Tax Avoidance and Financial Constraints: The influence of Firm- and Country-level Governance

(1)

1

Tax Avoidance and Financial Constraints:

The influence of Firm- and Country-level

Governance

Master thesis by Esperanza Maria van Keulen Student number: s2532549

MSc International Financial Management Supervisor: Dr. H. Gonenc

Co-assessor: Dr. M. Ararat

Abstract

This thesis examines the relationship between tax avoidance and financial constraints, by analyzing a sample of 1,833 firms (11,926 firm-year observations) over the period of 2002-2016. Moreover, because agency problems can lead to managerial rent extraction, the moderating influence of firm- and country-level governance on the relationship is addressed. The results indicate that in general, tax avoidance increases financial constraints because of the risks associated with tax avoidance. While no monitoring role for good country-level governance in efficiently utilizing the cash from avoided taxes is found, good firm-level governance, and the combination of good country-level governance and good firm-level governance, is found to weaken the negative effects of tax avoidance on financial constraints.

Keywords: Tax avoidance, financial constraints, firm-level governance governance,

country-level governance

(2)

2

List of tables

Table 1. Variable definitions ... 26

Table 2. Descriptive statistics by variable ... 28

Table 3. Descriptive distribution by year ... 29

Table 4. Descriptive statistics by country ... 30

Table 5. Univariate comparisons for constrained and unconstrained firms ... 32

Table 6. Correlation matrix ... 33

Table 7. Regression analysis financial constraints and tax avoidance ... 35

Table 8. 2SLS analysis financial constraints and tax avoidance ... 36

Table 9. Regression analysis of the effects of firm- and country-level governance ... 39

Table 10. Regression analysis of subsamples of good and weakly governed firms ... 40

Table 11. Regression analysis of the interplay between firm- and country-level governance ... 42

Table 12. Regression analysis of tax avoidance and financial constraints in subsamples ... 43

(4)

4

1. Introduction

Taxes are a major cost for individuals and companies, and not everyone perceives them as fair. Taxes represent a reduction in disposable income for individuals and a reduction in profits for companies. Consequently, many corporations have been relying on tax avoidance strategies to reduce their tax burden. This can be done legally, utilizing strategies in compliance with the law, but also with more aggressive methods in violation of the law, both belonging to the overarching term tax avoidance (Edwards, Schwab and Shevlin, 2016). Past research has extensively examined tax avoidance, although more attention has been given to the determinants rather than the consequences of tax avoidance (Hanlon, Maydew and Saavedra, 2017). Financial constraint is a determinant of tax avoidance, because financially constrained firms can engage in tax planning behaviour. Financial constraints influence tax avoidance behaviour because firms can reduce their constraints by reducing tax expenses, and as such increase the level of available cash (Edwards et al., 2016). However, the relationship between tax avoidance and financial constraints is actually two-sided. When focusing on this consequence perspective and examining the effect of tax avoidance on financial constraints, it becomes clear that the literature is inconclusive and describes contradicting effects. On the one hand, tax avoidance is associated with risks and can increase the costs of equity and debt (Goh Lee, Lim and Shevlin, 2016; Isin 2018). On the other hand, tax planning strategies can be used as a method to generate more internal funds and reduce financial constraints (Edwards et al., 2016).

(5)

5

(Desai and Dharmapala, 2009). This is an agency problem between shareholders and managers, and it could be addressed through proper governance mechanisms that would limit such self-serving behaviour (Eisenhardt, 1989). Good governance leads to better monitoring (Desai and Dharmapala, 2009), which leads to a reduction in managerial rent extraction. This can for instance reduce the risks for external financers (Isin, 2018) and it can also ensure that the cash generated through tax avoidance is directed towards company goals. Governance is a broad construct, and both firm-level and country-level governance can impact the relationship and lead managers to act more in the interest of shareholders; but this might be even more true for the combination of country- and firm-level governance (Seifert and Gonenc, 2018). This thesis focuses on the effect of tax avoidance on financial constraints by addressing the reverse causality concern, and it examines the impact of both firm- and country-level governance on this relation. It is expected that good firm-level governance, good country-level governance, and a potential interplay of these constructs, will moderate the relationship between tax avoidance and financial constraints. This means that when good governance is added to the relationship, there will either be a less pronounced increase in financial constraints, or a more pronounced decrease.

