Corporate effective tax rates and tax reform: evidence from the Netherlands

(1)

Corporate effective tax rates and tax

reform: evidence from the Netherlands

Jelle E. Tjerkstra

♣

April 2013

Abstract: This study focuses on determinants of the variability of corporate effective tax rates (ETRs) in the Netherlands spanning the Werken aan Winst (WaW) tax reform in 2007. This reform reduced corporate tax rates and broadened the tax base in line with recent trends in taxation in the European Union. Using statistical analysis, attention was paid to firm characteristics such as size, capital structure, asset mix and profitability and the consequences for these relationships after the WaW tax reform. The results show a non-linear relationship between firm size and ETR. Small firms face higher tax burdens as they grow, but the relationship is reversed for larger firms as they can benefit more from tax planning. The results also show a negative relationship between leverage and ETR and fixed assets and ETR. A positive relationship is found for the relationship between profitability and ETR. After the tax reform the overall tax burden is lowered from 29.1% in the period 2004-2006 to 23.7% in the period 2008-2010.The tax burden is shifted from small and profitable companies to larger and less profitable companies, as in line with the goals of the tax reform.

Key words: Corporate taxation, effective tax rate, tax reform.

♣

(2)

1. Introduction

Most business decisions have tax implications and taxation might have a significant impact on corporate decisions. Companies should take the tax burden of investment opportunities into account when designing their strategies. Taxes are of major importance when choosing a new location for a firm’s headquarters or when designing a new financing structure of a firm. Global firms aim to reduce their ETRs by taking advantage of the differences between national tax systems. Many companies use tax planning to create a competitive advantage. For example, General Electric has 1,000 employees working at its tax department. Its tax department is viewed as a profit center, not as a cost center. The global financing operations effectively reduce the taxes paid from its operational activities, making the global conglomerate more competitive (New York Times, 2012).

Recently, a lot of multinationals have been criticized for their aggressive tax planning and corresponding low tax burdens (BBC News, 2012). One of the most prominent examples is Starbucks and its low tax burden in the United Kingdom. According to research of press agency Reuters (2012), Starbucks paid only 8.6 million pounds of income taxes in the last 15 years, while reporting turnover of more than 3 billion pounds. 91 of the 100 largest companies in the world are present in the Netherlands, predominantly for tax purposes according to research of the Dutch newspaper De Volkskrant (2013). 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 taxes. Apple reported an ETR as low as 1.9% for their non US income as a result of tax planning via the Netherlands (BBC News, 2012).ETRs are often used by policy makers and interest groups as a tool to make inferences about the corporate tax systems. This research aims to provide evidence on the firm characteristics of companies reporting relatively low tax burdens and is focused on companies located in the Netherlands.

(4)

2 remainder of profit (exceeding €60,000) was lowered to 25.5%. One of the aims of the introduction of these lowered rates was to shift tax burdens from small and medium sized companies (SMEs) to multinational enterprises (MNEs). The newly introduced patent box provides a special tax regime for qualifying income from R&D activities and aims to attract and retain innovative activities. Also, limitations were introduced for the tax deductibility of interest and depreciation expenses in order to broaden the tax base (Memorie van Toelichting, 2006). The changes can be seen in line with tax reforms in other industrialized countries where lowered tax rates are financed by a broadened tax base.

The goal of this research is to identify the determinants of the variability in corporate effective tax rates in the Netherlands during the years 2004-2006 and 2008-2010 in order to investigate the consequences of the changes in the Corporate Income Tax Act 1969 which came into force per 1 January 2007. This research builds on earlier research on ETRs, which has mainly focused on US companies. There is a lack of research on ETRs and its determinants in the Netherlands. No earlier research has been done on the effects of tax reforms in the Netherlands.

The Dutch tax authorities did a descriptive study on the differences between the financial accounting tax expense and actual taxes paid by companies (Belastingdienst (1997)). It does not provide information on the differences observed in relation to firm characteristics. Janssen (2005) did a study on the relation between firm characteristics and ETRs of companies located in the Netherlands during the years 1994-1999. He found small differences between the statutory tax rates and effective tax rates. These differences can be explained by firm characteristics to a very small extent.

This study is similar to the research of Richardson and Lanis (2007) on the effect of the Ralph Review tax reform of corporate taxation in Australia. The new regulations became operative in the year 2000 and aimed to broaden the taxable base and reduce the tax rate. The reform measures were very similar to the measures introduced by the Bill Werken aan Winst. Richardson and Lanis (2007) find that the average effective tax rate is lower after the reform, especially for firms with relatively low debt levels. Capital intensive firms face increasing ETRs due to the removal of accelerated depreciation. These results were in line with the goals of the Ralph Review tax reform. Gupta and Newberry (1997) did a similar research on the tax reform in the United States on the Tax Reform Act of 1986. They cast doubt on the actual consequences of the Tax Reform Act 1986 which are not in line with the foreseen goal to establish a level playing field for all companies (irrespective of size or asset mix).

(5)

3 To what extent is the corporate effective tax rate of Dutch companies related to firm characteristics spanning the “Werken aan Winst” tax reform?

The following sub questions will be answered in this study:

- How do companies need to account for corporate income taxes in their financial statements? - What are the different calculation methods used to measure corporate effective tax rates? - What are the results of previous empirical research on the determinants of the corporate

effective tax rate?

- What are the backgrounds and goals of the measures introduced by the “Wet Werken aan Winst” in the Corporate Income Tax Act 1969?

An empirical study is used to answer the research question. This is in line with previous research focusing on ETRs. The data will be obtained from the Orbis database. The Orbis database covers over 100 million companies around the world. Company financials are presented in a standardized format in order to increase the comparability of financial statements. Orbis contains financial information on both listed and unlisted companies. The dataset will cover the years 2004-2006 and 2008-2010. The year 2007 will be excluded because prior research shows that firms normally respond to tax legislation changes one year after tax legislation becomes operative (e.g. Dhaliwal and Wang (1992); Scholes et al. (1992)). This empirical model is in line with the methods Richardson and Lanis (2007) used to test the Australian corporate income tax reform. They pooled all the data in one dataset and performed an ordinary least squares regression (OLS).

The focus of this study is only on the corporate income tax. Other studies focused on the shift from one tax to another, for example the shift from direct taxes to indirect taxes (Eurostat, 2011). The possible corresponding changes in other tax laws (e.g. personal income tax) are also not taken into account. Also, no distinction will be made for different legal forms. This is what Oestreicher et al. (2008) did in their study on the reform of business taxation in Germany, focusing on effective tax rates for different legal forms. They found that German business became more competitive from a tax perspective as effective tax burdens were lowered after the reform in 2007. Effective tax burdens were equalized for different legal forms as well as for national and international groups active in Germany.

(6)

4 This study only uses data from financial statements as information from the tax authorities is not publicly available.

(7)

5

2. Literature on effective tax rate and its determinants

This chapter aims to provide an overview of the literature on effective tax rates and its determinants. This chapter first focuses on ETR as a measure of the tax burden of a company and subsequently on the relation between firm characteristics and ETR. On the basis of existing literature, several hypotheses will be formulated which will be used in the empirical part of this paper.

