Ownership structure and corporate tax avoidance of publicly listed
European firms
Name: Julia Toroi Student number: s3425959 Study programme: MSc IFM Supervisor: Egle Karmaziene
Abstract: This thesis investigates the relationship between ownership structure and corporate tax avoidance of publicly listed European firms, by establishing the importance of controlling owner identity, ownership share, and the role of statutory corporate income tax rates. The findings are that firms exhibit differences in tax aggressiveness, depending on the identity of their controlling owner, with firms owned by financial institutions being the most tax aggressive ones, and firms owned by families being the least tax aggressive ones.
1 Table of contents
1. Introduction ... 2
2. Literature review and hypothesis development ... 4
Agency theory ... 5
Owner identities ... 6
2.2.1. Widely held corporation ... 7
2.2.2. Widely held financial institution ... 8
2.2.3. Families ... 8
2.2.4. Others ... 9
Ownership concentration ... 9
Statutory corporate income tax rate ... 11
3. Methodology ... 12
Measuring tax avoidance ... 12
Ownership structure ... 12
Research design ... 13
3.3.1. Ownership structure ... 13
3.3.2. Effects of corporate income tax rate ... 14
3.3.3. Controls ... 15
4. Data... 16
5. Results ... 22
Results for ownership structure estimates ... 22
Effect of corporate income tax rate ... 26
6. Robustness analysis ... 29
7. Conclusion ... 32
References ... 35
2 1. Introduction
Tax aggressive behaviour of firms has received substantial negative attention in media during the last few years. The attention has mainly been drawn to giant multinationals, such as Google, Amazon, Starbucks and Apple, because of their preposterously low tax expenses. The low tax rates have often resulted in public outcry from consumers, who find this behaviour unethical and unfair. While this response often is irrational and based on perceived unfairness, tax revenues evidently play an important role in society as they are needed to finance public goods (Lanis and Richardson, 2012). Executives of several of these firms, however, state that they do not see the problem with their tax planning as it is their fiduciary duty to maximise shareholder wealth. Some executives, for example the CEO of Google, have even gone as far as saying that they are proud of their clever tax strategies (Telegraph, 2012). The negative attention has led to the EU tax reform that is currently in progress, which is aimed at controlling tax avoidance (European Commission). The EU is the only supranational institution in the world with the power to enforce tax legislation on its member states. Even though the European Commission states that collecting taxes and combating tax fraud is the role of the EU Member states, it also states that it “provides a framework and offers instruments to effectively handle cross-border tax issues” as an increasingly globalised world poses challenges for national tax authorities (European Commission).
3 (Desai and Dharmapala, 2009). Thus it remains still unanswered, whether ownership structure affects firm value, at least to some degree, through tax avoidance.
Nevertheless, most studies have been done with US data only and thus, the typical ownership structure of listed firms in the US, due to the strong equity market, may make these specific for the US. Firms in continental Europe generally receive their financing from banks and institutional investors as the equity markets in these countries are considered as weak (Doupnik and Perera, 2015). It is not only the providers of financing to firms in general that differs between the US and continental Europe; continental European firms also typically have higher concentration of ownership (Thomsen and Pedersen, 2000). The high concentration of ownership causes continental European firms to have strong owners and weak managers, in contrast to the US, which is said to have a problem of strong managers and weak owners (La Porta et al., 1999; Becht, 1997). This may give owners in continental Europe a stronger say, also when it comes to their tax avoidance preferences. These factors make continental European firms different from US firms, with respect to their ownership structure, which make firms in EU 15 countries, with the UK and Ireland excluded, an interesting and insightful sample to study. Additionally, the identities of the typical investors in continental Europe are families, corporate owners, and the government, while in for example the US and the UK, the typical investors are financial institutions (Thomsen and Pedersen, 2000). This may have a substantial effect on the objectives of the shareholder as families, the government and corporate owners may have objectives different from maximising shareholder value (Thomsen and Pedersen, 2000). Because of this, studies performed on US or UK data generally don’t address the effects on owner identity, but instead focus solely on ownership concentration. This focus is, simply put, not appropriate in continental Europe where owner identities differ and a large portion of companies have a controlling shareholder (Pedersen and Thomsen, 2003).
4 will hopefully come to a stronger conclusion of under which circumstances the ownership structure of a firm affects its tax aggressiveness. Focusing on EU countries will provide a homogenous sample, as it will control for large differences in the regulatory environment that the firms in the sample operate in, which reduces external factors affecting the tax avoidance of firms (Remeur, 2015).
I find that firms owned by financial institutions behave more tax aggressively compared to other firms. Firms owned by families, on the other hand, appear to be less tax aggressive than other firms. Firms owned by widely held corporations do not exhibit a different degree of tax aggressiveness than the widely held corporations themselves. The findings hold at higher than average corporate income tax rates. Additionally, I find that the ownership concentration of a firm with a controlling owner does not further affect the tax avoidance behaviour of the firm. Thus, it can be assumed that being a controlling owner possesses the owner with enough power to line the incentives of the manager with its own interests. Further are statutory corporate income tax rates found to be the cause of the tax aggressiveness of firms owned by financial institutions.
This thesis contributes to the existing research on the determinants of tax avoidance by examining a new aspect of it, which is firm ownership structure. Previous research has not established whether the identity of a controlling owner, or the share it holds, affects tax avoidance of publicly listed European firms. The findings of this study may indicate whether there are gains to be made by shareholder by increasing their ownership in the firm into a controlling stake, as Thomsen and Pedersen (2000) point out that the corporate strategy needs to fit the objectives of the dominant owner. Also minority investors might, on the basis of the findings of this thesis, be able to identify firms with a level of tax aggressiveness that fits the preferences of the investor. Additionally, as corporate tax avoidance may have a negative impact on society, this research topic is relevant from the perspective of national governments and tax authorities, as they may be interested in knowing which firms are prone to show tax aggressiveness. Therefore, understanding the causes of tax avoidance supports policy makers in understanding the effect of any changes they make. In order to do so, the drivers of tax aggressiveness need to be identified.
2. Literature review and hypothesis development
5 However, when it comes to tax avoidance, the ultimate border between avoidance (legal) and evasion (illegal) is not always evident (Slemrod, 2004). As such it becomes a question of judgement and thus tax avoidance can be assumed to be associated with some degree of risk, depending on the extent of it. The cost of tax evasion, if caught, to the firm and its managers can be significant and it includes for example fees paid to tax experts, time spent on the resolution of tax audits, reputational loss, and penalties or fees paid to tax authorities (Badertscher et al., 2013). However, Dyreng, Hanlon and Maydew (2014) state that, if a firm has arranged its strategy to avoid taxes in a way that is compliant with rules and regulations, the risk associated with it could be quite small. Although, they also find that firms that avoid taxes are associated with higher uncertainties in taxes. This adds to the assumption that avoiding taxes increases the firm specific risk as uncertainties in taxes mean higher earnings volatility. Thus it is assumed in the supporting theory that tax avoidance implies increased risk for the firm. The assumption is necessary for the agency theory to hold; firm managers are generally assumed to be more risk averse than the shareholders and thus, they prefer less tax avoidance. Consequently, a conflict of interest may arise between owners and managers. This conflict of interest is, however, assumed to be only due to differences in risk preferences, compared to differences in personal virtues and beliefs about the ethics of tax payments, as these are assumed to be left aside (Slemrod, 2004).