The proposed relationships are tested with an international sample of 11,926 firm-year observations over 36 countries in the period 2002-2016. The data is gathered from Compustat Global, Asset4 and World Bank and is merged to compose the final sample. The relationships are tested using ordinary least squares (OLS) regressions with lagged variables, and a two-staged least squares (2SLS) regression with an instrumental variable, to control for endogeneity problems due to the reverse causality between financial constraints and tax avoidance.

(6)

6

support is found for a monitoring role of good firm-level governance in more efficiently utilizing cash from tax avoidance, but no support for such a role of good country-level governance is found. Lastly, there is some support for a substituting effect of firm- and country-level governance, in which firm-country-level governance partially reduces the negative effects of tax avoidance on financial constraints in countries with strong governance.

(7)

7

(8)

8

2. Literature review

2.1 Tax avoidance

Tax avoidance is broadly defined as anything that reduces the cash effective tax rate of the firm over a long period of time (Dyreng, Hanlon and Maydew, 2008). Tax planning and tax avoidance are used interchangeably to describe actions taken to reduce the firm’s cash tax liabilities. These can be legal tax strategies in compliance with the law, but also aggressive tax strategies and such methods can be in violation of the law (Edwards et al., 2016). There are many possible strategies a firm can apply to avoid taxes. An example being companies reversing their structure such that the parent company and a subsidiary located in a tax haven switch roles, without any operational changes, which provides significant tax savings (Desai and Dharmapala, 2009).

2.2 Financial constraints influence tax avoidance

There is a direct association between tax avoidance and financial constraints. However, the association is not one-sided but instead two-sided. On the one side, financial constraints can cause tax avoidance. Financial constraints can be defined as frictions preventing the firm from funding all investments desired. This investment inability might for example be a result of an inability to issue equity, credit constraints or illiquidity of assets (Lamont, Polk and Saá-Requejo, 2001).

(9)

9

variation in financial constraints of firms explains variation in tax aggressiveness of firms, because more financially constrained firms could use the aggressive tax strategies as a means of obtaining additional funds. Unlike other strategies to cut costs, tax planning will not negatively affect the operations of a firm and instead of the traditional methods of financing, which are more difficult to access in financially constrained periods, firms raise cash through escalating tax credits or decreasing reported income, which reduces the cash taxes paid (Edwards et al., 2016).

2.3 Tax avoidance influences financial constraints

On the other side, tax avoidance can also impact financial constraints. In this thesis, the focus is on this side of the relationship, examining the consequence perspective. However, literature is inconclusive regarding the nature of this relationship. Arguments can be made for two different impacts, namely tax avoidance increasing financial constraints and tax avoidance decreasing financial constraints.

2.3.1 Tax avoidance increases financial constraints

Literature has documented that tax avoidance is associated with a higher cost of debt (Shevlin, Urcan and Vasvari, 2013; Hasan, Hoi, Wu and Zhang, 2014), as banks bear risks related to the tax avoidance behaviour of the firm. Debt holders have fixed claims on the firm while anticipated tax avoidance benefits are mainly gathered by shareholders. Thus, tax avoidance is associated with downside risks for creditors because the benefits of tax avoidance do not accrue to the debt holders (Hasan et al., 2014). Because banks bear these risks, they protect their exposure by adding a premium (Isin, 2018).

(10)

10

avoidance could subject companies to increased regulatory investigation (Isin, 2018). Consequently, these arguments explain why cost of debt financing increases in association with engaging in tax avoidance. Beladi, Chao and Hu (2018) find that reductions in the terms of loans and increases in the costs of loans are debt financing consequences of tax avoidance.