2.1 Effective tax rate as a measure of the tax burden of a company

2.1.1 Marginal and average effective tax rates

The economic literature offers a range of approaches to compute corporate effective tax rates. Nicodeme (2001) distinguishes between studies on a macro level and on a micro level. Macro studies use aggregate data published by national or international statistics institutes. The goal of these studies is to provide high-level information on tax burdens of companies (Callihan, 1994). The effective tax burden of a country is measured by the ratio of taxes paid by corporations on a measure of the tax base. The tax base can be calculated as the corporate gross operating surplus or the aggregate corporate profit. Aggregate data are widely available and the ratio can be easily calculated. These aggregate data suffer from a few important shortcomings. The most important shortcoming is the mismatch between the numerator and denominator of the ratio. Corporate operating surplus may include interests, rents, and royalties paid by corporations while these sources of income may be taxed on a personal income tax level instead of on a business taxation level. This will lead to an underestimation of effective taxation.

On a micro level, studies focus on the tax burden of individual companies. Data are obtained from large databases containing financial information of firms all over the world. Two kinds of empirical ETR research can be distinguished: research on marginal ETRs and on average ETRs. The marginal ETR focuses on new investment opportunities. Marginal ETR is the rate of tax paid on an additional unit of income from a specific investment project. Studies using marginal ETRs are forward looking: it is relevant to know the marginal tax burden before investing in a new project. The marginal rate can also be relevant for investigating tax effects on financing decisions when new funds have to be attracted.

(8)

6 taxation. An advantage of using average ETRs compared to marginal ETRs is that it uses real life data. Actual tax burdens are used and no estimates for tax consequences have to be calculated. This allows researchers to make inferences about the equitability of the tax system and how the tax burdens are spread over different firms. Studies using average ETRs can also be used to identify possible relationships between the characteristics of a firm and its effective corporate taxation (Nicodeme, 2001).

2.1.2 Measures of the average effective tax rate

Fullerton (1984) provides a simple definition of the average effective tax rate: last year’s tax divided by last year’s income. Such a broad definition provides a great deal of latitude in the measurement of these two components. This will lead to various tax burden estimates as researchers use different accounting concepts for income and tax burden. The general calculation of average ETR of a company is the ratio of financial accounting tax expense (TE) over pre-tax financial accounting income (PTI) of a particular firm i (Callihan, 1994). TE can be defined as taxable income (TI) times the statutory tax rate (t). Thus:

=

(1) Omer et al. (1991) note that specification of a particular ETR measure depends on the research issue and the source of information used to provide estimates of tax and income. Financial reports are generally the only publicly available source for obtaining tax and income measures for individual firms. Spooner (1986) notes several problems associated with using financial statement information to calculate ETRs. These include among others industry differences, inclusion of other taxes in the tax expense account and treatment of foreign taxes. Greater disclosure of financial information for determining the present value of deferred tax assets and liabilities would improve ETR estimates. Omer et al. (1991) conclude that alternative ETR measures cause notable shifts in estimated ETRs and its estimated associations with firm characteristics.

(9)

7 Permanent differences will lead to elimination of part of the taxes payable and will not lead to any deferred tax assets or liabilities. (Sutton, 2004; p. 530-540).

Omer et al. (1991) show that systematic differences in the reporting of deferred tax liabilities can affect the estimation of ETRs. When holding income constant, estimated ETRs differ on average two to three percentage points because of differences in the measurement of deferred taxes. Including deferred taxes in the numerator has two possible shortcomings. Omer et al. (1991) show that deferred taxes are not always reversed; in such cases ETR is overstated. Also, companies are very prudent in reporting deferred tax assets – often as a result of taxable losses – since it is often uncertain whether they can recoup the losses (Sutton, 2004 p.545). The time value of deferred taxes is not taken into account, thereby also possibly overstating ETR. On the other hand, excluding deferred taxes also has several drawbacks. The tax expense remains an estimate of the actual tax burden which is prone to calculation errors and wrong assumptions. Also, certain non-codified tax rulings might have a lowering effect on the actual tax burden, but might not be reflected in the tax expense account (Janssen, 2000). Moreover, companies might have an incentive to show reasonable tax burdens in their financial statements. Reporting low tax burdens might provoke actions by tax authorities or interest groups like the Tax Justice Network or The Citizens For Tax Justice (see e.g. Citizens For Tax Justice, 1984, 1985, 1986) or TaxJusticeNL.1

The measures of income used in the denominators can be divided into three categories (Plesko, 2003). The first category tries to mimic taxable income. If taxable income is used, the ETR will only differ from the statutory rate for two reasons. First, corporate tax rates can be graduated (different rates for various tax brackets) and income below a certain threshold may be exempted. Second, current or past activities may generate tax credits or extra deductions that reduce the amount of taxes due. These can be extra deductions for particular investments or the offset of losses from previous years. The second category uses a (variant of) pretax book income, like earnings before income taxes (EBT) or earnings before interest and taxes (EBIT). If financial accounting income is used, the impact of tax preferences will also be taken into account. A firm might engage in specific activities due to a beneficial tax treatment. The third category uses operating cash flow as the denominator when calculating ETRs. This measure potentially controls for systematic differences in accounting method choices which are related to firm size (Zimmerman, 1983).

(10)

8 2.1.3 Alternatives to effective tax rate

Plesko (2003) criticizes the use of average ETRs based upon financial statement information. He suggests that researchers should choose to examine and utilize the firms’ tax return information rather than rely on information in financial statements. Plesko uses confidential tax return data to evaluate alternative measures of corporate tax rates. Plesko makes far-stretching inferences from his results and claims that ETR measures based on financial statements provide little insight about annual corporate tax burdens and may introduce substantial bias into statistical models. Shevlin (1999) disagrees with Plesko and believes the appropriate measure of ETR depends on the context and research question being examined. The purpose of calculating financial statement income (as a measure of operating performance) is different than the purpose of calculating taxable income (as a base on which taxes are calculated). If one wishes to relate current taxes payable to the firm’s actual performance, a strong argument can be made for financial statement based measures of earnings. This is what the Citizens For Tax Justice (1984, 1985, 1986) did in a widely cited series of studies, highlighting that many large firms were paying very low taxes while reporting large earnings to shareholders. Any study wishing to evaluate these studies can legitimately examine financial statement based ETRs. If one wishes to study the relationship between firms’ (financial) characteristics and the amount of taxes paid relative to their reported earnings, financial statement ETRs are also more appropriate to compare firms. Financial statement information is widely available and therefore a comparison can be made between different firms. All other financial characteristics of a firm are derived from the financial statements of a firm and a financial statement based ETR increases comparability. This explicitly holds for a possible relation between profitability and ETR since taxable income might significantly differ from financial statement income due to permanent and temporary differences. If financial statement income is used, also financial accounting tax expense should be used, following the same line of reasoning as above.

Hafkenscheid (2010) believes that the ETR is flawed as a measure of the tax burden of a company and the value created by a company’s tax department. Companies should use the Time & Risk Adjusted Cash Tax Ratio (TRACTR) to measure the performance of a tax department. The ETR is based on the company’s profit and loss account based on IFRS or a similar accounting standard, whereas TRACTR focuses on cash flows. Hafkenscheid notes three major shortcomings of ETR:

1. The time value of money is disregarded. All cash flows – future and present – are calculated on a nominal basis rather than on a net present value basis.

(11)

9 for unless the position is more likely than not to be disallowed by the tax authorities. This may lead to misleading estimates of the economic value of a tax position.