Agency theory
6 havens used to avoid taxes simultaneously provide opportunities for rent extraction by managers (Desai and Dharmapala, 2009). When income is less visible form tax authorities, it will also be less visible for shareholders. This means that the managers have more opportunities for expropriation of rents (Desai and Dharmapala, 2006).
Owner identities
Defining and identifying the ownership structure leads to greater understanding of the firm’s actions, as I already discussed the fact that ownership structure has an influence on the corporate governance and the performance of a firm. Thomsen and Pedersen (2000) find that the owner identity is an equally important determinant of the performance, as is ownership concentration, which will be discussed in section 2.3. below. On a general level, firms can be divided into those being widely held and those having a controlling owner. Owning more than 50 percent of the voting rights constitutes the definition of the controlling owner, due to the typically high ownership concentrations in European firms (Thomsen and Pedersen, 2000). A firm can have a majority owner that is not a controlling owner, if the majority owner holds less than, or exactly, 50 percent of the voting rights. The controlling owner is the ultimate owner, or the majority shareholder of the majority shareholder, etc., going far enough through the ownership chains until a shareholder controlling the majority shareholder right is found (LaPorta et al., 1999). Thus, firms that have a majority without a controlling share, will be classified as widely held. The identities of the controlling owner can be divided into five categories, namely families, the state, a widely held financial institution (e.g. a bank, a pension fund or an insurance company), a widely held corporation, and miscellaneous (La Porta et al., 1999). For the purpose of this thesis, the categories will be narrowed down so that miscellaneous also includes state owned firms, as the scope otherwise would become too wide.
7 the most common one, however all of them can be used by a shareholder to gain control rights in excess of its cash flow rights (La Porta et al., 1999).
The following section will present the different owner identities, and discuss the specific agency problems related to them, as well as possible preferences and goals of these, as they may depend on the owner identity (Pedersen and Thomsen, 2003). This applies only for companies with a controlling owner, thus owning over 50 percent of the voting rights in the company - companies without it will be classified as widely held.
2.2.1. Widely held corporation
A firm controlled by another corporation can be thought of as either being widely held, since the management of the controlling firm is itself not accountable to an ultimate owner, or as being controlled by that management (La Porta et al., 1999). This may cause the management of the parent firm has lower incentives to monitor and control the subsidiary, as they have less to gain personally, thus constituting an agency cost from the perspective of the shareholders of the parent firm. Additionally, being owned by a widely held corporation makes the control dispersed into more decision makers, which may make decisions regarding strategy harder to reach due to information asymmetry. It may thus also affect decisions regarding the tax strategies of the firm, which makes a clear effect on the tax aggressiveness difficult to conclude.
Corporate group structures or cross-ownership are generally a strategic choice in the sense that they may give the firms involved access to necessary resources (Kester, 1992). However, corporate groups may have size related disadvantages due to greater government scrutiny, and this “political cost” means that these firms pay more taxes (Zimmerman, 1983).
The aforementioned leads to a hypothesis that is ambiguous. It may lead to an insignificant result, as the managers of the owner has lower incentives for value maximising strategies and the more dispersed ownership makes decision making more difficult, or it could lead to these firms paying more taxes, due to the political cost. Of these theories I expect the one relying on agency theory to prevail, and that the result thus will be insignificant.
Hypothesis 1: Having a widely held corporation as the controlling owner has no effect on the tax
8 2.2.2. Widely held financial institution
Financial institutions, such as banks, pension funds, insurance companies, and other financial companies are assumed to have the main objective of maximising shareholder value. They possess large amounts of capital under their control, and this capital needs to be invested. Large investors face relatively lower transaction costs, which mean they can change their portfolios relatively easy, thus often having a shorter investment horizon (Gunthorpe and Levy, 1994). As their portfolios are held at arm’s length and they are mainly evaluated on their financial result, financial institutions will arguably aim for value maximising strategies (Pedersen and Thomsen, 2003). Typical characteristics of institutional investors are the facts that they are relatively specialized and they generally have a positive impact on shareholder value (Thomsen and Pedersen, 2000). Due to the relative low transaction cost and high specialization, it can be assumed that financial institutions generally have well diversified portfolios, meaning that firm specific risks of the shares they don’t affect the value of their portfolio. Therefore it is assumed that financial institutions as investors are risk neutral or risk taking, and that they prefer the firm to behave tax aggressively due to the fact that it creates value for the shareholders.
On the other hand, just like in section 2.3.2. above, the same information asymmetry issues may arise when widely held financial institutions are the controlling owners. However, due to the stronger objectives of shareholder value creation of widely held financial institutions, managers in the parent firm are expected to have stronger incentives in their contracts in order to minimize the aforementioned issues. Thus, it is assumed that they will prefer tax aggressiveness, as this should maximise shareholder value.
Hypothesis 2: Firms owned by widely held financial institutions are more tax aggressive than other firms.
2.2.3. Families
9 for longer time periods (Zellweger, 2007). This implies that they are interested in the long run profitability of the firm, instead of short term profits. It also makes family owned firms more concerned of their reputation (Chen et al., 2010). In addition, a large share of their personal wealth often is tied up in the firm, making family owners relatively risk averse (Thomsen and Pedersen, 2000).
The fact that the founding family is present and has high control within the company implies a greater agency conflict between large and minority shareholders. This is because the minority shareholders often have more diversified portfolios, and thus they prefer more risk, while the family owners often have a less diversified portfolio and a larger stake in the specific company. In addition to this, the risk of expropriation by the family, at the disadvantage of minority shareholders, worsens the conflict. At the same time, there is a smaller agency conflict between owners and managers as compared to non-family firms, as the owners are either represented in the management, or at least they work closely together. (Chen et al. 2010) In general, family owned firms are expected to be relatively risk averse (Pedersen and Thomsen, 2003) and thus, having a family as the ultimate owner is expected to lower the tax avoidance of a firm, which is also consistent with the result from Chen et al. 2010.
Hypothesis 4: Family owned firms are less tax aggressive than other firms.
2.2.4. Others
The “others” category holds all identities that do not fit into the categories above. These are government owned firms, foundations, research institutes, and firms with managers or employees as the controlling owner. This category is not relevant for the research of this thesis.
Ownership concentration
10 structures (Holmström, 1999). La Porta et al. (1999) elaborates on the fact that controlling shareholders have stronger incentives to monitor managers in order to maximise their own wealth. Firms with a controlling owner, so a shareholder that holds over 50 percent of the voting rights in the firm, can be assumed to have strong governance, as the majority owner has both incentive and power to monitor and control the management. Therefore, it can be assumed that tax avoidance is value creating when a firm has a controlling owner. Thus, higher ownership concentrations may imply a higher tax avoidance, however, it is not clear whether there is a linear relationship between the ownership concentration and tax avoidance, or if it is enough that one shareholder is in control for tax avoidance to be maximised. In other words, it is not clear whether these incentives increase with the ownership concentration, thus meaning falling agency costs with increasing concentration.
