial constraints on the value of firms is observable in stock returns. They found that stock returns of financially constrained firms move together over time. This suggests that constrained firms are subject to common shocks. Moreover, the authors created a new measure of financial constraint, the Kaplan-Zingales (KZ) index. Firms that are more constrained have a higher index level. Lamont et al. (2001) created a linear combination using five accounting ratios. These five variables are cash flow to total capital, the market-to-book ratio, debt to total capital, dividends to total capital and cash holdings to capital. Cash flow to total capital, dividend to total capital and cash holding to capital have a negative sign and the market-to-book ratio and debt to total capital have a positive sign. For example, if a firm has higher cash flows, it is less likely to be financially constrained.

(12)

12

Whited (1992) provides evidence supporting the theory that asymmetric information problems in debt markets affect the ability of financially constrained firms to obtain external financing. Consequently, this also affects their allocation of investment expenditures over time. The standard Euler equation of an optimising model of investment fits well for unconstrained firms, in contrast to constrained firms.

Based on the findings of the papers described in this paragraph, an increase in the corporate income tax rate has a stronger effect on the leverage ratio and investments of unconstrained firms than those of constrained firms.

2.4 Hypotheses

As described in the first section of this chapter, the existing literature has not provided a final answer to the question of the effects of taxation on the leverage ratio of firms. In line with the studies of Graham (1996), Heider and Ljungqvist (2015), MacKie-Mason (1990), and Rajan and Zingales (1995), I expect that the leverage ratio of firms will increase in response to an increase in the corporate income-tax rate. Moreover, consistent with the studies of Denis and Sibilkov (2010), Fazzari et al. (1988), Hennessy and Whited (2007), Lamont et al. (2001), and Whited (1992), I expect this effect to be more pronounced for unconstrained firms. This leads to the following hypotheses:

H1a: The leverage ratio of firms increases in response to an increase in the corporate income tax rate.

H1b: This effect is more pronounced for unconstrained firms than for constrained firms.

As described in the second section of this chapter, the existing literature has not yet provided a final answer to the question of the effects of taxation on the investments of firms. Following the findings of Auerbach and Hassett (1991), Cummins and Hassett (1992), and Summers et al. (1981), I expect that the investments of firms will decrease in response to an increase in the corporate income tax rate. In line with the studies of Denis and Sibilkov (2010), Fazzari et al. (1988), Hennessy and Whited (2007), Lamont et al. (2001), and Whited (1992), I expect this effect to be more pronounced for unconstrained firms. This leads to the following hypotheses:

H2a: The investments of firms decreases in response to an increase in the corporate income tax rate.

(13)

13

3. Methodology

This chapter describes the variables and the regression model used in this study. In the first section, the main variables are explained. The second section describes the control variables and the last, the regression model.

3.1 Main Variables

In order to investigate the effect of an increase in the corporate income tax rate on the leverage ratio and investments of firms, two dependent variables are formulated. The first (Dit) measures the long-term book leverage ratio. To provide more robustness, I use four

different measures of this ratio – long-term debt divided by total assets, long-term debt (including the portion due within 1 year) divided by total assets, short-term and long-term debt divided by total assets, and long-term debt divided by the market value of assets – as proposed by Heider and Ljungqvist (2015). The second dependent variable (Iit) measures

investment, where investment is calculated by dividing capital expenditure by the total assets of the previous year.

The main variable of interest in this thesis is that which measures the treatment effect of several increases in the corporate income tax rate (see Chapter 3.3 for further explanation of this variable). The co-efficient of this variable is expected to be positive in the regressions where leverage ratio is the dependent variable, and negative in the regressions where investment is the dependent variable.

To measure the impact of financial constraint, an interaction variable is included. This is an interaction between the variable that measures the treatment effect of several increases in the corporate income tax rate, and a dummy variable for financial constraint that takes the value 1 if a firm is classified as unconstrained, and 0 otherwise. The financial constraint measures that are used in this thesis are dividend payout and the KZ index, constructed following Lamont et al. (2001) (see Appendix A for definitions of the variables). When utilising dividend payout, firms are classified as constrained if they have not paid any dividend during the sample period (2008-2015). When utilising the KZ index, firms are classified as constrained if they have a KZ-score above the sample median. As described in the literature review, the effect of an increase in the corporate income tax rate is expected to be more pronounced for unconstrained firms. Therefore, the co-efficient of this interaction variable is expected to be

(14)

14

positive in the regressions with leverage ratio as the dependent variable, and negative in the regressions with investment as the dependent variable.

3.2 Control Variables

A number of other variables have an effect on the leverage ratio and investments of firms, but are not the primary focus of this thesis. These are the control variables; as used in this thesis, they are: return on assets, firm size, tangibility and investment opportunities. These variables are commonly found in empirical models of leverage and investment (Rajan & Zingales, 1995; Heider & Ljungqvist, 2015).