The increased regulatory audits can lead to firms having to pay extra taxes, or fines and penalties for their tax avoidance behaviour. As a consequence, greater uncertainty is anticipated by equity holders in future after-tax cash flow, which leads to an increase in variance and covariance (Goh et al., 2016). Furthermore, more aggressive tax avoidance strategies that can be employed by companies to reduce the overall tax burden involve complex methods to structure transactions. Such strategies can lead to an increase in risk and hence an increase in variance and covariance of the firm’s cash flows (Goh et al., 2016). This will lead to an increase in cost of equity capital.

These findings indicate that tax avoidance increases cost of debt and cost of equity, thus providing arguments that tax avoidance increases the financial constraints of firms.

2.3.2 Tax avoidance decreases financial constraints

(11)

11

The above arguments explain that tax avoidance can lead to a decrease in the costs of equity and debt. A decrease in such costs of external financing represents a decrease in financial constraints. There are however also other arguments that link tax avoidance to a decrease in financial constraints. According to traditional theories, companies avoid tax to increase firm value because it encompasses a wealth transfer from the government to shareholders (Khurana and Moser, 2013). Tax avoidance would thus lead to increased after-tax value of a company (Desai and Dharmapala, 2008). This stresses the firm benefits of corporate tax avoidance. As explained before, financial constraints can cause tax avoidance because tax avoidance can be used as a method to generate funds internally (Edwards et al., 2016). Here financial constraint is a determinant for tax avoidance, but consequently the increase in generated funds through tax planning means a reduction in experience of financial constraints, and the need to acquire external funds has been reduced through internal generation of funds.

However, Guenther, Njoroge and Williams (2016; 2017) examined how the cash generated through tax avoidance strategies is actually used and allocated. According to them the source of the cash is a determinant of its use. Cash acquired through tax avoidance is subject to more uncertainty because tax authorities may demand additional cash tax payments; after challenging the tax strategies, the authorities might demand repayment (Hanlon et al., 2017). Guenther et al. (2017) find that firms with low financial constraints invest more of the cash funds generated through tax avoidance, and firms with high financial constraints save more. Financially constrained firms keep more of the acquired cash as cash instead of using it to fund desired investments (Guenther et al., 2016).

(12)

12

tax avoidance, this can lead to higher cost of debt and equity. On the other hand, using tax planning as a method to generate funds internally can lead to a reduction in financial constraints. Because the literature is inconclusive, both sides are examined to discover which one is more pronounced.

This leads to the following hypotheses:

H1a: Tax avoidance increases financial constraints.

H1b: Tax avoidance decreases financial constraints.

2.4 Firm-level governance

The above described association between tax avoidance and financial constraints, regardless of the nature of the effect, is not so simple and straightforward. In contrast to the traditional view, the agency theory suggests that tax avoidance provides possibilities for rent extraction and managerial opportunism; managers can use their position of control to benefit privately, through rent diversion (Desai and Dharmapala, 2006). This influences the effects of tax avoidance; and the relation between tax avoidance and financial constraints is thus also affected. Broadly, the agency theory describes the relationship between an agent and a principal (Ross, 1973). A problem that often arises is when the goals and desires of the principal (e.g. shareholders) and the agent (e.g. managers) conflict. It is relatively difficult for the principal to monitor the agent’s behaviour, but proper governance mechanisms could limit the self-serving behaviour of the agent (Eisenhardt, 1989).

(13)

13

Dharmapala, 2006). This effect is particularly strong in poorly governed firms because good firm governance can lead to better monitoring of managerial performance, which would lead managers to act more in the interest of the firm; thus the net effect of tax avoidance on firm value is greater for firms with good governance (Desai and Dharmapala, 2009). A similar idea has been established for cash holdings. Dhaliwal, Huang, Moser and Pereira (2011) find a negative relation between tax avoidance and cash holdings, because cash is easily diverted and therefore more susceptive to agency conflicts. They also find that governance mechanisms limit the negative relationship between tax avoidance and corporate cash holdings as these can lead to better monitoring and therefore limit managerial opportunities for rent diversion, and thus weaken the negative influence on cash holdings and firm value.