This implies that a firm’s ownership structure is the primary determinant of how the relationship between controlling owners and managers are affected by agency problems. Nevertheless, it can be assumed that for firms with controlling owners, the decision to avoid taxes is caused by the preferences of the owner, and not the manager, as the manager on its own is not likely to be tax aggressive. The question is whether these agency problems can be assumed to be controlled for with strong corporate governance, independently of the extent of them. If the answer is no, there would be a bias against finding a significant relationship between firms with lower ownership concentration and larger incentives problem, and for finding a relationship between firms with higher ownership concentration and smaller incentives problem.
11 concentration and owner specific preferences, the ownership concentration may have a positive, a negative or an insignificant effect on tax avoidance. However, I expect the agency theory to have a dominant effect and that high concentration of ownership increases the incentive of the majority shareholder to monitor the managers.
Hypothesis 4: The effective tax rate of firms owned by a widely held corporation is not
affected by the ownership concentration.
Hypothesis 5: The effective tax rate of firms owned by a widely held corporation is negatively
affected by the ownership concentration.
Hypothesis 6: The effective tax rate of firms owned by a family is positively affected by the
ownership concentration.
Statutory corporate income tax rate
If some specific owner identities turn out to prefer to avoid less taxes, as expected, the question that remains is whether this actually is caused by the owner identity, or if the behaviour of the owner is caused by the level of the corporate income tax rate. The higher the STR, the more there is to gain by avoiding taxes. In high tax countries tax aggressive firms might have significantly lower ETR due to the fact that there is much to gain by avoiding taxes, and therefore these firms may end up having a substantial benefit over the firms that avoid less taxes. Thus, it is possible that with high STR, even firms with owners causing them to avoid little taxes will have to increase their tax avoidance, in order to stay competitive.
On the other hand, it could also be assumed that owners preferring to avoid more taxes maximise their tax avoidance at all levels of corporate income taxes and thus, cannot avoid more taxes without it becoming tax evasion. However, it can also be argued that when corporate income taxes are low, firms have low incentives to engage in clever tax strategies as these may be costly, since they anyway operate under income tax rates that are internationally competitive. Thus, this adds to the argument that firms become more tax aggressive when the STR is high, independent of the identity of the owner.
Hypothesis 7: A higher statutory corporate income tax rate will make firms controlled by a
widely held corporation more tax aggressive.
Hypothesis 8: A higher statutory corporate income tax rate will make firms controlled by a
12
Hypothesis 9: A higher statutory corporate income tax rate will make firms controlled by one
individual or a family more tax aggressive.
3. Methodology
Measuring tax avoidance
The dependent variable for the empirical test is tax avoidance. Tax avoidance is measured based on a firm’s effective tax rate (ETR), as this is the most commonly used measure (Lanis and Richardson, 2012; Slemrod, 2004; Dyreng et al. 2010, Chen et al., 2010). Thus, a low ETR means high tax avoidance, while a high ETR means low tax avoidance. ETR1, is found by research to proxy tax aggressiveness (Slemrod, 2004; Dyreng et al., 2008). ETR1 is measured as total tax expense divided by pre-tax income, for firm i at time t:
ETR1i,t = Total Tax Expensei,t / Pretax Incomei,t.
The total tax expense is the tax to recognize for the period, and thus it includes current income tax and deferred taxes for the period. This measures the aggressive tax planning of a firm through permanent book-tax differences, including for example the use of tax havens (Chen et al., 2010). Other factors affecting the taxes that firms actually pay in European countries are generally deduction of interest expense, depreciation tax shield (via expensing or rapid depreciation), tax benefit for R&D, varying use of LIFO for inventory accounting, tax-advantaged transfer pricing (Lee and Swenson, 2012), and having affiliates in tax havens (Jaafar and Thornton, 2015). Countries differ in their regulation of these aspects, and the tax rules are essentially as important as the STR itself in determining the ETR (Lee and Swenson, 2012).
Ownership structure
13 between these two is based on data availability, however, if both are available, the total ownership percentage is used.
The owner identity gives rise to concerns regarding endogeneity as there can be simultaneity between ETR and the identity of the controlling owner. For example, a shareholder could increase it share in the company and become the ultimate owner, if it prefers firms with a certain level of tax avoidance. This means that instead of the owner identity causing the level of tax avoidance of the firms, owners of specific identities sort to firms that already express this behaviour. The causality of the relationship is difficult to establish. It could be done if there are enough changes of owner identity in the sample, as the effect the change has on tax aggressiveness then could be estimated. This is, nevertheless, outside of the scope of this thesis.
Research design
3.3.1. Ownership structure
The empirical part of this research is based on multivariate analysis using estimated ordinary least squares (OLS) regressions. I use robust standard errors. For this research, a timeframe of ten years should be sufficient (2006-2016). To test whether the owner identity has an effect on tax avoidance, I estimate the following regression:
14 To estimate if, additionally, the ownership share of the controlling owner has an effect on tax avoidance, I estimate the following regression, separately for each of the owner identities:
ETR1i,t = 𝛼𝛼0 + 𝛽𝛽1 CONCi,t + Controls + 𝜀𝜀i,t . (2) CONCi,t is the ownership concentration of firm i at time t, as explained in section 3.2.
3.3.2. Effects of corporate income tax rate
For the corporate tax rate, the statutory corporate income tax rate will be used for each country in the sample. To test for differences in the effects of the controlling owner identities caused by the statutory corporate income tax rate, I estimate the following regression separately for subsamples with only firms owned by widely held financial institutions, only widely held corporations, and only family owned firms:
ETR1i,t = 𝛼𝛼0 + 𝛽𝛽1 STRj,t, + Controls + 𝜀𝜀i,t. (3)
STRj,t is the statutory corporate income tax rate in country j, where firm i is incorporated, at time t. An endogeneity issue rises from the equation, as it is possible that firms sort into countries with low corporate taxes, to decrease their ETR. A firm’s decision to locate in a specific country may be affected by the statutory corporate tax rate of the country, as well as the rules and regulations regarding the tax rate. This has caused countries to compete for firms by decreasing corporate income tax rates (Bénassy-Quéré, 2014). As such, this could mean that tax aggressive firms sort into countries with low STRs. The causality of this relationship can, however, not be tested within the scope of this thesis.
To test for the interaction effects between the corporate income tax rate and owner identities, I estimate the following regression for the full sample:
ETR1i,t = 𝛼𝛼0 + 𝛽𝛽1 FINANCIALi,t +𝛽𝛽2 CORPORATIONi,t + 𝛽𝛽3 FAMILYi,t + 𝛽𝛽4 (FINANCIALi,t × highSTRj,t) +𝛽𝛽5 (CORPORATIONi,t × highSTRj,t) + 𝛽𝛽6 (FAMILYi,t × highSTRj,t) + 𝛽𝛽7
15 where highSTRj,t is a dummy interaction variable indicating firm-years where the STR is above the median. Studying the main relationship together with the interaction effect from STR may show how the main relationship varies across different levels of STRs. The moderator variable also has a direct effect on the dependent variable, as one of the main drivers of a firm’s ETR is the STR. However, this effect is distinctively different from its moderating effect. The moderation effect of highSTR works through the incentives the high levels of corporate tax rates gives rise to. A higher STR implies that there is more to gain by avoiding the taxes and thus, this effect should not be confused with the direct effect. The use of highSTR comes with endogeneity issues. The problem is that it can’t be known if it acts as a moderating variable on the relationship between owner identity and tax avoidance, of if, in fact, the owner identity is the moderator of the relationship between highSTR and tax avoidance. The second option is, nevertheless, less likely as it does not hold much theoretical ground.