Return on assets (ROA) is measured by dividing the operating income before depreciation by the total assets. The pecking order theory predicts a negative relation between profitability and debt ratios. This theory states that firms prefer to use retained earnings to finance their investments. Therefore, if a firm has a higher profitability, it needs less debt to finance new investments. Moreover, as both Ozkan (2001) and Heider and Ljungqvist (2015) found, profitability has a significant negative effect on the leverage ratio of firms. Therefore, I expect the sign of return on assets (a measure for profitability) regressed on the leverage ratio to be negative. Myers (1984) and Myers and Majluf (1984) state that profitability is positively related with investment. Therefore, I expect the sign of return on assets regressed on investment to be positive.

Firm size (FirmSize) is measured by taking the natural logarithm of the total assets. Homaifar et al. (1994) and Titman and Wessels (1988) have argued that larger firms have higher leverage ratios. Heider and Ljungqvist (2015) also found a positive relationship between firm size and leverage ratio. This is explained by the fact that there is less information asymmetry in large firms (Collins, 1987). Lower information asymmetry allows firms to borrow more easily. Large firms are less dependent on internal financing than smaller firms, so they can invest more easily. Therefore, the relationship between firm size and investment is also expected to be positive. In conclusion, the expected sign of the firm-size variable is expected to be positive in all regressions.

Tangibility (Tangibility) is measured by dividing (net) property, plant and equipment by the total assets. Almeida and Campello (2007) have argued that firms with more tangible assets are less likely to be financially constrained. This means they have easier access to

(15)

15

external financing. Therefore, I expect a positive relationship between tangibility and leverage, as well as between tangibility and investment.

The market-to-book ratio (Markettobook) is used as a proxy for investment opportunities. This ratio is measured by multiplying the year-end closing price by the total shares outstanding. This is added to the long- and short-term debt and preferred stock. Next, deferred taxes and investment tax credits are subtracted. The result is divided by the total assets. Almeida and Campello (2007) and Erickson and Whited (2000) found a significant positive relationship between investment opportunities and investment; that is, when firms have more investment opportunities available, they will invest more. Therefore, I expect a positive relationship between the market-to-book ratio and investment. The relationship between investment opportunities and leverage is also expected to be positive, as higher demand for investment results in higher demand for external financing (Erickson & Whited, 2000).

Due to multicollinearity, independent variables should be excluded if they have a high correlation with other independent variables, especially with the main variable of interest. Based on the table of the correlation co-efficients (see Appendix B), no variable was removed as a result of a high correlation with other independent variables. However, the variable Post

Yeart was excluded as the regressions also include time-fixed effects. If both were included,

the regressions would exhibit perfect multicollinearity. Moreover, in the regressions with fixed effects, the variables Germani, Unconstrainedi and Germani*Unconstrainedi are also excluded

as they are constant within firms over time.

3.3 The Regression Models

To test the effect of corporate taxes on the leverage ratio and investments of firms, a differences-in-differences method is used. The following regression models were constructed: 𝐷𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2 ∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 + 𝛽3∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽4 ∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽5∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽6∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑡+ 𝜀𝑖𝑡

𝐼𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 + 𝛽3∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽4∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽5∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽6∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑡+ 𝜀𝑖𝑡

Where 𝑖 indexes firms and 𝑡 fiscal years. Year-fixed effects are included to exclude unobserved time-varying shocks. The variable of interest in this thesis is the differences-in-differences estimator (Germani*Post Yeart). This variable is an interaction between Post Yeart

(16)

16

(a dummy variable taking the value 1 for the years after a tax increase, otherwise 0) and

Germani (a dummy variable taking the value 1 for German firms, and 0 otherwise). As

explained in the introduction, Germany faced several increases in the corporate income tax rate in the period 2012-2015. Therefore, Post Yeart is equal to 1 in the period 2012-2015, and

equal to 0 in the period 2008-2011. If the co-efficient of Germani*Post Yeart is positive, it

means that firms are increasing their leverage ratio or investments as a response to an increase in the corporate income-tax rate. As explained in the literature review, this variable is expected to be positive in the leverage regressions and negative in the investment regressions. The results of these regressions are displayed in Chapter 5.

To deal with possible endogeneity issues, year-fixed effects are included. This reduces the omitted variable bias. Firms in the treatment group are German firms that faced several increases in the corporate income-tax rate during the 2011-2016 period. Firms in the control group are firms listed in the Netherlands, where the corporate income tax rate remained the same during that period. Using a differences-in-differences method reduces endogeneity because it minimises the effect of unobserved trends in local economic conditions. Firms listed in the Netherlands are a suitable control group as the Netherlands is adjacent to Germany, the two countries have many similar characteristics, and therefore they probably have similar economic trends.

In order to investigate the impact of financial constraint, the following regression models were constructed:

𝐷𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2 ∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 + 𝛽3∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖 + 𝛽4∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖 + 𝛽5∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖 + 𝛽6∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡 ∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖+ 𝛽7∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽8∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽9∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽10∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑡+ 𝜀𝑖𝑡 𝐼𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 + 𝛽3∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖+ 𝛽4∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖 + 𝛽5∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖 + 𝛽6∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡 ∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖+ 𝛽7∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽8∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽9∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽10∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑡+ 𝜀𝑖𝑡

As above, 𝑖 indexes firms and 𝑡 fiscal years. Compared to the previous set of equations, these have several additional variables. As explained in section 3.1, the interaction variable

(17)

17

different measures of financial constraint (dividend payout and the KZ index), these equations will be regressed twice. The results of these regressions are presented in Chapter 5.