The relationship between tax avoidance and financial constraints is also influenced by agency problems, thus making governance crucial (Bayar et al., 2018). With a focus on the consequence perspective of tax avoidance, Bayar et al. (2018) find the relationship between tax avoidance and financial constraints to be dependent on governance. They find that in firms with weak governance mechanisms, tax avoidance is associated with greater financial constraints because in firms with bad governance, tax avoidance coincides with cloudy information environments in which managers can extract rents for private benefits.

(14)

14

and financial constraints. This is consistent with the notion that in firms with good governance, tax avoidance is used effectively and does not exacerbate financial constraints.

Beladi et al. (2018) find that increased information transparency, which indicates good governance, reduces the negative effects tax avoidance has on debt financing. Also, Isin (2018) finds that the premiums for loans, as a consequence of risks associated with tax avoidance, are broadly eliminated for loans that facilitate the alignment of incentives between borrowers and lenders through performance sensitive provisions. These results suggest that good governance can also influence the association. The mechanisms of good governance that lead to increased transparency and incentive alignment, can improve the expected relationship between tax avoidance and financial constraints.

This expected relationship between tax avoidance and financial constraints can also be that more tax avoidance leads to a decrease in financial constraints. Based on the above mentioned arguments, this would mean that good governance would exacerbate the effects of tax avoidance on financial constraints. Good governance leads to better monitoring (Desai and Dharmapala, 2009) and the cash generated through tax avoidance would be directed towards financial constraints reduction more efficiently. A lack of good governance mechanisms, which equals bad governance, would provide more possibilities for managerial rent extraction (Desai and Dharmapala, 2009) and thus weaken the effects of tax avoidance in reducing financial constraints. It would lead to a less efficient allocation of the cash generated through tax avoidance.

(15)

15

moderating effect of governance can be on either one of the expected relationships. When tax avoidance increases financial constraints, good corporate governance will weaken this relationship. Good governance will ensure that tax avoidance leads to a less pronounced increase in financial constraints. The better the corporate governance, the fewer opportunities there are for managerial rent extraction (Desai and Dharmapala, 2009) which reduces the risks for external financers of engaging with the firm (Isin, 2018), this in turn reduces the financial constraints. When tax avoidance leads to a decrease in financial constraints, good corporate governance will strengthen this relationship. Good governance will ensure that tax avoidance leads to a more pronounced decrease in financial constraints. The better the corporate governance, the fewer opportunities there are for managerial rent extraction (Desai and Dharmapala, 2009), the more the cash from tax avoidance will actually be directed towards financial constraints reduction.

This leads to the following hypotheses for empirical examination:

H2a: If tax avoidance increases financial constraints: the better the firm-level governance, the smaller the effect of tax avoidance on financial constraints.

H2b: If tax avoidance decreases financial constraints: the better the firm-level governance, the bigger the effect of tax avoidance on financial constraints.

2.5 Country-level governance

(16)

16

which authority is exercised in a country (Garrido, Gomez, Maicas and Orcos, 2014). Several studies have already documented the influence country-level factors have on tax avoidance.

For example, Bame-Aldred, Cullen, Martin and Parboteeah (2013) examine the relationship between multiple cultural values at the country-level and the likelihood of engaging in illegal tax avoidance behaviour. As such, they find that individualism is positively related to the likelihood of illegal tax behaviour. Humane orientation, defined as the degree to which friendly, caring and generous behaviour towards others is encouraged (House, Hanges, Javidan, Dorfmand and Gupta, 2004), is negatively related. DeBacker, Heim and Tran (2015) discuss the relation between corruption and illegal tax avoidance behaviour. They find that companies in the US, with international owners originating from corrupt countries, avoid more taxes, suggesting the influence of cultural norms and values. These are a few examples of country-level factors influencing tax avoidance behaviour. But given the theoretically established relationship between tax avoidance and financial constraints, it is most interesting and relevant to see how the country-level factor country governance influences this relationship.

Desai, Dyck and Zingales (2007) describe the notion that shareholders and tax authorities have the common objective to reduce managerial rent extraction. Shareholders want the cash generated through tax avoidance to stay within the company, while tax authorities want to reduce tax avoidance altogether, but both are achieved through better monitoring. This statement can be used to argue that an increase in country-level governance, such as through regulatory environmental changes (Jiménez-Angueira, 2018) that lead to better monitoring, can lead to a reduction in managerial rent extraction and as such influence and moderate the relationship between tax avoidance and financial constraints.