3.3.3. Controls
16 Badertscher et al. (2013) finds that a high institutional ownership has a positive and statistically significant effect as an interaction term on the relationship between tax avoidance and firm value. They use institutional ownership as a measure of the governance of the firm, and define a high institutional ownership as a fraction that exceeds 0.6. The motivation for this is that institutional investors have greater incentive and capacity to monitor the managers, which thus controls for some of the agency problems. In my analysis, I will extend this assumption from institutional investors to all investors, so that any controlling owner of any identity has the incentive and capacity to monitor managers. Instead of the threshold of 0.6, I will use 0.5001, as this is my definition for being a controlling owner and thus, this assumption applies only for firms with controlling owners. Thus, an additional control variable for corporate governance is not needed. A summary of all variables is found in the appendix, for ease of reading.
CountryDummies controls for country-wise differences in corporate taxes, differences in tax rules, as well as cultural differences. YearDummies controls for year fixed effects allow for variation within the firms over the years. Chen et al., 2010)
4. Data
The data is withdrawn from the ORBIS database by Bureau van Dijk, for the years 2005 to 2016. The sample begins with all publicly listed firms, both currently active and inactive, in the following EU countries: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Italy, Luxembourg, Netherlands, Portugal, Spain and Sweden. This sample is intended to minimize measuring errors since differences in accounting rules are expected to be minor, as argued by Lee and Swenson (2012). The countries in the sample follow EU legislation and thus, firms make their financial reports in accordance with IFRS since 2005. Thus it can be assumed that differences in factors affecting ETRs, as mentioned in the literature review, are small.
The database contains extensive ownership data, including the ultimate shareholder of a firm, its identity, and its direct and total ownership percentage. In addition to this, ORBIS provides the necessary financial data of the firms.
17 pre-tax income, income taxation, and controlling owner identity data. Additionally, I require firms to have positive values of pre-tax income. A negative income before tax, i.e. a loss, would give a negative ETR, and this would affect the validity of the study. In addition to this, countries differ in their treatment of loss carryforwards, a factor that also reduces the ETR and thus excluding firms with negative income eliminates the influence of differences in treatment of loss carryforwards (Lee and Swenson, 2012).
To mitigate effects of outliers, I trim both ETR measures on [0, 1], winsorize RD, SIZE and ROA at the 1st and 99th percentiles (Dyreng et al., 2016), and REV, LEV and PPE at 1st and 95th percentiles, because of their initial skewness.
Figure 1 presents two graphs. The left side shows the mean share of each owner identity for the countries in the sample. The full bar shows the mean percentage of firms with a controlling owner. In my sample, the countries with the most firms with controlling owners are Italy, France and Germany – 57.3, 46.1 and 42.1 percent of publicly listed firms have controlling owners in these countries, respectively. The countries with the smallest percentage of controlling owners are Finland, Sweden and the Netherlands – only 7.8, 13.4 and 17.0 percent of the firms in the sample have controlling owners, respectively. In the full sample, 36.5 percent of firms have controlling owners. 5.7 percent are owned by a financial institution, 12.5 by a widely held corporation and 17.3 by families. The right side graph of figure 1 shows the ownership concentrations for each country in the sample. These are rather even for the whole sample - the mean for the full sample is 69.5 percent and the median is 66.1 percent. It can be noted, however, that the ownership concentrations are available only for a limited part of the sample, due to missing values in the dataset. Only about 25 percent of the full sample has data for ownership concentration.
18
Figure 1
Mean percentages of owner identity categories (left) and mean and median ownership concentrations (right), for each country in the sample.
Figure 2
Time series of mean ETR1 and STR.
Table 1 presents the descriptive statistics for the data for the full sample. Even after winsorizing, REV, LEV and PPE have extreme maximum values, and medians that are relatively far from the mean, thus indicating some skewness. However, to keep the sample truthful, these were not winsorized more. The sample has an ETR1 mean of 28.1 percent, which is higher than the mean STR of 26.7 percent. Of the control variables it is notable that RD has a median of 0, due to the low number of firms which have research and development expenses.
Table 2 presents t-tests for differences in means of two subsamples that the full sample has been split into. The subsamples are constructed based on a criteria of having either a STR of less than or equal to 26 percent, or over 26 percent. A STR of 26 percent is chosen as the threshold for the low and high
0% 10% 20% 30% 40% 50% 60% 70% AT BE DE DK ES FI FR GR IT LU NL PT SE To tal
FINANCIAL CORPORATION FAMILY
0% 20% 40% 60% 80% 100% AT BE DE DK ES FI FR GR IT LU NL PT SE To tal
Mean (CONC) Median (CONC)
0% 5% 10% 15% 20% 25% 30% 35% 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
19 corporate income tax rate subsamples, because it is the median of the full sample, and thus splitting at it will create two samples of close to equal size. Based on the results of the t-tests, the subsamples have statistically significant differences in the means of most of the variables. This argues the case for testing the hypotheses separately for each subsample, in addition to testing them on the full sample, as these differences may cause the result to change.
Table 3 presents the Pearson correlation for the variables, as well as their p-values on the second row. The highest correlation in the table is between SIZE and REV. The correlation is -0.76 and strongly significant. However, it is not above the rule of thumb of +/- 0.8, so both variables will be kept, in order to measure different aspects of firm size. Other than that, the correlations are generally low and thus, multicollinearity isn’t an issue.
Table 1
Descriptive statistics of data. The data consists of 15,329 firm-years from 3,137 firms in selected European countries. ETR1 is the tax aggressiveness measure. CONC indicates the ownership percentage of the controlling owner. STR is the statutory corporate income tax rate of the county the firm is incorporated in. FINANCIAL is a dummy variable indicating firms controlled by a widely held financial institution. CORPORATION is a dummy variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets. LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year.
N Mean Median Sd. Min P25 P75 Max
20
Table 2
T-testing means of variables in subsamples. The subsample "low"consist of firm-years with a STR of 26 percent or below. The subsample "high" consists of firm-years with a STR of above 26 percent. The third column reports the difference in menas between the "low" and the "high" subsample. The last column reports the t-value of t-tests used to test the difference in means. ETR1 is the tax aggressiveness measure. CONC indicates the ownership percentage of the controlling owner. STR is the statutory corporate income tax rate of the county the firm is incorporated in. FINANCIAL is a dummy variable indicating firms controlled by a widely held financial institution. CORPORATION is a dummy variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. LEV is total debt and liabilities. PPE is property, pland and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets. LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year.