In the robustness tests, I also include firm-fixed effects in order to exclude unobserved firm characteristics that remain constant over time. As indicated above, the variables Germani, Unconstrainedi and Germani*Unconstrainedi are excluded from these regressions. This results

in the following equations:

𝐷𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2 ∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽3∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽4∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽5∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑖𝑡+ 𝜀𝑖𝑡

𝐼𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽3∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽4∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽5∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑖𝑡+ 𝜀𝑖𝑡

In order to investigate the impact of financial constraint, the following regression models were constructed:

𝐷𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖+ 𝛽3∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡 ∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖+ 𝛽4∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽5 ∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽6∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽7∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑖𝑡+ 𝜀𝑖𝑡 𝐼𝑖𝑡 = 𝛽1∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡+ 𝛽2∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖 + 𝛽3∗ 𝐺𝑒𝑟𝑚𝑎𝑛𝑖 ∗ 𝑃𝑜𝑠𝑡 𝑌𝑒𝑎𝑟𝑡 ∗ 𝑈𝑛𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑𝑖+ 𝛽4∗ 𝑅𝑂𝐴𝑖𝑡−1+ 𝛽5∗ 𝐹𝑖𝑟𝑚𝑠𝑖𝑧𝑒𝑖𝑡−1+ 𝛽6∗ 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡−1+ 𝛽7∗ 𝑀𝑎𝑟𝑘𝑒𝑡𝑡𝑜𝑏𝑜𝑜𝑘𝑖𝑡−1+ 𝛼𝑖𝑡+ 𝜀𝑖𝑡

The results of these regressions are presented in Chapter 6.

(18)

18

4 Data and Descriptive Statistics

This chapter discusses the data sample and descriptive statistics. The first section explains how the sample is formed, and the second section presents the main characteristics of the data.

4.1 Sample Selection

The initial sample consisted of all publicly listed firms in Germany and the Netherlands for the period 2008-2015. From the Compustat Global Fundamentals Annual database, I excluded utilities (SIC code 49; 29 observations), financial firms (SIC code 6; 3 observations) and public sector entities (SIC code 9; 22 observations). I also exclude firm-years with missing return on assets (37 observations), negative or missing total assets (14 observations), or firms that were delisted during the sample period (62 observations).

After correcting for these criteria and removing transactions with missing data, the final sample consisted of 6,159 firm-years for 1,088 firms.

4.2 Distribution and Descriptive Statistics

Table 1 presents the distribution of constrained and unconstrained firms in the total sample of 1,088 firms. According to the first measure of financial constraint (dividend payout), approximately 53% of the firms are classified as constrained and the remainder as unconstrained. According to the second measure of financial constraint (the Kaplan-Zingales index), approximately 63% of the firms are classified as constrained and the remainder as unconstrained. According to both measures the majority of the firms were classified as constrained. However, the KZ index classified 104 more firms as constrained than did the dividend payout measure.

Table 2 shows the distribution of the treatment group (German firms) and the control group (Dutch firms) in the total sample of 1,088 firms. The control group consists of 17% of the total sample (186 firms), and the treatment group 83% (902 firms). As it could be argued that the fact that the treatment group is significantly larger than the control group may lead to biased results, I also include a robustness test.

(19)

19

Table 1: Sample Distribution – Constrained and Unconstrained Firms

Table 1 displays the distribution of constrained and unconstrained firms across the total sample. The sample consists of 6,159 firm-years for all non-financial and non-utility German and Netherlands companies that are publicly trading over the period 2008-2015, as per the Compustat Global Fundamentals Annual database.

Measure of constraint

Dividend KZ index

Constrained 580 684

Unconstrained 508 404

Total 1,088 1,088

Table 2: Sample Distribution – Treatment and Control Group

Table 2 displays the distribution of the treatment group (German firms) and the control group (Dutch firms) across the total sample. The sample consists of 6,159 firm-years for all financial and non-utility German and Netherlands companies that are publicly trading over the period 2008-2015, as per the Compustat Global Fundamentals Annual database.

Country: N %

Germany 902 83

Netherlands 186 17

Total 1,088 100

Table 3 displays some of the descriptive statistics pertaining to the dependent and independent variables used in this research. There are various means of measuring how much debt a firm uses to finance its operating activities. It can be expected that long-term book leverage will be the most sensitive to changes in the tax rate. Short-term debt is unlikely to be altered in response to a tax increase, because it is often used as working capital. Moreover, if the yield curve is slopes upward (interest rates are increasing with maturity), long-term debt generates a larger tax shield than short-term debt. In addition, firms have less control over market leverage than over book leverage. Therefore, book leverage is a cleaner measure than market leverage. As Table 3 indicates, the long-term book-leverage ratio is on average 15.4%, while the long-term market-leverage ratio is on average 17.5%. There are large differences in the amount of long-term debt, as can be seen from the standard deviation, which is around €10,542.86 million. Investment varies between 0 and €21.81 million, with the average investment valued at €261,000. As is displayed in Table 3, the average firm in the sample has return on assets of 6.9%, and €8,692.76million in total assets; 19.3% of these assets are tangible (on average). The average firm here trades at a market-to-book ratio of 1.20.