(17)

17

Kanagaretnam and Lobo (2014), and they find it to be negative. In societies with high trust norms, managers are more likely to accord to the trust they receive and consider fairness more important (Lee et al., 2014). This higher sensitivity to fairness considerations could either lead to managers being less likely to avoid taxes, or in case of tax avoidance behaviour, a lower likelihood of diverting the extracted rents for private benefits. In societies with high trust norms, it is expected that shareholders treat managers fairly, and rent extraction through agency problems is less likely (Lee et al., 2014). Thus, good country-level governance can lead to trust, which in turn can lead to a reduction in opportunistic behaviour. Therefore, the cash gained from tax avoidance behaviour is more likely to be directed towards the purpose of reducing financial constraints.

When tax avoidance increases financial constraints, good country-level governance will weaken this relationship. Good country-level governance will ensure that tax avoidance leads to a less pronounced increase in financial constraints. When tax avoidance decreases financial constraints, good level governance will strengthen this relationship. Good country-level governance will ensure that tax avoidance leads to a more pronounced decrease in financial constraints.

This leads to the following hypotheses for empirical examination:

H3a: If tax avoidance increases financial constraints: the better the country-level governance, the smaller the effect of tax avoidance on financial constraints.

(18)

18

2.6 The relationship between firm-level and country-level governance

The effectiveness of governance mechanisms should not merely be examined separately, but also together (Cuomo, Mallin and Zattoni, 2016), because governance mechanisms can complement and substitute each other (Ward, Brown and Rodriguez, 2009). Governance is a multidimensional construct, and the total effect is a function of both firm and country governance (Seifert and Gonenc, 2018). As previously explained, firm-level governance and country-level governance can individually lead managers to act more in the interest of shareholders, but this is even more true for the combination of country- and firm-level governance (Seifert and Gonenc, 2018). Internal governance and country characteristics, such as investor protection, are complements (Aggarwal, Erel, Stulz and Williamson, 2009). There is an interplay between internal and external corporate governance mechanisms and this affects corporate tax avoidance (Jiménez-Angueira, 2018). This interplay between firm-level and country-level governance would thus affect the relationship between tax avoidance and financial constraints. These arguments suggest a complementary relationship, which means that firm-level governance and country-level governance strengthen each other and their effect on the main relationship.

(19)

19

controls can take resources away from tax planning strategies, resulting in lower tax avoidance (Jiménez-Angueira, 2018). Furthermore, tighter monitoring and stricter regulations increase the riskiness of tax avoidance as a means to achieve short-term goals, and therefore it induces a reduction in this tax avoidance behaviour (Jiménez-Angueira, 2018). Extending this effect to the relationship between tax avoidance and financial constraints would suggest that, if a firm engages in the risky behaviour of tax avoidance, the improved external monitoring would redirect the manager’s tax avoidance behaviour towards company goals instead of towards their

own private benefits. These arguments lead Jiménez-Angueira (2018) to conclude that weak corporate governance firms reduce their tax avoidance behaviour when regulations are tightened, and as such the positive relation between tax avoidance behaviour and weak corporate governance is dependent on external governance.

Thus, country-level governance and firm-level governance both influence the relationship between tax avoidance and financial constraints. Whether these two governance constructs complement or substitute each other is debatable. On the one hand, the effects can strengthen each other. Good country-level governance in combination with good firm-level governance would together lead to a stronger effect than individually. Thus it would either lead to a less pronounced increase, or a more pronounced decrease in financial constraints as a consequence of tax avoidance. On the other hand, the effects can substitute each other. Good country-level governance in combination with good firm-level governance would not lead to a more pronounced effect together than they would individually.