Mean "low" Mean "high" Difference in menas t-value
22 5. Results
Results for ownership structure estimates
Table 4 presents the results of regressing owner identities on the effective tax rate, while controlling for year and country characteristics. The results suggest that hypotheses 1, 2 and 3 are supported, as statistically significant and consistent relationships are found between FINANCIAL and ETR1, and FAMILY and ETR1, and an insignificant coefficient for CORPORATION, as expected. Firms owned by widely held financial institutions are more tax aggressive than other firms, and this is supported by the results of all models. Thus it can be said that in the preferred model, which is model 3, and at a 10 percent significance level, these firms have up to 1.1 percent lower effective tax rates than widely held listed firms. This confirms the theory that financial institutions are specialized owners with strong incentives to monitor and control the firms they own, and that their objective is to create shareholder value. Family owned firms have 1.0 percent higher effective tax rate than widely held firms, and the result holds at a 1 percent significance level. Family owned firms thus pay more taxes than other firms do, which is in line with the hypothesis that families or individuals are risk averse owners. They are long term shareholders, who care about the reputation of the firm, and thus they have other objectives than minimizing taxes. Firms owed by widely held corporations have an effective tax rate that is not different from the one of widely held listed firms. This confirms the theory that owners of these firms have too low incentive to engage in active tax avoidance, or that the decision-making is dispersed into too many persons to reach an optimal tax strategy.
Moreover, the controls are generally highly significant, with the exception of RD. On the other hand, the median RD is 0, and the result could have been expected. LEV and ROA lowers the effective tax rate. The effect of LEV is very close to zero, but a one percent increase in ROA lowers ETR1 by 0.3 percent. The positive coefficient of SIZE and REV imply that the size of a firm increases the effective tax rate of the firm, meaning that large firms pay relatively more taxes than small firms, which is consistent with the “political cost” hypothesis from Zimmerman (1983).
23 aggressiveness depending on the identity of their controlling owner is confirmed only for financially owned firms.
Table 4
OLS regression result for full sample. ETR1 is the tax aggressiveness measure. FINANCIAL is a dummy variable indicating firms controlled by a widely held financial institution. CORPORATION is a dummy variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets. LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year. Robust standard errors in square brackets. *p<0.10 **p<0.05 ***p<0.01.
ETR1
1 2 3
b/se b/se b/se
FINANCIAL 0.013** -0.013** -0.011* [0.006] [0.006] [0.006] CORPORATION 0.012*** 0.001 0.003 [0.004] [0.004] [0.004] FAMILY 0.027*** 0.009** 0.010*** [0.003] [0.004] [0.004] LEV -0.000* -0.000** [0.000] [0.000] PPE 0.000*** 0.000*** [0.000] [0.000] RD 0 -0.000* [0.000] [0.000] ROA -0.003*** -0.003*** [0.000] [0.000] REV 0.000*** 0.000*** [0.000] [0.000] SIZE 0.005*** 0.005*** [0.000] [0.000] Constant 0.275*** 0.180*** 0.194*** [0.001] [0.009] [0.010] YearDummies Yes
CountryDummies Yes Yes
R-squared 0.004 0.094 0.103
Adjusted R-squared 0.004 0.092 0.101
Observations 17669 14180 14180
24 owned by financial institutions have on average the same effective tax rate as widely held firms, and thus, the result does not hold when statutory corporate income tax rates are low. For the owner identity CORPORATION, the coefficients insignificant in both of the subsamples. Thus, the result holds under different levels of corporate tax rates. Firms with the owner identity FAMILY have a 1.2 percent higher effective tax rate than widely held firms, which is 0.2 percent more than in the full sample. In the low tax sample the coefficient is insignificant, implying that family owned firms are as tax aggressive as widely held firms when corporate tax rates are low. Thus, the result for family owned firms does not hold at low corporate income tax rates.
Table 5
OLS regression result for sample split at median STR. ETR1 is the tax aggressiveness measure. FINANCIAL is a dummy variable indicating firms controlled by a widely held financial institution. CORPORATION is a dummy variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets.
LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year. Robust standard errors
in square brackets. *p<0.10 **p<0.05 ***p<0.01. High Low ETR1 1 2 b/se b/se FINANCIAL -0.027*** 0.016 [0.007] [0.010] CORPORATION 0.003 0.002 [0.006] [0.006] FAMILY 0.012** 0.006 [0.005] [0.006] LEV 0 -0.000** [0.000] [0.000] PPE 0.000*** 0.000*** [0.000] [0.000] RD 0 -0.000*** [0.000] [0.000] ROA -0.002*** -0.004*** [0.000] [0.000] REV 0.000*** 0.000*** [0.000] [0.000] SIZE 0.004*** 0.006*** [0.001] [0.001] Constant 0.226*** 0.167*** [0.013] [0.014]
YearDummies Yes Yes
CountryDummies Yes Yes
R-squared 0.11 0.101
Adjusted R-squared 0.106 0.097
25 Table 6 presents the result of regressing ownership concentration on the effective tax rate for each owner identity subsample. If differences between countries are ignored, the coefficient for CONC for family owned firms becomes significant and negative. However, ignoring the fact that differences in culture and corporate income tax rates and rules exist is not feasible, as there at least are difference in the corporate income tax rate, which is not controlled for here. The result shows that ownership concentration does not have a linear relationship with the effective tax rate of a firm when year and country fixed effects are controlled for. Using a continuous variable, such as ownership concentration, tests for a linear relationship with the dependent variables. If the dependent variables don’t change linearly with CONC, as was discussed in section 2.2.4., there will be a bias against finding a relationship. Another option, in line with the hypothesis as described in section 2.2., it is possible that the ownership concentrations of the firm in the sample are at a level where the owners are right in between being tax aggressive and becoming risk averse. This means that owners are indifferent with wanting to avoid more taxes. If this is the case, the predicted coefficient should be insignificant. Nevertheless, the results show that ownership concentration does not affect ETR1. This implies that the results found regarding the identities of controlling owners hold at all levels of concentration, meaning that being a controlling owner possesses the owner with enough power to control the manager. However, as already commented in section 4, the data on ownership concentration is limited in the dataset and thus, it is also possible that the result is insignificant due to too small samples. Still, based on these results, hypothesis 4, 5 and 6 are rejected.
26
Table 6
OLS regression result for subsamples with each owner identity. The first two columns contain results for firms controlled by a widely held financial institution. The middle two columns contain the result for firms controlled by a widely held corporation. The last two columns contain the result for firms controlled by one individual or a family. ETR1 is the tax aggressiveness measure. CONC indicates the ownership percentage of the controlling owner. FINANCIAL is a dummy variable indicating firms controlled by a widely held financial institution. CORPORATION is a dummy variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets. LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year. Robust standard errors in square brackets. *p<0.10 **p<0.05 ***p<0.01.