(20)

20 Table 3: Summary Statistics

The sample consists of 6,159 firm-years for 1,088 firms and includes all German and Dutch non-financial and non-utility companies that were trading publicly during the 2008-2015 period, as per the Compustat Global Fundamentals Annual database. The table reports summary statistics for the dependent variables and the controls. Investment is measured by dividing capital expenditures by the total assets of the previous year. Return on assets is measured by dividing the operating income before depreciation by the total assets. Tangibility is measured by dividing (net) property, plant and equipment by the total assets. Market/book is measured by multiplying the year-end closing price by the total shares outstanding. This is added to the long- and short-term debt and preferred stock. Next, deferred taxes and investment tax credits are subtracted. The result is divided by the total assets. Investment, leverage, return on assets, firm size, tangibility and market/book are winsorized 0.5% in each tail. All firm-years percentile

Mean S.D. N Min 25th 50th 75th Max

Firm leverage long-term debt / assets 0.154 0.171 5801 0 0.010 0.103 0.237 0.864

long-term debt (incl. current

portion) / assets 0.194 0.201 4482 0 0.009 0.146 0.304 1.036

(short-term and

long-term debt) / assets 0.231 0.229 5801 -0.888 0.100 0.217 0.350 1.089

long-term debt /

market value of assets 0.175 0.208 4859 0 0.004 0.107 0.270 1.110

Long-term debt (€m) 1730.813 10542.860 5801 0 0.715 12.500 164.005 213108 Investment 0.261 1.803 6158 0 0.014 0.034 0.071 21.809 Firm characteristics ROA 0.069 0.146 6159 -1.106 0.036 0.083 0.129 0.448 Total assets (€m) 8692.761 43081.500 6159 0.312 38.061 156.972 1028.987 427090 Tangibility 0.193 0.197 5752 0 0.029 0.136 0.294 0.874 Market/book 1.197 1.283 4859 -0.109 0.568 0.865 1.311 13.258

(21)

21

5. Results

This chapter provides an overview of the principal results of the regressions. The first section presents the results in three tables, and the section following provides further explanation.

5.1 Empirical Results

Table 4 presents an overview of the key results of the regressions, which measure the impact of an increase in the corporate income-tax rate on the leverage ratio and investments of German firms. The impact of financial constraint is displayed in tables 5 and 6. All the tables comprise five columns, the first four having long-term book leverage as dependent variable, and the last, investment as dependent variable. In order to increase robustness, I use four different definitions of long-term book leverage. In column 1, the dependent variable is equal to long-term debt divided by total assets. In column 2, it is equal to long-term debt (including the current portion) divided by total assets. In column 3, the dependent variable is equal to short- and long-term debt divided by total assets. In column 4, it is equal to long-term debt divided by the market value of assets. In the final column, the dependent variable is equal to investment (defined as capital expenditure divided by the total assets of the previous year.)

Table 4 shows a positive average treatment effect (ATE) for German firms in the first four columns, and a negative average treatment effect in the fifth column. However, the treatment effect is only significant (at the 5% level) in regression 5. The positive ATE in the first four regressions is consistent with the findings of Graham (1996), Heider and Ljungqvist (2015), MacKie-Mason (1990) and Rajan and Zingales (1995) and with hypothesis H1a. The significantly negative ATE in the last regression is consistent with the findings of Auerbach and Hassett (1991), Cummins and Hassett (1992), and Summers et al. (1981), and with hypothesis H2a. However, it is not consistent with the findings of Hall and Jorgenson (1969). The insignificant ATE in regressions 1-4 is consistent with the findings of Kraus and Litzenberger (1973), and with the tradeoff theory.

Return on assets is significant in regressions 1-3 (at the 1% level) and in regression 4 (at the 10% level); the effect on leverage is negative and the effect on investment is positive. This is consistent with the findings of Heider and Ljungqvist (2015), Myers (1984), Myers and Majluf (1984), Ozkan (2001), and with the pecking order theory. Firm size is positively significant at the 1% level in all regressions. This is consistent with the findings

(22)