This leads to the following hypotheses:

(20)

20

(21)

21

3. Methodology

3.1 Data and sample selection

The international sample consists of observations from listed firms to ensure higher data availability, for the period 2002 till 2016. The financial statement data is retrieved from Compustat Global database, the firm-level governance data is retrieved from Asset4 ESG database, and the country-level governance data is retrieved from the World Bank. The sample of observations extracted from Compustat Global is matched and merged with the samples extracted from Asset4 and World Bank. Based on Bayar et al (2018), and in line with other tax avoidance studies, non-positive pretax income observations are deleted because for taxation purposes these are treated differently. Furthermore, financial and utility firms (SIC 4000-4999, 6000-6999 and 9000-9999) are excluded. The final sample consists of 1,833 firms, and 11,926 firm-year observations, in 36 countries for the period 2002 till 2016.

3.2 Variables

3.2.1 Independent variable

(22)

22

obscure making inferences about the tax avoidance behaviour of the firm (Dyreng et al., 2008). Thus using the long run CashETR reduces some of the volatility in the annual CashETR. The data for this variable is collected from Compustat. Since CashETR represents the effective tax rate, firms that avoid more taxes, are assumed to have lower CashETRs.

𝐶𝑎𝑠ℎ𝐸𝑇𝑅𝑖𝑡 = ∑𝑁 𝐶𝑎𝑠ℎ 𝑇𝑎𝑥 𝑃𝑎𝑖𝑑_𝑖𝑡 𝑡=5 ∑𝑁 (𝑃𝑟𝑒𝑡𝑎𝑥 𝐼𝑛𝑐𝑜𝑚𝑒_𝑖𝑡− 𝑆𝑝𝑒𝑐𝑖𝑎𝑙 𝐼𝑡𝑒𝑚𝑠_𝑖𝑡) 𝑡=5 (1) This variable represents the long run CashETR, centered over a five year period. The lag values of one and two years, and the lead values of one and two years are needed to compute this variable. In order to sustain the observations from 2002 and 2003 and 2015 and 2016, which are the first and last two years of the sample period, one year lag values and one year lead values are replaced with the regular value if these lag and lead values are missing. Two year lag and lead values are replaced with the one year lag and lead values if the two year lag and lead values are missing. This is a trade-off between either replacing values or reducing observations, both affecting the accuracy of the results. As stated by Tsikriktsis (2005), often the least accurate method to deal with missing data is deleting observations. This replacement method seems most appropriate because the CashETR is centered around these lag and lead values merely to reduce some of the volatility in the annual CashETR measure. Thus, this method is expected to lead to more accuracy than using the annual CashETR. In case of non-contiguous firm-year observations, a similar replacement method is used to minimize the observations that need to be dropped.

3.2.2 Dependent variable

(23)

23

the SA index (equation 2). In their article, Hadlock and Pierce (2010) define size as the log of assets. They provide the number of years the firm is public as a general definition of age. Given the greater availability of data on the founding year of firms in the dataset of this thesis, age will be defined as the number of years since the firm has been established, and this definition is used to compute the SA index. In this sample, the SA index varies between approximately -2.7 and -10.8. In which -10.8 represents the lowest financial constraints and --2.7 represents the most financial constraints. Thus an increase in SA index represents an increase in financial constraints.

𝑆𝐴 𝑖𝑛𝑑𝑒𝑥 = (−0.737 ∗ 𝑠𝑖𝑧𝑒) + (0.043 ∗ 𝑠𝑖𝑧𝑒2_{) − (0.040 ∗ 𝑎𝑔𝑒)}

(2) The literature does not agree on one overall measure as the best proxy, therefore a robustness analysis is performed with several alternative and additional measures for financial constraints: 1. Kadapakkam, Kumar and Riddick (1998) describe size as an appropriate measure for financial constraints, because smaller companies have fewer access to external funds than larger companies. Different studies (e.g. Almeida, Campello and Weisbach, 2004; Clearly, 2006) use size as a measure to divide their sample into financially constrained and unconstrained firms. The previously established definition of size as the log of total assets is used. Based on Clearly (2006), the sample will be divided in two groups, based on the median of size.

(24)

24

is used. The sample will be divided into financially constrained and financially unconstrained firms based on the median of age.