3. Accounting for deferred taxes depends on shock factors which are difficult to predict. For example, loss carry over rules can be changed or tax rates may be altered over the years. This can have significant consequences for the accounting of deferred taxes.

The primary focus of TRACTR is on the performance of the tax department of a company and not on the (relative) tax burden of a company. TRACTR is defined as the ratio between the actual tax cash flow and the cash tax flow that would be payable if the company did not engage in tax planning (‘natural cash flow’). All cash flows are calculated on the basis of net present value. TRACTR is more suited to measure the (relative) performance of tax departments and the extent to which companies engage in tax planning. It is difficult to compare companies because a lot of (confidential) financial data is needed to retrieve actual and ‘natural’ cash flows since it is not possible to retrieve ‘natural’ cash flows from publicly available financial statements. The TRACTR is more suitable for internal use than for a comparison between multiple firms.

2.2 Determinants of the effective tax rate

Research on the determinants of ETRs begins with Stickney and McGee (1982) and Zimmerman (1983). The results show that differences in ETRs are related to a number of firm characteristics, such as size, capital structure and asset mix. The remainder of this chapter will focus on the results of prior research regarding determinants of ETRs. Subsequently, the hypotheses will be formulated for the empirical part of this study.

2.2.1 Firm size and effective tax rate

Company size is the most widely used variable in prior research on the determinants of ETRs. The debate has been dominated by the political cost and political power hypotheses from the early 1980s. This discussion will be summarized first, and subsequently the consequences of recent developments in the global economy for tax strategies will be discussed. As firms become more global they are confronted with a wide range of tax systems of different countries. Global firms can benefit from differences in tax systems of countries by organizing their activities in a tax friendly manner.

(12)

10 ETRs (Zimmerman, 1983). Scholes and Wolfson (1992) argue that large and mature companies may find it difficult to aggressively pursue tax planning, because that may interfere with other tax related contracts. An opposite – and more dominant – stream of papers shows a negative relation between size and tax burden (e.g. Porcano (1986); Kim and Limpaphayom (1998), CTJ (1986)). These results support the view that larger firms have lower ETRs because they have more political power. Siegfried (1972) argues that these firms have substantial resources available to manipulate the political process in their favor. Also, they have more possibilities to structure their activities in a tax friendly manner. All these studies used different empirical procedures, including time periods, sample selection, ETR definitions and firm-size proxies. Wilkie and Limberg (1990) did an attempt to reconcile the results of Zimmerman (1983) and Porcano (1986) and found that their different result were largely due to differences in the empirical methods used. Kern and Morris (1992) extended the research of Wilkie and Limberg (1992) and found that Zimmerman’s results were more robust to a series of empirical procedures than Porcano’s (1986) results. Gupta and Newberry (1997) asserted that there is no general relationship between firm size and ETR since the divergent results show that firm-size effects are sample specific. Richardson and Lanis (2007) contradict this view by arguing that all recent research outside the US on the relationship between firm characteristics and ETR show a negative relationship, thus supporting the political power hypothesis. Only the results of US research show the tendencies as documented by Gupta and Newberry.

Another argument for a negative relationship between firm size and ETR is that larger firms have more possibilities to engage in tax-planning and organize their activities in order to achieve optimal tax savings. This is supported by the results of Fernandez and Martinez (2011). They test whether a non-linear relationship might exist between the size of a firm and the reported ETR. Although their sample consists only of large-size, listed firms, the results show that there is a turning point leading to a change of sign in the “size-ETR relation” after a certain point. They find that relatively small firms face a greater tax burden as they grow. After the turning point, they find an opposite relationship: the larger the firm, the lower its ETR. These results might indicate that small firms are stimulated by tax incentives, but as they grow they do not longer fall within the scope of the measures supporting SMEs. As a firm becomes even larger and has activities in several countries it might develop a tax strategy focusing on a tax efficient corporate structure supported by external tax advisors. This will eventually lead to a drop in the firm’s ETR.

As the results of previous research are inconclusive regarding the relation between firm size and ETR, I propose a non-directional hypothesis. Formally:

(13)

11 2.2.2 Capital structure and effective tax rate

The relation between a firm’s capital mix and tax burden has been extensively studied in the literature. Modigliani and Miller (1958) demonstrate that the value of a firm is independent of capital structure, assuming no taxes and transactions costs. Firms’ financing decisions are however influenced by tax considerations. Tax regulations normally allow differential tax treatment for the capital structure of firms. Interest is generally tax deductible, whereas dividends are generally not deductible. There is however a wide range of exceptions to this general treatment. Some countries allow a deduction for equity (e.g. the Belgian notional interest deduction) and a lot of countries limit the deductibility of interest. This This could reduce the cost of debt financing in comparison with equity financing. In a later paper, Modigliani and Miller (1963) argue that capital structure affects the value of a firm when accounting for taxes. They model the value of a firm as an increasing function of the amount of debt in the firms’ capital:

= + ₍₂₎ where Vl is the value of a levered firm, Vu the value of an unlevered firm, D the value of the firm’s

outstanding debt and t the firm’s average corporate income tax rate. The value of a firm increases by the product tD, also called the tax shield. In practice, applying formula will not lead to an exact estimation of the value of a company.

(14)

12 After this turning point, the coefficient has an opposing sign and a positive relationship exists between ETR and leverage. Firms with high marginal tax rates have an incentive to increase their debt levels as long as the interest on the debt is tax deductible. It should however be stressed that tax is not the only determinant of the leverage of a firm, since it is also highly dependent of the activities, culture and risk level of a firm. However, this goes beyond the scope of this study.

Given the opposite research streams regarding the relationship between leverage and ETR, I propose a non-directional hypothesis. Formally:

H2: ETRs are associated with firm leverage. 2.2.3 Asset mix and effective tax rate

ETRs might also be influenced by operational activities and the asset mix of a firm. Prior research generally shows that a greater weight of tangible fixed assets (property, plant and equipment) negatively influences the ETR (e.g. Stickney and McGee,1982; Gupta and Newberry, 1997; Richardson and Lanis, 2007). Most tax regimes permit taxpayers to write off the cost of depreciable assets over periods shorter than their economic lives. Also, governments stimulate investments by granting superdeductions for specific investment categories (e.g. innovation or environment friendly investments). Firms investing more in long term fixed assets are expected to have lower ETRs. However, the sector of activity determines to a certain extent the asset mix. Real estate companies will have higher levels of tangible fixed assets than internet startups for example. The possibility to invest in fixed assets depends on the volume of current assets that firms need for their activity. Investment in inventories is considered as an alternative way of using funds to the investment in property, plant and equipment and limits the possibility of reducing ETR. Inventory intensity could therefore be considered to lead to a higher ETR. Empirical research supports the positive and statistically significant relationship between the level of inventory and ETR (Gupta and Newberry, 1997; Richardson and Lanis 2007; Fernandez-Rodriguez and Martinez-Arias, 2011). Fernandez and Martinez (2011) consider the possibility of a non-linear relationship between the asset mix and effective tax rate. Their results show that companies with a ratio of property, plant and equipment over total assets below 37.98% are subject to a greater tax burden, but from that level ETRs decrease as capital intensity rises.

(15)

13 between R&D activities and ETR. A lot of studies on the determinants of ETRs do not include a measure of R&D activities. This may be due to difficulties measuring innovation or R&D activities within a company or the availability of data.