22 Table 4: Effect of tax changes on leverage and investment

I estimate standard regressions to test whether, and by how much, firms change their leverage or investments in response to changes in their corporate income tax rate. Post Year is a dummy variable taking the value 1 for the years after a tax increase (2012 until 2015) and 0 otherwise. German is a dummy variable taking the value 1 for German firms (firms in the treatment group) and 0 for Netherlands firms (firms in the control group). German*Post Year is the interaction term of these two variables (the main variable of interest) and measures the average treatment effect of an increase in the corporate income tax rate. Return on assets is measured by dividing the operating income before depreciation by the total assets. Firm size is measured by taking the natural logarithm of the total assets. Tangibility is measured by dividing (net) property, plant and equipment by the total assets. Market/book is measured by multiplying the year-end closing price by the total shares outstanding. This is added to the long- and short-term debt and preferred stock. Next, deferred taxes and investment tax credits are subtracted. The result is divided by the total assets. In column 1, the dependent variable is equal to long-term debt / assets. In column 2, the dependent variable is equal to long-long-term debt (incl. current portion) / assets. In column 3, the dependent variable is equal to short and long-term debt / assets. In column 4, the dependent variable is equal to long-term debt / market value of assets. In the last column, the dependent variable is equal to investment (defined as capital expenditures divided by the total assets of the previous year.) All regressions are differences-in-differences regressions; the German*Post Year-variable is the differences-in-differences estimator. All specifications include year fixed effects to remove unobserved time-varying shocks. The fixed effects are not reported for brevity. Heteroskedasticity-consistent standard errors are shown in parentheses underneath the coefficient estimates. I use ***, **, and * to denote significance at the 1%, 5%, and 10% level (two-sided), respectively.

Dep. Var: Long-term book leverage Dep. Var: Long-term

debt

Long-term debt Short and Long-term debt/

Investment (incl. current portion)

Long-term debt Market value assets (1) (2) (3) (4) (5) German*Post Year 0.009 0.010 0.009 0.013 -4.406** (0.019) (0.023) (0.019) (0.034) (1.710) German -0.004 -0.013 -0.004 0.004 0.522 (0.013) (0.016) (0.013) (0.024) (1.187) Lagged ... ROA -0.085*** -0.031*** -0.915*** -0.019* 0.184 (0.006) (0.007) (0.006) (0.011) (0.542) Firm Size 0.009*** 0.011*** 0.009*** 0.020*** 0.396*** (0.002) (0.002) (0.002) (0.003) (0.143) Tangibility 0.267*** 0.334*** 0.267*** 0.382*** 2.085 (0.018) (0.021) (0.018) (0.033) (1.648) Market/book 0.028*** 0.010*** 0.028*** 0.009*** 0.080 (0.002) (0.002) (0.002) (0.004) (0.177)

Year FE Yes Yes Yes Yes Yes

Number of observations 4859 3889 4859 4859 4858 R-squared 0.09 0.08 0.94 0.05 0.01

(23)

23

Table 5: Effect of tax changes and financial constraint on leverage and investment

I estimate standard regressions to test whether, and by how much, firms change their leverage or investments in response to changes in their corporate income tax rate. Post Year is a dummy variable taking the value 1 for the years after a tax increase (2012 until 2015) and 0 otherwise. German is a dummy variable taking the value 1 for German firms (firms in the treatment group) and 0 for Netherlands firms (firms in the control group). German*Post Year is the interaction term of these two variables (the main variable of interest) and measures the average treatment effect of an increase in the corporate income tax rate. German*Post Year*Unconstrained is an interaction variable of the German*Post Year variable and a dummy variable for financial constraint. This dummy takes the value 1 if a firm is classified as unconstrained (according to their dividend payout) and 0 otherwise. Return on assets is measured by dividing the operating income before depreciation by the total assets. Firm size is measured by taking the natural logarithm of the total assets. Tangibility is measured by dividing (net) property, plant and equipment by the total assets. Market/book is measured by multiplying the year-end closing price by the total shares outstanding. This is added to the long- and short-term debt and preferred stock. Next, deferred taxes and investment tax credits are subtracted. The result is divided by the total assets. In column 1, the dependent variable is equal to long-term debt / assets. In column 2, the dependent variable is equal to long-term debt (incl. current portion) / assets. In column 3, the dependent variable is equal to short and long-term debt / assets. In column 4, the dependent variable is equal to long-term debt / market value of assets. In the last column, the dependent variable is equal to investment (defined as capital expenditures divided by the total assets of the previous year.) All regressions are differences-in-differences regressions; the German*Post Year-variable is the differences-in-differences estimator. All specifications include year fixed effects to remove unobserved time-varying shocks. The fixed effects are not reported for brevity. Heteroskedasticity-consistent standard errors are shown in parentheses underneath the coefficient estimates. I use ***, **, and * to denote significance at the 1%, 5%, and 10% level (two-sided), respectively.

Dep. Var: Long-term book leverage Dep. Var:

Long-term debt Long-term debt Short and Long-term debt/ Investment (incl. current portion) Long-term debt Market value assets

(1) (2) (3) (4) (5) German*Post Year 0.018 0.042 0.018 0.012 -9.598*** (0.028) (0.033) (0.028) (0.051) (2.527) German 0.028 -0.022 0.028 0.039 0.576 (0.020) (0.022) (0.020) (0.035) (1.756) Unconstrained (Dividend) 0.017 0.077*** 0.017 -0.030 -0.664 (0.024) (0.029) (0.024) (0.043) (2.158) German*Unconstrained 0.052** 0.031 0.052** -0.050 0.098 (0.026) (0.031) (0.026) (0.048) (2.358) Post Year*Unconstrained -0.012 0.030 -0.012 0.001 -9.577*** (0.035) (0.042) (0.035) (0.063) (3.133) German*Post Year*Un- 0.018 0.055 0.018 0.006 -9.706*** constrained (0.038) (0.046) (0.038) (0.069) (3.427) Lagged ... ROA -0.085*** -0.029*** -0.915*** -0.021* 0.110 (0.006) (0.007) (0.006) (0.011) (0.542) Firm Size 0.012*** 0.016*** 0.012*** 0.027*** 0.481*** (0.002) (0.002) (0.002) (0.003) (0.156) Tangibility 0.271*** 0.336*** 0.271*** 0.388*** 2.125 (0.018) (0.021) (0.018) (0.033) (1.647) Market/book 0.029*** 0.009*** 0.029*** 0.009*** 0.054 (0.002) (0.002) (0.002) (0.004) (0.177)