FINANCIAL CORPORATION FAMILY
ETR1
1 2 3 4 5 6
b/se b/se b/se b/se b/se b/se
CONC -0.025 -0.009 -0.022 -0.002 -0.062** -0.039 [0.042] [0.044] [0.028] [0.027] [0.025] [0.025] LEV 0 0.000* -0.000** -0.000** -0.000** 0 [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] PPE 0.000* 0 0.000* 0.000* 0.000** 0.000** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] RD 0.000* 0 0 0 0.000** 0 [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] ROA -0.002 -0.002* -0.003*** -0.003*** -0.003*** -0.003*** [0.001] [0.001] [0.001] [0.001] [0.000] [0.000] REV 0.000*** 0.000*** 0.000*** 0.000*** 0.000*** 0.000*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] SIZE 0.011*** 0.008*** 0.004*** 0.005*** 0.006*** 0.007*** [0.002] [0.002] [0.002] [0.002] [0.001] [0.001] Constant 0.175*** 0.196** 0.313*** 0.182*** 0.299*** 0.186*** [0.068] [0.086] [0.034] [0.039] [0.027] [0.032]
YearDummies Yes Yes Yes Yes Yes Yes
CountryDummies Yes Yes Yes
R-squared 0.102 0.207 0.062 0.13 0.065 0.107
Adjusted R-squared 0.071 0.161 0.048 0.108 0.055 0.09 Observations 553 553 1195 1195 1669 1669
Effect of corporate income tax rate
27 can be seen as slightly ambiguous mainly due to the negative coefficients of FINANCIAL. Finding a linear negative relationship between the variables is unexpected, although, it could be explained by the theory mentioned earlier that simultaneously with declining corporate income tax rates, corporate tax aggressiveness has also been declining.
Table 8 presents the result of regressing owner identity on effective tax rate, while including interaction effects between the owner identity and a dummy variable for high corporate income tax rate. For firms with the identity CORPORATION, the coefficient stays consistent with the previous results. The coefficient for both CORPORATION and its interaction effect with highSTR is insignificant, implying that firms owned by widely held corporations in general have similar effective tax rates as the widely held corporations themselves. Thus, the results show that the tax aggressiveness of firms owned by widely held corporations does not change because of high corporate income taxes. This rejects hypothesis 7.
Table 7
OLS regression result for subsamples with each owner identity. The first two columns contain results for firms controlled by a widely held financial institution. The middle two columns contain the result for firms controlled by a widely held
corporation. The last two columns contain the result for firms controlled by one individual or a family. ETR1 is the tax aggressiveness measure. STR is statutory corporate income tax rate of the country the firm is incorporated in. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets. LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year. Robust standard errors in square brackets. *p<0.10 **p<0.05 ***p<0.01.
FINANCIAL CORPORATION FAMILY
ETR1
1 2 3 4 5 6
b/se b/se b/se b/se b/se b/se
STR -0.204** -0.521** 0.039 0.302* 0.100** 0.234* [0.086] [0.259] [0.058] [0.167] [0.041] [0.130] LEV 0 0.000* -0.000** -0.000** -0.000** -0.000* [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] PPE 0 0 0.000** 0.000** 0.000*** 0.000*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] RD 0 0 0 0 0 0 [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] ROA -0.002* -0.002* -0.003*** -0.003*** -0.004*** -0.003*** [0.001] [0.001] [0.001] [0.001] [0.000] [0.000] REV 0.000*** 0.000*** 0.000*** 0.000*** 0.000*** 0.000*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] SIZE 0.012*** 0.008*** 0.005*** 0.006*** 0.004*** 0.005*** [0.002] [0.002] [0.001] [0.001] [0.001] [0.001] Constant 0.223*** 0.300*** 0.253*** 0.067 0.247*** 0.151*** [0.050] [0.084] [0.029] [0.052] [0.023] [0.044]
YearDummies Yes Yes Yes Yes Yes Yes
CountryDummies Yes Yes Yes
R-squared 0.086 0.179 0.055 0.112 0.057 0.1
28 For firms in the owner identity category FINANCIAL, the results have changed significantly from the results in section 5.1. When country fixed effects are included, the coefficient for FINANCIAL becomes insignificant. However, the coefficient for the interaction effect is relatively consistent and significant. The coefficient is negative, and implies that it high corporate income tax rates change the behaviour of these firms, by making them avoid more taxes. The results shows that firms owned by financial institutions on average have a 3.9 percent lower effective tax rate than widely held firms when corporate income tax rates are high. This confirms the theory from section 5.1. of firms owned by financial institutions becoming tax aggressive when corporate income taxes are high. This accepts hypothesis 8.
When it comes to the category FAMILY, the result from table 8 contradicts earlier results. The coefficient for FAMILY, as well as its interaction effect with highSTR, is insignificant. This implies than family owned firms are not less tax aggressive than other firms, with the exception of financially owned firms. The result also contradicts the result from Chen et al. (2010) of family firm firms paying more taxes than other firms. However, if the insignificance is ignored, the result is in line with other results. Nevertheless, hypothesis 9 is rejected as family owned firms do not become more tax aggressive at higher corporate tax rates.
29
Table 8
OLS regression result for full sample. ETR1 is the tax aggressiveness measure. FINANCIAL is a dummy variable indicating firms controlled by a widely held financial institution. CORPORATION is a dummy variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. highSTR is a dummy variable indicating firm-years with a corporate income tax rate above 26 percent. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets.
ROA is return on assets. LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year. Robust
standard errors in square brackets. *p<0.10 **p<0.05 ***p<0.01.
ETR1
1 2 3
b/se b/se b/se
FINANCIAL 0.019* 0.013 0.013 [0.010] [0.010] [0.010] FINANCIAL x highSTR -0.031** -0.039*** -0.039*** [0.012] [0.012] [0.012] CORPORATION 0.006 -0.002 -0.001 [0.006] [0.006] [0.006] CORPORATION x highSTR 0.017** 0.007 0.007 [0.008] [0.008] [0.008] FAMILY 0.020*** 0.005 0.003 [0.006] [0.006] [0.006] FAMILY x highSTR 0.009 0.008 0.011 [0.007] [0.007] [0.007] highSTR 0.023*** 0.034*** 0.019*** [0.003] [0.004] [0.005] LEV -0.000*** -0.000** -0.000** [0.000] [0.000] [0.000] PPE 0.000*** 0.000*** 0.000*** [0.000] [0.000] [0.000] RD 0 0 -0.000* [0.000] [0.000] [0.000] ROA -0.003*** -0.003*** -0.003*** [0.000] [0.000] [0.000] REV 0.000*** 0.000*** 0.000*** [0.000] [0.000] [0.000] SIZE 0.005*** 0.005*** 0.005*** [0.000] [0.000] [0.000] Constant 0.218*** 0.177*** 0.187*** [0.007] [0.009] [0.010] YearDummies Yes
CountryDummies Yes Yes
R-squared 0.045 0.1 0.105
Adjusted R-squared 0.045 0.098 0.103
Observations 14180 14180 14180
6. Robustness analysis
30 changed once during the timeframe, I assume the owner identity of the firm to be the one that it has been for at least half of the time.