As both inventory intensity and capital intensity can be seen as a proxy for a firm’s asset mix, two hypotheses regarding the relationship between asset mix and ETR are formulated. Formally:

H3:ETRs are associated with firm capital intensity. H4: ETRs are associated with firm inventory intensity. 2.2.4 Firm profitability and effective tax rate

Firm profitability is the most obvious determining factor for the tax burden of a company. The most profitable firms have larger profits, so they pay tax every year. Less profitable firms will show lower profit levels and will have a lower tax burden. Less profitable firms are also more prone to incur losses and can reduce their tax liability by carrying over losses. This reasoning represents a benefit in terms of tax burden for firms incurring losses. The results of previous empirical research on this matter support this positive relationship (Stickney and McGee, 1982; Wilkie and Limberg, 1990; Plesko, 2003; Richardson and Lanis, 2007; Chen, 2010). This leads to the following hypothesis:

H5: ETRs are positively associated with firm profitability.

(16)

14

3. “Werken aan Winst” reform of the Corporate Income Tax

Act 1969

3.1 Trends in corporate taxation in the European Union

Considerable reforms to corporate income tax regimes have taken place in industrialized countries during the last three decades. Statutory tax rates have fallen from an average of 48% in the early 1980s to 23.5% in 2012 in the EU member states. Table 1 shows an overview of the statutory corporate tax rates over time for the EU Member States. The EU appointed Ruding Committee proposed a minimum statutory tax rate of 30% in 1992. Two decades later, only five member states have a statutory rate higher than the proposed minimum (Eurostat, 2011). There is wide interest in the development of corporate income taxes for a number of reasons. The increase of capital mobility has raised concerns among policymakers and interest groups that high levels of taxation might reduce investments and create incentives to relocate book profits (Nota, 2005). Low tax burdens are seen as an instrument for attracting foreign capital investments by offering an attractive tax treatment. Taxes may have distortive effects on the market, particularly in highly integrated areas like the European internal market. These factors are the driving force for the European Commission to stimulate convergence among Member states in the taxation of corporations (Prammer, 2011). Also, decisions of the Court of Justice regarding the treatment of domestic and cross-border activities have reduced the freedom of individual Member States in the design and application of their corporate tax codes.

Two prominent trends are recognizable in corporate income tax codes in the EU. First, Member States have moved towards lowering tax rates. The cut in corporate tax rates started in the early 1980s but the trend has weakened in recent years. The financial and economic crisis that started in 2008 has resulted in a significant deterioration of public finances across all Member States. As a result, some Member States have introduced a (temporary) crisis levy. Broadening the taxable base is the second trend in corporate taxation. Fiscal depreciation rules have become less generous and strict limitations are introduced for deductions. This trend is also due to the Code of Conduct for business taxation and the strict EU rules on State aid to enterprises. The policy of reducing rates while broadening the tax base is usually referred to as ‘tax rate cut cum base broadening reforms’ (Eurostat, 2011). Belgium forms a striking exception to this general trend by introducing an allowance for corporate equity, called notional interest deduction.2 Devereux et al. (2008) investigated how governments respond to changes in other countries’ tax systems. They did this after significant

2

(17)

15 reforms to corporation taxation in industrialized countries. They found evidence that countries actually respond to each other in terms of tax legislation. They found two forms of tax competition: over statutory tax rates for mobile profits and over effective marginal tax rates for capital. Chau Le Thi Nguyet (2006) reached the same conclusion in her dissertation for the countries Malaysia, the Philippines and Vietnam.

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 '97-'10 BE 40.2 40.2 40.2 40.2 40.2 40.2 34.0 34.0 34.0 34.0 34.0 34.0 34.0 34.0 -6.0 BG 40.2 37.0 34.3 32.5 28.0 23.5 23.5 19.5 15.0 15.0 10.0 10.0 10.0 10.0 -30.2 CZ 39.0 35.0 35.0 31.0 31.0 31.0 31.0 28.0 26.0 24.0 24.0 21.0 20.0 19.0 -20.0 DK 34.0 34.0 32.0 32.0 30.0 30.0 30.0 30.0 28.0 28.0 25.0 25.0 25.0 25.0 -9.0 DE 56.7 56.0 51.6 51.6 38.3 38.3 39.6 38.3 38.7 38.7 38.7 29.8 29.8 29.8 -27,0 EE 26.0 26.0 26.0 26.0 26.0 26.0 26.0 26.0 24.0 23.0 22.0 21.0 21.0 21.0 -5.0 IE 36.0 32.0 28.0 24.0 20.0 16.0 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 -23.5 EL 40.0 40.0 40.0 40.0 37.5 35.0 35.0 35.0 32.0 29.0 25.0 25.0 25.0 24.0 -16.0 ES 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 32.5 30.0 30.0 30.0 -5.0 FR 41.7 41.7 40.0 37.8 36.4 35.4 35.4 35.4 35.0 34.4 34.4 34.4 34.4 34.4 -7.3 IT 53.2 41.3 41.3 41.3 40.3 40.3 38.3 37.3 37.3 37.3 37.3 31.4 31.4 31.4 -21.8 CY 25.0 25.0 25.0 29.0 28.0 28.0 15.0 15.0 10.0 10.0 10.0 10.0 10.0 10.0 -15.0 LV 25.0 25.0 25.0 25.0 25.0 22.0 19.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 -10.0 LT 29.0 29.0 29.0 24.0 24.0 15.0 15.0 15.0 15.0 19.0 18.0 15.0 20.0 15.0 -14.0 LU 39.3 37.5 37.5 37.5 37.5 30.4 30.4 30.4 30.4 29.6 29.6 29.6 28.6 28.6 -10.7 HU 19.6 19.6 19.6 19.6 19.6 19.6 19.6 17.6 17.5 17.5 21.3 21.3 21.3 20.6 1.0 MT 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 0.0 NL 35.0 35.0 35.0 35.0 35.0 34.5 34.5 34.5 31.5 29.6 25.5 25.5 25.5 25.5 -9.5 AT 34.0 34.0 34.0 34.0 34.0 34.0 34.0 34.0 25.0 25.0 25.0 25.0 25.0 25.0 -9.0 PL 38.0 36.0 34.0 30.0 28.0 28.0 27.0 19.0 19.0 19.0 19.0 19.0 19.0 19.0 -19.0 PT 39.6 37.4 37.4 35.2 35.2 33.0 33.0 27.5 27.5 27.5 26.5 26.5 26.5 29.0 -10.6 RO 38.0 38.0 38.0 25.0 25.0 25.0 25.0 25.0 16.0 16.0 16.0 16.0 16.0 16.0 -22.0 SI 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 23.0 22.0 21.0 20.0 -5.0 SK 40.0 40.0 40.0 29.0 29.0 25.0 25.0 19.0 19.0 19.0 19.0 19.0 19.0 19.0 -21.0 FI 28.0 28.0 28.0 29.0 29.0 29.0 29.0 29.0 26.0 26.0 26.0 26.0 26.0 26.0 -2.0 SE 28.0 28.0 28.0 28.0 28.0 28.0 28.0 28.0 28.0 28.0 28.0 28.0 26.3 26.3 -1.7 UK 31.0 31.0 30.0 30.0 30.0 30.0 30.0 30.0 30.0 30.0 30.0 30.0 28.0 28.0 -3.0 EU27 35.2 34.1 33.5 31.9 30.7 29.3 28.3 27.0 25.5 25.3 24.5 23.6 23.5 23.3 -12.2 EA17 37.0 35.8 35.2 34.4 33.0 31.8 30.4 29.6 28.1 27.7 26.8 25.7 25.6 25.6 -11.4

Table 1 Top statutory tax rates on corporate income in the years 1995-2011 in %. Source: Eurostat (2011)

(18)

16 2011). The effective tax levels in the EU have not come down by the same level as the statutory corporate tax rates due to the tax base broadening reforms.