Year FE Yes Yes Yes Yes Yes

Number of observations 4859 3889 4859 4859 4858

R-squared 0.09 0.09 0.94 0.05 0.01

Dep. Var: Long-term book leverage Dep. Var:

Long-term debt Long-term debt Short and Long-term debt/ Investment (incl. current portion) Long-term debt Market value assets

(1) (2) (3) (4) (5) German*Post Year 0.001 0.014 0.001 0.028 -9.654*** (0.028) (0.032) (0.028) (0.050) (2.486) German 0.019 0.031 0.019 0.034 0.413 (0.019) (0.022) (0.019) (0.034) (1.710) Unconstrained (KZ index) -0.030 -0.011 -0.030 -0.034 -0.156

(24)

24

Table 6: Effect of tax changes and financial constraint on leverage and investment

I estimate standard regressions to test whether, and by how much, firms change their leverage or investments in response to changes in their corporate income tax rate. Post Year is a dummy variable taking the value 1 for the years after a tax increase (2012 until 2015) and 0 otherwise. German is a dummy variable taking the value 1 for German firms (firms in the treatment group) and 0 for Netherlands firms (firms in the control group). German*Post Year is the interaction term of these two variables (the main variable of interest) and measures the average treatment effect of an increase in the corporate income tax rate. German*Post Year*Unconstrained is an interaction variable of the German*Post Year variable and a dummy variable for financial constraint. This dummy takes the value 1 if a firm is classified as unconstrained (according to the Kaplan-Zingales index) and 0 otherwise. Return on assets is measured by dividing the operating income before depreciation by the total assets. Firm size is measured by taking the natural logarithm of the total assets. Tangibility is measured by dividing (net) property, plant and equipment by the total assets. Market/book is measured by multiplying the year-end closing price by the total shares outstanding. This is added to the long- and short-term debt and preferred stock. Next, deferred taxes and investment tax credits are subtracted. The result is divided by the total assets. In column 1, the dependent variable is equal to long-term debt / assets. In column 2, the dependent variable is equal to long-term debt (incl. current portion) / assets. In column 3, the dependent variable is equal to short and long-term debt / assets. In column 4, the dependent variable is equal to long-term debt / market value of assets. In the last column, the dependent variable is equal to investment (defined as capital expenditures divided by the total assets of the previous year.) All regressions are differences-in-differences regressions; the German*Post Year-variable is the differences-in-differences estimator. All specifications include year fixed effects to remove unobserved time-varying shocks. The fixed effects are not reported for brevity. Heteroskedasticity-consistent standard errors are shown in parentheses underneath the coefficient estimates. I use ***, **, and * to denote significance at the 1%, 5%, and 10% level (two-sided), respectively.

Dep. Var: Long-term book leverage Dep. Var:

Long-term debt Long-term debt Short and Long-term debt/ Investment (incl. current portion) Long-term debt Market value assets

(1) (2) (3) (4) (5) German*Post Year 0.001 0.014 0.001 0.028 -9.654*** (0.028) (0.032) (0.028) (0.050) (2.486) German 0.019 0.031 0.019 0.034 0.413 (0.019) (0.022) (0.019) (0.034) (1.710) Unconstrained (KZ index) -0.030 -0.011 -0.030 -0.034 -0.156 (0.024) (0.029) (0.024) (0.043) (2.160) German*Unconstrained -0.041 0.086*** -0.041 -0.054 0.148 (0.026) (0.031) (0.026) (0.047) (2.349) Post Year*Unconstrained -0.011 0.087** -0.011 0.021 -9.110*** (0.035) (0.042) (0.035) (0.063) (3.127) German*Post Year*Un- 0.008 0.065 0.008 0.021 -9.877*** constrained (0.038) (0.045) (0.038) (0.069) (3.422) Lagged ... ROA -0.085*** -0.032*** -0.915*** -0.019* 0.094 (0.006) (0.007) (0.006) (0.011) (0.542) Firm Size 0.010*** 0.011*** 0.010*** 0.021*** 0.371*** (0.002) (0.002) (0.002) (0.003) (0.143) Tangibility 0.170*** 0.200*** 0.170*** 0.275*** 1.945 (0.021) (0.024) (0.021) (0.038) (1.908) Market/book 0.029*** 0.010*** 0.029*** 0.009*** 0.042 (0.002) (0.002) (0.002) (0.004) (0.177)

Year FE Yes Yes Yes Yes Yes

Number of observations 4859 3889 4859 4859 4858

R-squared 0.10 0.11 0.95 0.06 0.01

(25)

25

of Collins (1987), Heider and Ljungqvist (2015), Homaifar et al. (1994), and Titman and Wessels (1988).