3. According to Fazzari, Hubbard and Petersen (1988), the likelihood of higher dividend payout ratios is higher with unconstrained firms, and the likelihood of lower dividend payout ratios is higher with financially constrained firms. Based on Whited and Wu (2006), dividends are measured using a dummy variable, that takes the value of 1 if a firm pays dividends and 0 if it does not. Firms that pay dividends are classified as unconstrained, and firms that do not pay dividends are classified as constrained.

3.2.3 Firm-level moderator

Firm-level governance data is extracted from Asset4 ESG database. Throughout the literature, there are many different measures used to capture corporate governance. Corporate governance can be measured with many different constructs capturing for example institutional ownership (Bayar et al., 2018; Jiminéz-Angueira, 2018), board characteristics (Bayar et al., 2018; Seifert and Gonenc, 2018; Jiminéz Angueira, 2018), shareholders rights (Seifert and Gonenc, 2018; Jiminéz-Angueira, 2018), audit committee attributes (Jiminéz-Angueira, 2018) and CEO wealth (Bayar et al., 2018). Asset4 database has an overall corporate governance score which is used by, for example, Cheng, Ioannou and Serafeim (2014). This overall score is comprised of multiple separate constructs such as board functions, board structure, compensation policy, vision and strategy and shareholder rights, and is used as the corporate governance measure in this thesis. The scores can vary from 0 to 100 with higher values corresponding with better corporate governance.

3.2.4 Country-level moderator

(25)

25

2018) and consists of six governance dimensions: 1. voice and accountability, 2. political stability and absence of violence and terrorism, 3. government effectiveness, 4. regulatory quality, 5. rule of law, and 6. control of corruption. Country-level governance is measured as the average of these six categories. The scores can vary from -2.5 to 2.5, with higher values corresponding with better country governance.

3.2.4 Control variables

Following Bayar et al. (2018), the results are controlled for firm-specific variables that influence financial constraints. Cash holdings, computed as cash and short-term investments divided by total assets, is controlled for, because firms that are financially constrained might hold more cash and reduce the dependency on external funds (Denis and Sibilkov, 2010). Capital expenditures, defined as capital expenditures divided by total assets, market to book ratio, defined as market value of equity divided by book value of equity and net working capital, defined as net working capital minus cash divided by total assets, are used to control for investment requirements of a firm. Firms with more capital expenditures and greater opportunities for growth are expected to be financially more constrained because they need more funds for such investments (Korajczyk and Levy, 2003). When the investments of a firm are inefficient, they might face more financial constraints (Subramaniam, Tang, Yue and Zhou, 2011), therefore inefficiency is also controlled for. Following Subramaniam et al. (2011) and Bayar et al. (2018), it is measured as the interaction between a dummy variable that takes the value of one when the Tobin’s Q of a firm is lower than the mean of the industry, and capital

(26)

26

Table 1. Variable definitions

Variable name Definition

Age The number of years the firm is public CapEx Capital expenditures / total assets

CashETR Long run cash effective tax rate; The five-year-centered moving sum for the cash tax paid divided by the five-year-centered moving sum of pretax income minus special items

CashHold Cash and short-term investments / total assets CountryGov Country governance score

DivDum Dividend dummy that takes the value of 1 if a firm pays dividends FirmGov Corporate governance score

Fitted CashETR Estimated value of tax avoidance acquired from stage 1 of the 2SLS analysis

GDPG GDP growth

IndMedCashETR Industry median cash effective tax rate

Ineff Interaction between CapEx and a dummy variable that takes the value of 1 when the Tobin's Q of a firm is lower than the mean of the industry MTB Market value of equity / book value of equity

NWC Net working capital minus cash / total assets

SA Index for financial constraints; measured as -0.737 * Size + 0.043 * Size^2 - 0.040 * Age

Size Log of total assets

Note: This table provides the definitions of the variables used in this thesis.