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 BE 34.5 34.5 34.5 34.4 34.5 29.5 29.5 29.5 25.7 25.4 24.9 24.9 24.6 BG 32.0 29.7 28.1 24.2 20.4 20.5 17.1 13.2 13.2 8.8 8.9 8.8 8.8 CZ 26.4 25.4 23.6 23.6 23.6 23.6 24.6 22.7 21.0 21.0 18.4 17.5 16.7 DK 30.0 28.3 28.3 26.8 26.8 26.8 26.8 25.1 25.1 22.5 22.5 22.5 22.5 DE 41.2 40.4 40.4 35.8 35.8 37.0 35.8 35.8 35.5 35.5 28.2 28.0 28.0 EE 22.4 22.4 20.4 20.4 20.4 20.4 20.4 18.8 18.1 17.3 16.5 16.5 16.5 IE 9.4 9.4 9.4 9.4 12.3 14.3 14.3 14.3 14.4 14.4 14.4 14.4 14.4 EL 30.4 30.4 30.4 30.4 30.4 30.4 30.4 27.8 25.2 21.7 21.8 30.5 41.5 ES 36.5 36.5 36.5 36.5 36.5 36.5 36.5 36.5 36.5 34.5 32.8 32.8 32.8 FR 39.8 38.4 36.6 35.8 34.9 35.0 35.0 34.8 34.4 34.6 34.6 34.7 31.0 IT 32.0 32.0 31.3 30.7 34.3 32.6 31.8 31.8 31.8 31.8 27.2 27.4 27.4 CY 27.5 27.5 27.5 26.5 26.9 14.8 14.8 10.6 10.6 10.6 10.6 10.6 10.6 LV 22.7 22.7 22.7 22.7 20.2 17.7 14.3 14.3 14.3 14.3 13.8 13.8 12.6 LT 23.0 23.0 19.1 19.1 12.7 12.7 12.7 12.7 16.0 15.2 12.7 16.8 12.7 LU 32.6 32.6 32.6 32.6 26.5 26.5 26.5 26.5 25.9 25.9 25.9 25.0 25.0 HU 19.0 19.3 19.7 19.7 19.7 19.7 17.8 16.6 16.3 19.5 19.5 19.5 19.1 MT 32.2 32.2 32.2 32.2 32.2 32.2 32.2 32.2 32.2 32.2 32.2 32.2 32.2 NL 32.3 32.3 32.3 32.3 31.9 31.9 31.9 29.1 27.4 23.7 23.7 23.6 23.6 AT 29.7 29.7 29.7 31.2 31.0 31.0 31.2 23.0 23.0 23.0 23.0 22.7 22.7 PL 32.4 30.6 27.1 25.3 25.3 24.2 17.1 17.1 17.1 17.4 17.4 17.5 17.5 PT 33.4 33.4 31.5 31.5 29.5 29.4 24.6 24.6 24.6 23.7 23.7 23.7 23.9 RO 34.0 34.4 22.7 22.7 22.9 22.7 22.4 14.7 14.7 14.8 14.8 14.8 14.8 SI 20.9 20.9 20.9 20.9 20.9 21.5 21.5 22.1 22.3 20.9 20.0 19.1 18.2 SK 36.7 36.7 25.8 25.8 22.3 21.9 16.5 16.8 16.8 16.8 16.8 16.8 16.8 FI 25.9 26.1 27.2 27.2 27.2 27.2 27.2 24.5 24.5 24.5 24.5 23.6 23.8 SE 23.8 23.8 23.8 23.1 23.1 23.1 23.1 24.6 24.6 24.6 24.6 23.2 23.2 UK 29.7 28.9 28.7 28.7 29.3 29.3 29.3 29.3 29.3 29.3 28.0 28.3 28.4 EU27 29.3 28.9 27.5 27.0 26.4 25.6 24.6 23.3 23.0 22.4 21.5 21.8 21.8

Table 2 Effective average tax rates for the non-financial sector in the years 1998-2010 in %. Source: Eurostat (2011)

3.2 Background and goals of the Werken aan Winst reform

On 21 July 2004, the Government released a “Growth Letter” in which outlines were given for fundamental changes in the Dutch economy. The need to reform the Dutch Corporate Income Tax Act 1969 (CITA 1969) stems from the wish to strengthen the Dutch economy. The Netherlands experienced a very slow economic recovery after the dot-com bubble burst in 2000. No significant reforms have taken place after the economic downturn. As a result, the Netherlands plunged to rank 15 in the Institute for Management Development competitiveness ranking3 of 2004 after a series of

(19)

17 top-5 rankings (Growth Letter, p. 5). The Netherlands have been known for its investment climate due to its central location in Europe, a robust physical infrastructure, well-educated labour force and favourable tax system. Corporate income taxes are seen as a relevant factor by improving the investment climate since taxes are of great importance when making investment decisions.

On 28 November 2006, the Dutch Senate approved the bill “Werken aan Winst”, containing considerable reforms of the CITA 1969. The changes should be seen within the broad context of the then Government’s ongoing efforts to further improve the Dutch investment climate. The changes aim to lower tax rates and broaden the tax base, which is in line with the observed trends in other industrialized countries. The revised tax system should become simpler and companies should be confronted with less compliance costs. Also, the changes aim to lower tax burdens of small and medium enterprises (SMEs). The share of total CIT revenues borne by SMEs has increased from 34,1% in the period 1995-1998 to 37,8% in the years 1999-2002. In the same period, the share of enterprises with international activities decreased from 66,1% to 61,5%. This may be explained by companies shifting profits to more favourable tax regimes or the relocation of (parts of) enterprises. The goal of the 2006 plans was to stop this movement by reducing the rates. A reduction of 1 percent point in the tax rate will result in an increase of the taxable base by 0,63%, according to a study of the Centraal Planbureau (Nota, 2005). Another goal of the WaW tax reform was to stimulate innovation and employment. More innovative companies should attract highly qualified personnel which will significantly improve the competiveness of the Dutch economy. Highly innovative companies show higher returns since they possess specific knowledge (Devereux et al., 2002). Companies which are very profitable will also benefit more from a tax rate cut cum base broadening reforms. They are generally less hit by base broadening reforms (e.g. limitations for loss carry over rules) but will benefit more from tax rate deductions since their large profits are taxed at a lower rate. Several measures aim to reduce undesirable tax planning and abuse and can be seen as base broadening rules. The need to comply with EU law was also one of the goals of the WaW tax reform, after a series of decisions of the Court of Justice.

As the aim of the reform measures is to improve the Dutch investment climate and the net corporate tax revenue impact is negative (see also par. 3.4), I expect ETRs to be lower after the WaW tax reform. Formally:

H6: ETRs are lower in the period after the WaW tax reform compared to ETRs in the period before the WaW tax reform.

infrastructure, labor market, tax regime and the ability to innovate are the most important drivers of a

(20)

18

3.3 Tax reform measures

This paragraph will describe the most important changes as effective per 1 January 2007, introduced by the bill “Werken aan Winst”.