Tangibility is positively significant at the 1% level in regressions 1-4, but insignificant in regression 5. This significant positive relationship is consistent with the findings of Almeida and Campello (2007), and Heider and Ljungqvist (2015). The market-to-book ratio (a measure of investment opportunities) is positively significant at the 1% level in regressions 1-4, but insignificant in regression 5. This significant positive relationship is consistent with the findings of Almeida and Campello (2007) and Erickson and Whited (2000). The R-squared is highest in the third regression (0.94), and lowest in the fifth regression (0.01).

Tables 5 and 6 provide further information about the impact of financial constraint. As explained in the methodology chapter, I use two different measures of financial constraint: dividend payout and the KZ index. Table 5 shows the regression results where I use dividend payout, and Table 6 the results where I use the KZ index. Both tables comprise five columns, representing the same dependent variables as in Table 4.

Table 5 displays similar regression results to those of Table 4: the average treatment effect is not significant in regressions 1-4. Moreover, I find no evidence for a significant effect for unconstrained firms. These results are not consistent with the findings of Denis and Sibilkov (2010), Fazzari et al. (1988), Hennessy and Whited (2007), Lamont et al. (2001) and Whited (1992), or with hypothesis H1b. Column 5 indicates a significantly negative ATE for unconstrained German firms (at the 1% level). This result is consistent with the findings of Denis and Sibilkov (2010), Fazzari et al. (1988), Hennessy and Whited (2007), Lamont et al. (2001) and Whited (1992), and with hypothesis H2b. The results for the control variables are similar to the results in Table 4. The R-squared is highest in the third regression (0.94), and lowest in the fifth regression (0.01).

Table 6 displays similar results to those of Tables 4 and 5: the average treatment effect is not significant in regressions 1-4. These results are inconsistent with the findings of Graham (1996), Heider and Ljungqvist (2015), MacKie-Mason (1990), Rajan and Zingales (1995), and with hypothesis H1a. However, they are consistent with those of Kraus and Litzenberger (1973), and with the tradeoff theory. Column 5 displays a significantly negative ATE for German firms (at the 1% level). This is consistent with the findings of Auerbach and Hassett (1991), Cummins and Hassett (1992), and Summers et al. (1981), and with hypothesis H2a. However, this is not consistent with the findings of Hall and Jorgenson (1969). Table 6

(26)

26

illustrates no significant ATE for unconstrained firms in regressions 1-4. These results are not consistent with the findings of Denis and Sibilkov (2010), Fazzari et al. (1988), Hennessy and Whited (2007), Lamont et al. (2001) and Whited (1992), nor with hypothesis H1b. Column 5 displays a significantly negative ATE for German unconstrained firms (at the 1% level). These results are consistent with the findings of Denis and Sibilkov (2010), Fazzari et al. (1988), Hennessy and Whited (2007), Lamont et al. (2001) and Whited (1992), and with hypothesis H2b. The results for the control variables are similar to those in tables 4 and 5. The R-squared is highest in the third regression (0.94), and lowest in the fifth regression (0.01).

5.2 Explanation

Columns 1-4 of tables 4, 5 and 6 indicate no evidence for an effect of corporate taxes on leverage. All co-efficients are insignificant at the 1%, 5% and 10% levels. A possible reason for these insignificant results is that the tax increases were not large enough to incentivise German firms to raise their debt levels: the German corporate income-tax rate increased with 0.1, 0.05 and 0.07 percentage points (see Figure 1) in the sample period. Moreover, other studies (e.g. Heider & Ljungqvist, 2015) investigating increases in the top corporate income-tax rate conclude that firms may respond more strongly to changes in the top corporate-income-tax rate than to changes in the total corporate tax rate. Moreover, selection bias or a lack of relevant data in the sample are also possible explanations for the insignificant results in all regressions.

A further explanation for these results is the fact that firms have different levels of tax exposure. If firms have substantial operations (and hence tax liabilities) outside the country where they are headquartered, the regressions will underestimate the sensitivity of debt and investment to taxes. A firm would have a stronger response to taxation if it were operating only in the country where it is headquartered.

In column 5 of Table 4, it is indicated that German firms decreased their investments with €4.406 million after a tax increase. In column 5 of Table 5, it is indicated that German constrained firms decreased their investments with €9.598 million after a tax increase, while German unconstrained firms decreased theirs with €19.304 million. Similarly, in column 5 of Table 6, it is indicated that German constrained firms decreased their investments with €9.654 million after a tax increase, while German unconstrained firms decreased theirs with €19.531 million.

(27)

27

The figures in columns 1 and 3 of Table 5, indicate that German unconstrained firms have leverage ratios that are 5.2% higher than firms in the Netherlands or constrained firms. This effect is significant at the 5% level. Column 2 of Table 5 indicates that unconstrained firms have leverage ratios that are 7.7% higher than constrained firms. This effect is significant at the 1% level. In column 2 of Table 6, it is indicated that German unconstrained firms have leverage ratios that are 8.6% higher than firms in the Netherlands or constrained firms. This effect is significant at the 1% level. Column 2 of Table 6 indicates that unconstrained firms raise their leverage ratio with 8.7% in the years after a tax increase. This effect is significant at the 5% level.