The result of robustness analysis is presented in table 9. The result differs in most aspects from the main result. The coefficient of FINANCIAL is smaller and not significant. Nevertheless, the coefficient is still negative, implying that firms owned by financial institutions may be more tax aggressive than other firms. It is possible that the dataset does not contain sufficient data to statistically support this. The coefficient of CORPORATION, on the other hand, becomes positive and significant. Firms with the owner identity FAMILY are the only ones for which the result stays similar as it is positive and significant. Nevertheless, the coefficient is higher than in the main result. What is notable is the change in the constant. In table 5, when only country dummies were included, it was 0.194. Here, the corresponding number is 0.139. This suggests the overall decrease in corporate income tax rates, which has been mentioned earlier (Bénassy-Quéré, 2014). As was concluded in section 5.1., the result did not hold at lower corporate income tax rates. However, despite discussion of decreasing corporate tax rates, the mean STR of the years used in the robustness check is 26.1 percent. This is only 0.6 percent lower than in the full sample. Thus any conclusions based on this are too ambiguous.
It is difficult to evaluate whether the result presented here is due to a misrepresentative sample, or if the main result actually is incorrect. However, due to the manipulation of the data that was necessary to conduct the robustness check, the dataset risks being bias, and thus I assume that the result of the robustness check does not prevail over the main result.
As an additional robustness check, a second measure of tax aggressiveness will be used, following Jaafar and Thornton (2015). This is calculated as total tax expense divided by operational cash flows:
ETR2i,t = Total Tax Expensei,t / Operational Cash Flowi,t.
31
Table 9
OLS regression result for cross section using five-year averages of data of years 2011-2016. Sample contains 1,593 firms. ETR1i = 𝛼𝛼0 + 𝛽𝛽1 FINANCIALi +𝛽𝛽2 CORPORATIONi + 𝛽𝛽3 FAMILYi + Controls + 𝜀𝜀i. ETR1 is the tax aggressiveness measure.
FINANCIAL is a dummy variable indicating firms controlled by a widely held financial institution. CORPORATION is a dummy
variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development.
REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets. LEV, PPE, RD and REV are scaled by total
assets. REV and SIZE are lagged by one year. Robust standard errors in square brackets. *p<0.10 **p<0.05 ***p<0.01.
ETR1
1 2 4
b/se b/se b/se
FINANCIAL -0.005 -0.004 -0.003 [0.015] [0.015] [0.014] CORPORATION 0.050*** 0.042*** 0.038*** [0.013] [0.013] [0.012] FAMILY 0.057*** 0.050*** 0.030** [0.014] [0.013] [0.013] PPE 0.000** 0.000*** [0.000] [0.000] REV 0.000*** 0.000*** [0.000] [0.000] SIZE 0.005*** 0.005*** [0.001] [0.001] RD 0 0 [0.000] [0.000] ROA -0.001*** -0.001** [0.000] [0.000] Constant 0.233*** 0.186*** 0.139*** [0.013] [0.019] [0.022] CountryDummies Yes R-squared 0.03 0.054 0.153 Adjusted R-squared 0.028 0.049 0.142 Observations 1593 1585 1585
Table 10 reports the result of regressing owner identity on tax aggressiveness, using ETR2 to measure it. The results are not robust to using this measure. The coefficient of FINANCIAL becomes insignificant in all models. The coefficient of FAMILY, on the other hand, are in line with the previous results when country fixed effects are not added. The coefficient of CORPORATION is insignificant, which is in line with the previous result. The correlation between ETR1 and ETR2 is only 0.39 (significant at a 1 percent level), which is low for two variables that are supposed to measure similar effects.
32 To conclude the robustness section, it is mainly the result of firms owned by financial institution that change in the two estimations presented. Firms owned by financial institutions constituted the smallest category of firms in the full sample of the main analysis. As both of the robustness checks were restricted by data availability, it is possible that this category of firms became very small. Because of possibility of this bias, the results of the robustness analysis are not in general seen as representative.
Table 10
OLS regression result for full sample. ETR2i,t = 𝛼𝛼0 + 𝛽𝛽1 FINANCIALi,t +𝛽𝛽2 CORPORATIONi,t + 𝛽𝛽3 FAMILYi,t + Controls + 𝜀𝜀i,t.
ETR2 is the tax aggressiveness measure. FINANCIAL is a dummy variable indicating firms controlled by a widely held financial
institution. CORPORATION is a dummy variable indicating firms controlled by a widely held corporation. FAMILY ins a dummy variable indicating firms controlled by an individual or a family. LEV is total debt and liabilities. PPE is property, plant and equipment. RD is research and development. REV is total revenue. SIZE is the logarithm of total assets. ROA is return on assets.
LEV, PPE, RD and REV are scaled by total assets. REV and SIZE are lagged by one year. Robust standard errors in square
brackets. *p<0.10 **p<0.05 ***p<0.01.
ETR2
1 2 3
b/se b/se b/se
FINANCIAL 0.004 0.01 0.01 [0.020] [0.019] [0.019] CORPORATION -0.001 0.007 0.002 [0.011] [0.010] [0.011] FAMILY 0.025** 0.028*** 0.015 [0.010] [0.010] [0.010] LEV 0 0 [0.000] [0.000] PPE 0.000* 0.000** [0.000] [0.000] RD 0 0 [0.000] [0.000] ROA 0.006*** 0.007*** [0.000] [0.001] REV 0 0 [0.000] [0.000] SIZE 0 0 [0.001] [0.001] Constant 0.255*** 0.198*** 0.177*** [0.003] [0.015] [0.025]
YearDummies Yes Yes
CountryDummies Yes
R-squared 0.001 0.093 0.122
Adjusted R-squared 0.001 0.09 0.114
Observations 4647 4622 3663
7. Conclusion
33 tax avoidance may be a smaller problem than thought. As Lee and Swenson (2012) states, firms in the EU seem to prefer to increase their income, even if this happens at the expense of having to pay more taxes. The findings in this thesis are also more or less in line with this. There is only one group of firms, which is found to aim to be more tax aggressive than other firms, and that is firms owned by widely held financial institutions.
The empirical results of this thesis confirms that controlling owners of different identities have different preferences when it comes to tax avoidance. Financial institutions seem to be tax aggressive owners, as firms they own generally have the lowest effective tax rates. The findings are presumably due to the specialization and the strong objective of value maximisation that this owner category is argued to have. Family owned firms, on the other hand, appear to have other objectives than maximising firm value through tax avoidance as they, on average, have higher effective tax rates than other firms. This is likely to be due to the relative risk averseness of family owners, as they are argued to be long term shareholders and also, they are particular about the reputation of the firm. Firms controlled by widely held corporations fall somewhere in the middle, and do not appear to be more tax aggressive than widely held corporations themselves in general. The insignificant result is possibly due to the management of the owner having low incentives to pursue tax aggressive strategies or, due to the dispersed ownership causing decision making regarding optimal decisions more difficult. The result is only found to hold at corporate income tax rates that are above the EU median of 26 percent. This may be because firms don’t have enough to gain in tax savings when tax rates are low, or because the taxes are so low that the firms already have a competitive advantage and thus in neither case, have strong enough incentives to engage in clever tax strategies.
The effect the ownership concentration has on the tax avoidance remains ambiguous. The analysis does not find a linear relationship between the ownership concentration of a firm with a controlling owner, and its tax aggressiveness. This finding implies that the tax avoidance preference of firms owned by a specific owner type holds at any level of ownership concentration. Thus, it can be concluded that being a controlling owner imposes strong enough governance on the firm to minimize the incentives problem between the owner and the managers, independently of their extent.