3.3.1 Reduction of tax rates

The Dutch corporate tax rates were reduced from 29.1% to 25.5%. Lower rates were introduced for the first €25,000 of taxable income (20%) and for the income between € 25,000 and € 60,000 (23.5%). Also, the dividend withholding tax rate will be reduced from 25% to 15%. This reduction is less relevant to the international investor who typically can avail themselves of the benefits of a tax treaty or the Parent-Subsidiary Directive. This reduction may be a true benefit for investors who are not eligible for these benefits and is especially relevant for (intermediate) holding companies. As rates in the first tax brackets are lowered more than the general rate, I expect smaller companies to benefit more from the rate reduction.

H7: The association between ETRs and firm size is positively impacted upon by the Werken aan Winst tax reform.

3.3.2 Introduction of patent box

Self-developed intellectual property (IP) that has been patented after 1 January 2007 can be allocated to a patent box. An effective tax rate of 10 per cent applies to any income resulting from such IP. Previously deducted development expenses need to be recaptured to the extent the taxpayer derives income from the IP. The total eligible profits are limited to four times the amount of the development costs. The patent box has been renamed innovation box as per 1 January 2010. The effective rate is reduced to 5% and the amount of eligible profits is no longer limited.

3.3.3. Revised anti base erosion provisions

Specific interest deduction deferral rules restricting highly leveraged acquisitions have been replaced by new measures. Under the new measures, the acquisition of shares in a related company and the acquisition of shares in an unrelated company that will become related after the acquisition are considered ‘tainted’ transaction when financed with a loan from a related party. The anti-base erosion rules restrict the deduction of interest expenses on a related party loan if the proceeds have been used for such a ‘tainted’ transaction, unless the ‘business reason’ or ‘reasonable rate of taxation’ exception applies.

(21)

19 were considered when calculating the debt-to-equity ratio. As per 1 January 2013, the thin capitalization rules have been abolished in order to reduce overkill.

Given the less favourable treatment of debt financing after the WaW tax reform, I expect the relationship between firm capital mix and ETR to be impacted after the reform:

H8: The association between ETRs and firm leverage is positively impacted upon by the Werken aan Winst tax reform.

3.3.4. Limited depreciation of assets

A minimum deprecation term of 5 years has been introduced on assets such as inventory, cars and computers whereas this period will be ten years for goodwill. If the useful life of an asset is longer, this longer period must be taken into account. However, further depreciation may be possible if the value of an asset is lower than its book value.

Under the tax law prior to 1 January 2007, real property could be depreciated to the value of the historic acquisition cost minus its estimated value at the end of the useful life. Under the new measures, real property can in principle only be depreciated to the extent the book value of the real property does not drop below the threshold value. The threshold value depends on the use of the real property and is related to the WOZ value of the real property. The WOZ value of a real property is determined by the local municipality and is the tax base for municipal real estate tax. If the property is held as a portfolio investment, the threshold is 100% of the WOZ value whereas the threshold equals 50% of the WOZ value if the real property is used by the taxpayer or a related party. Given the less favourable impact of the WaW tax reform on depreciation rules of assets, I expect the relationship between asset mix to be impacted by the reform:

H9: The association between ETRs and firm capital intensity is positively impacted upon by the Werken aan Winst tax reform.

H10: The association between ETRs and firm inventory intensity is negatively impacted upon by the Werken aan Winst tax reform.

3.3.5 Restriction of loss carry back/forward

(22)

20 As loss carry over rules are limited and tax rates are lowered, I expect high profitable companies to benefit more from the WaW tax reform:

H11: The association between ETRs and firm profitability is negatively impacted upon by the Werken aan Winst tax reform.

3.3.6 Group interest box

Under this optional regime, an effective tax rate of 5 per cent applies to the balance of interest income and expense on intercompany loans. The box only applies to interest income and expenses, not to capital gains and losses or currency results. The European Commission did not provide immediate approval since the regime might constitute state aid and the introduction of the group interest box was postponed. Eventually, in 2009, the Commission approved the group interest box and confirmed that the regime did not constitute state aid. This approval did not trigger the Dutch Government to subsequently introduce the group interest box, probably due to the deterioration of the Dutch Budget.

3.4 Budgetary implications

(23)

21

€mln €mln

Tax reforms increasing corporate tax revenue

Restriction of loss carry back/forward 720

Restriction depreciation of real property 915 Minimum depreciation term of goodwill to ten years 110 Minimum depreciation term of other assets to 5 years 85

Other measures 285

2115 Tax reforms decreasing corporate tax revenue

Patent box 300

Reduction dividend withholding tax 40

Reduced tariffs for lower tax bracket 340

Reduced general tariff to 25,5% 1425

2105

Net corporate tax revenue impact -10

Table 3 Estimated effects of the Werken aan Winst tax reform on tax revenue in € millions. Source: Memorie van Toelichting (2006, p. 30)

Table 4 shows the estimated effects of the Werken aan Winst tax reform. It should be noted that these effects also include the proposed group interest box. This regime had an estimated negative impact on corporate tax revenues of € 475 mln but the introduction was postponed and cancelled due to belayed approval of the European Commission.

Effect

Size differences

Small and medium sized enterprises 2,2%

Multionational enterprises 1,9%

Per sector

Agriculture 2,2%

Industries 4,6%

Hospitality/trade/transport/communications 1,1%

Banks and insurance 1,9%

Financial service industry 2,7%

Service industry 2,3%

Real estate -4,3%

Other business 2,1%

Average 2,1%

(24)

22

3.5 Stimulus package in the years 2008 until 2010

(25)

23

4. Research design

4.1 Sample and data

The sample consists of a single panel of listed and non-listed Dutch firms collected from the Orbis Database over the period 2004-2010. The year 2007 was excluded because this was a transitional year. Prior research shows that firms normally respond to changes in tax law one year after the legislation becomes operative (Dhaliwal and Wang, 1992; Guenther, 1994). Arguably, the years 2008 and 2009 may also represent transition years. Some of the provisions were phased in over those years and an additional stimulus package of the Dutch government also influenced the tax regime during the years 2008 and 2009. Hence, the post-WaW data may be partially contaminated and likely has noise.

On the basis of previous research, firms in the sample have to meet several criteria. The final sample consists of 1033 firms after excluding firms that fall into the following categories:

1. Firms with missing data for any of the six sample years;

2. Firms classified as bank or insurance company since these firms face specific regulation that might affect their ETRs differently from other firms (Richardson and Lanis, 2007);

3. Firms with a negative tax expense since this will distort the calculation of their ETRs (Zimmerman, 1983; Omer et al. 1993);

4. Firms with negative income before taxes, since this will distort the calculation of their ETRs (Zimmerman, 1983; Omer et al. 1993);

5. Firms with carry over net operating losses, since their ETRs are difficult to interpret (e.g. Wang, 1991; Wilkie and Limberg, 1993);

6. Firms with ETRs exceeding the value of one, since these are not included in previous ETR research. ETRs exceeding one might cause model estimation problems (e.g. Stickney and Mcgee, 1982).