According to tables 4, 5, and 6, financial constraint does not have an additional impact on leverage. This differed from expectations. A potential cause for the insignificant results obtained in this chapter is that dividend payout and the KZ index do not (perfectly) measure financial constraint. Farre-Mensa and Ljungqvist (2015) studied the effectiveness of the five most commonly used financial-constraint measures in debt and equity markets (including dividend payout and the KZ index). They found that firms classified as constrained have no trouble raising debt in response to a tax increase and engage in equity-recycling. This suggests that firms classified as constrained do not behave as if they were constrained. Hadlock and Pierce (2010) evaluated methods commonly used in the literature to measure financial constraints (especially the KZ index). The five components of the KZ index (described in Chapter 2) are all regressed to a firm’s categorised level of constraints. In the estimated ordered logit models, only two of the five KZ index components, cashflow and leverage, are consistently significant with the sign according to the KZ index. For two other components, Tobin’s Q and dividends, the co-efficients have changing signs across estimated models and are insignificant in many cases. The fifth component, cash holdings, has a positive and significant impact in the models predicting constraints, where the sign in the KZ index is negative.

According to the researches of Farre-Mensa and Ljungqvist (2015), and Hadlock and Pierce (2010) it is possible that financial constraint measures do not perfectly measure financial constraint. A better measure of financial constraint is probably the method Kaplan and Zingales (1997) employed: classifying firms on the basis of the discussion of financial constraints in their 10-K statements. This is, however, a time-consuming method.

(28)

28

6. Robustness Checks

This chapter provides robustness checks and presents additional results. The first section shows the results in three tables, and the following section provides further explanation.

6.1 Empirical Results

Tables 7, 8 and 9 provide an overview of the main results of the regressions. These regressions measure the impact of an increase in the corporate income tax rate on the leverage ratio and investments of German firms. Moreover, Tables 8 and 9 show the additional impact of financial constraint. Table 8 shows the regression results where I use dividend payout and Table 9 shows the regression results where the Kaplan-Zingales index is used. All tables comprise five columns, representing the same dependent variables as in Table 4. In this section, I also include firm-fixed effects in order to exclude unobserved firm characteristics that remain constant over time.

Table 7 shows a positive ATE for German firms in the first four columns, and a negative ATE in the fifth. However, the treatment effect is not significant in all regressions. The positive ATE in the first four regressions is consistent with the findings of Graham (1996), Heider and Ljungqvist (2015), MacKie-Mason (1990), Rajan and Zingales (1995), and and with hypothesis H1a. The negative ATE in the fifth regression is consistent with the findings of Auerbach and Hassett (1991), Cummins and Hassett (1992), and Summers et al. (1981), and with hypothesis H2a. However, this is not consistent with the findings of Hall and Jorgenson (1969). The insignificant ATE in all regressions is consistent with the findings of Kraus and Litzenberger (1973), and with the tradeoff theory.

Return on assets is significant in regressions 1, 2 and 5 (at the 1% level), and in regression 3 (at the 10% level); the effect on leverage is negative and the effect on investment is positive. This is consistent with the findings of Heider and Ljungqvist (2015), Myers (1984), Myers and Majluf (1984), Ozkan (2001), and with the pecking order theory. Firm size is positively significant at the 1% level in all regressions except regression 3). This is consistent with the findings of Collins (1987), Heider and Ljungqvist (2015), Homaifar et al. (1994), and Titman and Wessels (1988).

Tangibility is positively significant at the 1% level in regressions 1, 2, 3 and 5, but insignificant in regression 4. The significant positive relationship is consistent with the findings of Almeida and Campello (2007) and Heider and Ljungqvist (2015). The market-to-book ratio

Referenties

GERELATEERDE DOCUMENTEN

The effective tax rate (ETR) is a widely used measure for the tax burden borne by companies and can be defined as corporate income taxes divided by income before

Model 4 presents the separate effect of corporate income tax rate and personal income tax on both interest (PITI) and dividends (PITD) with 2 lags to assess whether there is

The independent variables are: leverage t−1 , a lagged level of the dependent variable; effective tax rate, a ratio of income taxes to earnings before taxes; intangibility, a ratio of

Table 2 reports the descriptive statistics for all the variables used in the full sample, which are the Tobin’s Q-ratio, return on assets (ROA), ES (environmental and

Column 1 provides consistent results with the first hypothesis stating that under normal conditions (i.e. no crisis), the interaction variable is positive and

The effect of a country's economic development on the relationship between the three forms of ownership and tax avoidance is supported and implies that the higher (lower) the

For the moderating effect, it is statistically significant that firm size will strengthen the association between effective tax rate rise and firm risk-taking, which means that

The results from the robustness test show that social CSR activities are negatively related to tax avoidance when using the total book-tax difference as a proxy measure