34 strategies they may have an international disadvantage. Thus, it is not necessarily in the primary behaviour of these owners to do so but instead, high tax rates demand it from the firm, in order to stay competitive internationally.
The findings of this thesis give further understanding to the tax avoidance done by firms. The results have several implications. For example, governments can put firms controlled by financial institutions under greater scrutiny, as these are in general expected to be more tax aggressive. The findings also suggest that shareholders may be able to affect the level of tax aggressiveness of a firm by becoming controlling owners. Additionally, individual minority investors may use the findings to invest in firms that are more likely to express a level of tax avoidance in line with the risk preferences of the investor.
The findings of the study are, nevertheless, subject to several limitations. A limitation of the empirical part is the exclusion of a dummy variable indicating whether firms have operations in countries known as tax havens. Previous research includes this as a control variable, however, identifying these firms was considered to be outside the scope of this thesis. Another concern is that the empirical part is not addressing the endogeneity concerns mentioned in section 3. Establishing causality between, for example, owner identity and tax avoidance was not possible with the dataset used for this thesis. An additional limitation of the study is that the results are not necessarily applicable in countries with different institutional environment, different tax rules, and different ownership structures of firms. The study was on a sample of firms from EU countries.
35 References
Badertscher, B. A., Katz, S. P., & Rego, S. O. (2013). The separation of ownership and control and corporate tax avoidance. Journal of Accounting and Economics, 56(2), 228-250.
Becht, M. (1997). Strong blockholders, weak owners and the need for European mandatory disclosure. The separation of ownership and control: A survey of 7 European countries.
Bénassy-Quéré, A., Trannoy, A., & Wolff, G. (2014). Tax harmonization in Europe: Moving forward. Les notes du conseil d’analyse économique, 14.
Chen, S., Chen, X., Cheng, Q., & Shevlin, T. (2010). Are family firms more tax aggressive than non-family firms?. Journal of Financial Economics, 95(1), 41-61.
Desai, M. A., & Dharmapala, D. (2006). Corporate tax avoidance and high-powered incentives. Journal of Financial Economics, 79(1), 145-179.
Desai, M. A., & Dharmapala, D. (2009). Corporate tax avoidance and firm value. The review of Economics and Statistics, 91(3), 537-546.
Doupnik, T. and Perera, H. (2015). International Accounting, McGraw-Hill Education: New York, US.
Dyreng, S. D., Hanlon, M., & Maydew, E. L. (2008). Long-run corporate tax avoidance. The accounting review, 83(1), 61-82.
Dyreng, S. D., Hanlon, M., & Maydew, E. L. (2010). The effects of executives on corporate tax avoidance. The Accounting Review, 85, 1163–1189.
Dyreng, S., Hanlon, M., & Maydew, E. L. (2014). Rolling the dice: When does tax avoidance result in tax uncertainty. Available at SSRN, 2374945.
36 European Commission. Taxation and Customs Union. The fight against tax fraud and tax evasion: Role of the EU. URL:
https://ec.europa.eu/taxation_customs/fight-against-tax-fraud-tax-evasion/role-eu_en.
Faccio, M., & Lang, L. H. (2002). The ultimate ownership of Western European corporations. Journal of financial economics, 65(3), 365-395.
Faccio, M., Marchica, M. T., & Mura, R. (2011). Large shareholder diversification and corporate risk-taking. The Review of Financial Studies, 24(11), 3601-3641.
Gunthorpe, D., & Levy, H. (1994). Portfolio composition and the investment horizon. Financial Analysts Journal, 50(1), 51-56.
Gupta, S., & Newberry, K. (1997). Determinants of the variability in corporate effective tax rates: Evidence from longitudinal data. Journal of Accounting and Public Policy, 16(1), 1–34.
Holmström, B. (1999). Managerial incentive problems: A dynamic perspective. The review of Economic studies, 66(1), 169-182.
Jaafar, A., & Thornton, J. (2015). Tax havens and effective tax rates: An analysis of private versus public European firms. The International Journal of Accounting, 50(4), 435-457.
Kester, C.W. (1992). Industrial Groups as Systems of Contractual Governance, Oxford Review of Economic Policy, 8(3): 24–44.
Lanis, R., & Richardson, G. (2012). Corporate social responsibility and tax aggressiveness: An empirical analysis. Journal of Accounting and Public Policy, 31(1), 86-108.
La Porta, R., Lopez‐de‐Silanes, F., & Shleifer, A. (1999). Corporate ownership around the world. The journal of finance, 54(2), 471-517.
37 Lee, N., & Swenson, C. (2012). Are multinational corporate tax rules as important as tax rates?. The International Journal of Accounting, 47(2), 155-167.
Lemmon, M. L., & Lins, K. V. (2003). Ownership structure, corporate governance, and firm value: Evidence from the East Asian financial crisis. The journal of finance, 58(4), 1445-1468.
Pedersen, T., & Thomsen, S. (2003). Ownership structure and value of the largest European firms: The importance of owner identity. Journal of Management and Governance, 7(1), 27-55.
Rego, S. O. (2003). Tax‐avoidance activities of US multinational corporations. Contemporary Accounting Research, 20(4), 805-833.
Remeur, C. (2015) Tax policy in the EU: Issues and challenges. European Parliamentary Research Service. PE 549.001.
Sikka, P., & Willmott, H. (2010). The dark side of transfer pricing: Its role in tax avoidance and wealth retentiveness. Critical Perspectives on Accounting, 21(4), 342-356.
Slemrod, J. (2004). The economics of corporate tax selfishness. National Bureau of Economic Research. No. w10858.
Telegraph, (2012). Google’s tax avoidance is called ‘capitalism’, says chairman Eric Schmidt. December 12, 2012.
Thomsen, S., & Pedersen, T. (2000). Ownership structure and economic performance in the largest European companies. Strategic Management Journal, 689-705.
Zellweger, T. (2007). Time horizon, costs of equity capital, and generic investment strategies of firms. Family business review, 20(1), 1-15.
38 Appendix
Variable definitions.
ETR1i,t = tax aggressiveness of firm i at time t, measured as total tax expensei,t / pretax incomei,t. CONCi,t = the percentage of voting rights held by the majority shareholder
FINANCIALi,t = dummy variable taking the value on if controlling owner is a widely held financial institution, zero otherwise
CORPORATIONi,t = dummy variable taking the value on if controlling owner is the a widely held corporation, zero otherwise
FAMILYi,t = dummy variable taking the value on if controlling owner is one person or several persons from the same family, zero otherwise
STRj,t = statutory corporate income tax rate country j (where the firm is incorporated) at time t LEVi,t = total debt and liabilities / total assets
PPEi,t = plant, property and equipment / total assets
SIZEi,t = the logarithm of the value of total assets, lagged by one year REVi,t = total revenues / total assets, lagged by one year
RDi,t = research and development / total assets (set to 0 when missing)
ROAi,t = profitability of the firm measured as return on assets using P/L before taxes YearDummies = year fixed effects