Step Sample firms

All active firms in the Orbis database without missing data 3177

Firms classified as bank or insurance company -69 3108

Firms with tax refunds or negative income -1769 1339

Firms with net operating losses -266 1073

Firms with ETRs exceeding one -40 1033

Final sample (number of firms) 1033

Final sample (firm years) 6198

(26)

24

4.2 Dependent variable

The dependent variable is represented by the average effective tax rate (ETR) of a company i: ETRi. As

mentioned in paragraph 3.1, different methods to calculate ETRs can be used. The ETR used in this research is based on information collected from financial statements and is defined as current tax expense divided by income before interest and taxes (EBIT). EBIT is often referred to as operational income and is also used in other empirical studies (e.g. Gupta and Newberry, 1997). Income tax expense is used as the numerator of the ETR equation: deferred tax assets and liabilities are not accounted for. This has two main reasons: the first is to be in line with most prior research in which deferred tax assets and liabilities are excluded (e.g. Fernandez and Martinez, 2011; and Richardson and Lanis, 2007). Second, only limited data are available in the Orbis database on deferred tax assets and liabilities. If deferred tax assets and liabilities would have been included as mandatory variable for all 6 sample years, the database would consist of less than 50 companies. Such a small dataset would not be suitable for sound statistical analysis.

Empirical studies on the determinants of ETRs often include another measure for the tax burden of a company to improve the robustness of the results. As taxable base does not account for specific tax benefits, a proxy for cash flow from operations is used. Earnings before interest taxes depreciation and amortization (EBITDA) is used as an alternative measure. This measure mimics operational cash flow and is used to control for systematic differences in accounting method choices that are related to firm size (Zimmerman, 1983).

In short, two different ETR measures are employed as dependent variable in order to improve the validity of the results. The first measure (ETR1) is defined as income tax expense divided by EBIT. The second measure (ETR2) is defined as income tax expense divided by EBITDA.

4.3 Independent variables

Firm size (SIZE) is the most widely used variable in prior ETR research. Firm size is measured as the natural logarithm of the firm’s total assets (at book value). An asset based firm size measure is used – instead of an income based size measure – in order to facilitate comparison with prior ETR studies. Wilkie and Limberg (1990) also argue that income based measures can yield misleading inferences about the ETR-frim size relation because larger firms tend to have higher profitability which, in turn, results in higher ETRs.

(27)

25 underestimate the leverage of a firm since this would not account for a series of non-equity financing products that has been introduced in recent years.

Two proxies for the asset mix are included to capture firms’ investment decisions. The first asset mix variable is capital intensity (CAPINT), defined as the ratio of fixed assets to total assets (both at book values). The amount for property, plant and equipment is often used in the denominator in previous research (e.g. Richardson and Lanis, 2007). In this study fixed assets is used because financial fixed assets and immaterial assets are also included in this variable. For both asset categories, the WaW tax reform should have negative consequences due to the adjustments to the participation exemption regime and the limitation of deductible depreciation expenses for goodwill. Participations qualified as ‘low taxed portfolio participations’ can no longer benefit from an exemption, but a tax credit applies. Goodwill should be depreciated over a minimum term of more than ten years. The second variable is inventory intensity (INVINT) and is defined as the ratio of inventory to total assets. Capital intensive firms should face lower ETRs due to the tax benefits associated with capital investments (e.g. preferential depreciation schemes). To the extent that INVINT is a substitute for CAPINT, inventory intensive firms should face higher ETRs.

Since profitable firms are expected to face higher tax burdens, a proxy for profitability is also included in the model. Return on assets (ROA) is defined as the income before taxes divided by total assets (book value).

To test the WaW tax reform impact on ETRs, a period dummy variable (WAW) is included in the model. The dummy is coded 1 if the observations are in the period after the tax-reform, 0 otherwise. The WAW coefficient provides a test of the shift of the mean ETR spanning the WaW tax reform. Also, all the independent variables are multiplied by the WaW dummy variable. This permits testing for slope shifts in each of the firm-specific variables after tax reform and it is therefore possible to determine whether these associations have changed after the WaW tax reform.

(28)

out-26 innovate large companies because those large companies have all the money. Larger companies can be woefully bureaucratic, slow, inefficient and risk averse. That leaves plenty of opportunity for smaller companies to out innovate the larger ones despite the appearance of a disadvantage in money and scale. In most cases, small and innovative companies do not file a patent for their new ideas due to the expensive application process. Thus, the number of patents can’t be seen as a very suitable proxy for innovation and will not be added as a variable to the regression model.

4.4 Control variables

ETRs can fluctuate across sectors. As described in paragraph 3.4 the consequences of the WaW tax reform are expected to range from a 4,3% increase in tax burdens for the real estate sector to a 4,6% decrease in tax burdens for industrial firms. Six dummy variables (INSEC) are incorporated in the study: agriculture, industries, hospitality/trade/transport/communications, financial service industry, real estate and other business. These are derived from the eight industry sectors as described in paragraph 3.4. ‘Other business’ and ‘service industry’ are merged into the category ‘other business’ and the category ‘banks and insurance’ is not taken into account since all bank and insurance companies are removed from the sample. The classification server NACE is used in order to assign the companies to one of the categories. NACE is derived from the French title “Nomenclature générale des Activités économiques dans les Communautés Européennes” (Statistical classification of economic activities in the European Communities). The current version NACE Rev.2 is developed by Eurostat in 2006. The Orbis database provides a division number for each of the firms in the sample. These classifications have been translated into the six sectors as described in paragraph 3.4. An overview of the structure of NACE Rev. 2 compared to the six sectors used in this study can be found in Table 6.

Number of firms in sample NACE Rev.2 Divisions

Agriculture 11 01-09

Industries 240 10-44

Hospitality/trade/transport/communications 409 44-63

Financial service industry 273 64-66

Real estate 16 68

Other business and service 84 69-99

Table 6 Firms in the sample per industry

4.5 Regression models

(29)

27 ETRit = β0 + β1SIZEit + β2LEVit + β3CAPINTit + β4INVINTit + β5ROAit + β6-11INSECit + β12WAWit + β13SIZEit *

WAW + β14LEVit * WAW + β15CAPINTit * WAW + β16INVINTit * WAW + β17ROAit * WAW + β18-23INSECit *

WAW + ԑit (3)

where ETRit is the dependent variable and is a proxy for the corporate effective tax rate for firm i in

year t. The independent variables (subscripts omitted) include proxies for firm size (SIZE), capital structure (LEV), asset mix (CAPINT and INVINT), firm profitability (ROA), industry sector (INSEC), the Werken aan Winst tax reform (WAW) and interaction terms for the variables after the WaW reform (SIZE * WAW, CAPINT * WAW, INVINT * WAW, ROA * WAW, INSEC *WAW).

In order to test the relationship between ETR and firm size, capital structure, asset mix, firm profitability and the WaW tax reform a regression will be run without the interaction terms. The definitions of these variables and measurement issues affecting their computation are discussed above. All data on these variables are based on financial statement data drawn up from the Orbis database.

In order to test the hypothesis 1 to 6 the interaction variables will not be included in the first model. This will give the following regression model:

ETRit = β0 + β1SIZEit + β2LEVit + β3CAPINTit + β4INVINTit + β5ROAit + β6-11INSECit + β12WAWit +ԑit (4)

(30)

28

5. Results

5.1 Descriptive statistics

Table 7 reports descriptive statistics for dependent variables ETR1 and ETR2 and for independent variables over the period 2004-2010. Also, descriptive statistics are reported for the periods before and after the WaW tax reform. The dependent variable ETR1 has a mean of 26,40% and a median of 25,93%. Clear differences exist between the periods before and after the WaW tax reform. The mean shows a decrease of 5,37% and the median a decrease of 5,62% for ETR1. Similar observations can be made for dependent variable ETR2, which has a mean of 19,72% and a median of 20,25% for the full test period. Again, clear differences exist between the periods before and after the tax reform with a decrease of 4,13% for the mean and 4,45% for the median. A paired sample t-test has been executed and the results show that the changes of the mean are significant (results in right column). Given that both ETR measures have the same numerator (financial accounting tax expense) and that EBITDA is usually greater than EBIT, the mean and median for ETR1 are greater than for ETR2, as can be expected. Figure 1 (see next page) shows a graph with ETR1 and ETR2 for all sample years compared with the ETRs for the Netherlands as calculated by Eurostat (2011). All sectors experienced a decrease in ETR in the range of 4 – 5 per cent points (sector ETRs not reported).

Paired sample t-tests have been executed in order to compare the means of the independent variables in the pre and post WaW tax reform period. The independent variables CAPINT and INVINT do not show a significant change over time. Firms in the sample have become smaller and more profitable over time. Also, their finance structure changed as leverage has become significantly lower after the WaW tax reform.

2004-2010 2004-2006 2008-2010

(6198 firm years) (3099 firm years) (3099 firm years)

Mean

Std.

dev. Median Mean

Std.

dev. Median Mean

(31)

29 Figure 1 Corporate effective tax rates before and after WaW tax reform. ETR1 is income tax expense divided by EBIT, ETR2 is income tax expenses divided by EBITDA and ETR3 are corporate effective tax rates for the Netherlands as published by Eurostat in their 2011 Taxation Trends publication.

Table 8 gives the Pearson correlation matrix and shows that all the correlations are statistically significant, except for the one between inventory intensity and ETR2. Logically, ETR1 and ETR2 are heavily correlated since they use the same numerator and the proxies in the denominator of the ratio differ only to a very small extent. Also, capital intensity is positively correlated with leverage and negatively correlated with inventory intensity. As almost all independent variables are significantly correlated, attention has to be paid to multicollinearity in the regression model. When two or more independent variables in a multiple regression model are highly correlated the coefficient estimates may change erratically in response to small changes in the model or the date. The multiple regression model may not give valid results about individual predictors in the model. Variance inflation factors (VIFs) are used to estimate the degree of multicollinearity in the regression models.

ETR1 ETR2 SIZE LEV CAPINT INVINT ROA

ETR1 1 ETR2 0,710** 1 SIZE -0,072** -0,098** 1 LEV -0,179** -0,316** 0,016** 1 CAPINT -0,013** -0,368** 0,315** 0,522** 1 INVINT -0,120** -0,003 -0,147** -0,145** -0,460** 1 ROA 0,138** 0,425** -0,163** -0,229** -0,191** -0,064** 1 ** Correlation is significant at the 0,01 level (2-tailed)

* Correlation is significant at the 0,05 level (2-tailed)

(32)

30

5.2 Regression results

Table 9 provides OLS regression results of the relation between ETRs and several firm characteristics over the period 2004-2010, using standardized coefficients. Standardized coefficients are the estimates resulting from an analysis carried out on independent variables that have been standardized so that their variances are 1. Therefore, standardized coefficients refer to how many standard deviations a dependent variable will change, per standard deviation increase in the predictor variable. First, regressions are run without any interaction variables and thus including the independent variables SIZE, LEV, CAPINT, INVINT, ROA and WAW and the industry dummy variables. Both models are significant at the 0,01 level. The ETR 1 regression model reports an adjusted R² of 15% whereas the ETR2 regression model reports an adjusted R² of 36%. The adjusted R² is the amount of variance in the outcome that the model explains in the population. The Durbin-Watson statistic is a test statistic used to detect the presence of autocorrelation in the residuals from a regression analysis. Autocorrelation is important when analyzing time series data since standard errors in the OLS regression model may be underestimated. The Durbin-Watson statistic varies from 0 to 2. No autocorrelation exists when the value is close to 2; close to 0 autocorrelation will exist (Wonnacott and Wonnacott, 1970, p544-545). The Durbin-Watson statistic is close to 2 for both the ETR 1 model (1,936) and the ETR 2 model (1,878). This is caused by the relatively large sample size (1033 firms) compared to the length of the period observed (6 years). Variance inflation factors (VIFs) are used to estimate the degree of multicollinearity in the regression models. As a rule of thumb, VIF greater than 5 may indicate there is a multicollinearity problem. VIFs do not exceed the value of 2 in both models, thereby minimizing the problem of multicollinearity (results not reported).

expected ETR1 ETR2

SIZE ? 0,006 0,099** LEV ? -0,165** -0,134** CAPINT ? -0,121** -0,336** INVINT ? -0,188** -0,132** ROA + 0,058** 0,337** WAW - -0,280** -0,261**

Industry dummies yes yes

adjusted R² 0,15 0,36

Durbin-Watson 1,878 1,923

** Significant at the 0,01 level (2-tailed) * Significant at the 0,05 level (2-tailed)

(33)

31 Firm size has an insignificant positive relationship with ETR1 and a significant positive relationship with ETR2. The coefficients are relatively small compared to the other independent variables. The relationship between size and ETR is the only one without comparable results for the ETR1 and ETR2 model. Therefore, hypothesis 1 suggesting an association between firm size and ETR cannot be accepted on the basis of these linear regression models. These findings may be in line with Gupta and Newberry’s (1997) assertion that the size-ETR relationship may be sample specific and Fernandez and Martinez’ (2011) results suggesting that there might be a nonlinear relationship between size and ETR. In the next paragraph an additional test will be done to test for a possible non-linear relationship that might explain these results.

Leverage is included as a proxy for the capital structure of a firm. The results in both models show a negative relationship between leverage and ETR, as is in line with most prior research (e.g. Richardson and Lanis, 2007; Gupta and Newberry 1997). Therefore, hypothesis 2 suggesting an association between firm leverage and ETR can be accepted on the basis of these results. The two proxies for the asset mix (CAPINT and INVINT) show significant results as well. Both variables are negatively correlated as described in the Pearson correlation matrix in Table 8 and opposing signs might be expected as inventory can be seen as a substitute for capital. However, regression results show a negative relationship for both variables in both models. Based on these results, both hypothesis 3 and 4 can be accepted. The coefficient for CAPINT in the ETR2 model is much higher than the coefficient in the ETR1 model. This may be explained by the fact that the difference between EBIT (used in ETR 1) and EBITDA (used in ETR2) is the amount of depreciation and amortization. This difference will be large for firms with a lot of capital since they incur relatively more depreciation and amortization expenses. Therefore, EBITDA will be higher than EBIT and ETRs calculated on the basis of EBITDA will be lower (since both ETR1 and ETR2 have the same numerator).

As for profitability, ROA is included to measure the relationship between firm profitability and ETR. The results show that it has a significant positive association with ETR1 and ETR2, as expected by hypothesis 5. On the basis of these results, hypothesis 5 can be accepted.

Corporate effective tax rates and tax reform: evidence from the Netherlands