3.3 Empirical method

(27)

27 𝑆𝐴_𝑖,𝑡 = 𝛽₀+ 𝛽₁𝐶𝑎𝑠ℎ𝐸𝑇𝑅_{𝑖,𝑡−1}+ 𝛽₂𝐹𝑖𝑟𝑚𝐺𝑜𝑣_{𝑖,𝑡−1}+ 𝛽₃𝐶𝑜𝑢𝑛𝑡𝑟𝑦𝐺𝑜𝑣_{𝑖,𝑡−1} + 𝛽₄𝐶𝑎𝑠ℎ𝐸𝑇𝑅_{𝑖,𝑡−1}∗ 𝐹𝑖𝑟𝑚𝐺𝑜𝑣_{𝑖,𝑡−1}+ 𝛽₅𝐶𝑎𝑠ℎ𝐸𝑇𝑅_{𝑖,𝑡−1}∗ 𝐶𝑜𝑢𝑛𝑡𝑟𝑦𝐺𝑜𝑣_{𝑖,𝑡−1} + 𝛽₆𝐶𝑎𝑠ℎ𝐻𝑜𝑙𝑑_{𝑖,𝑡−1}+ 𝛽₇𝐶𝑎𝑝𝐸𝑥_{𝑖,𝑡−1}+ 𝛽₈𝑀𝑇𝐵_{𝑖,𝑡−1}+ 𝛽₉𝑁𝑊𝐶_{𝑖,𝑡−1} + 𝛽₁₀𝐼𝑛𝑒𝑓𝑓_{𝑖,𝑡−1}+ 𝛽₁₁𝐺𝐷𝑃𝐺_{𝑖,𝑡−1}+ ∑ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠_𝑖,𝑡 + ∑ 𝑌𝑒𝑎𝑟𝐷𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀𝑖,𝑡 (3) As an additional analysis for hypothesis 1, a 2SLS with instrumental variables is performed. Bayar et al. (2018) use this method, because it deals with the problem of potential endogeneity in the relationship because of reverse causality. In the first stage of the regression, the tax avoidance measure is predicted using an instrumental variable. Based on Bayar et al. (2018), the instrumental variable for CashETR is the industry median CashETR. The engagement of a firm in tax avoidance is likely influenced by the behaviour of other firms in the industry and furthermore, it is unlikely that the industry median CashETR would have an effect on a firm’s financial constraints (Bayar et al., 2018). Thus, industry median CashETR is a suitable instrumental variable as it is correlated with tax avoidance, which satisfies the validity assumption, and the exclusion restriction is also satisfied because it is not correlated with financial constraints. The first stage of the regression creates a new tax avoidance variable through the use of an instrumental variable (equation 4).

𝐶𝑎𝑠ℎ 𝐸𝑇𝑅𝑖,𝑡 = 𝛾0+ 𝛼1𝐼𝑛𝑑𝑀𝑒𝑑𝐶𝑎𝑠ℎ𝐸𝑇𝑅𝑗,𝑡+ 𝛾1𝐶𝑎𝑠ℎ𝐻𝑜𝑙𝑑𝑖,𝑡+ 𝛾2𝐶𝑎𝑝𝐸𝑥𝑖,𝑡+ 𝛾3𝑀𝑇𝐵𝑖,𝑡

+ 𝛾4𝑁𝑊𝐶𝑖,𝑡+ 𝛾5𝐼𝑛𝑒𝑓𝑓𝑖,𝑡+ 𝛾6𝐺𝐷𝑃𝐺𝑖,𝑡−1+ ∑ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠𝑖,𝑡

+ ∑ 𝑌𝑒𝑎𝑟𝐷𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀𝑖,𝑡

(4) In the second stage of the regression, the model estimated value of tax avoidance is used as the independent variable (equation 5).

𝑆𝐴𝑖,𝑡 = 𝛽0+ 𝛽1𝐹𝑖𝑡𝑡𝑒𝑑 𝐶𝑎𝑠ℎ𝐸𝑇𝑅𝑖,𝑡+ 𝛽2𝐹𝑖𝑟𝑚𝐺𝑜𝑣𝑖,𝑡+ 𝛽3𝐶𝑜𝑢𝑛𝑡𝑟𝑦𝐺𝑜𝑣𝑖,𝑡+ 𝛽4𝐶𝑎𝑠ℎ𝐻𝑜𝑙𝑑𝑖,𝑡

+ 𝛽₅𝐶𝑎𝑝𝐸𝑥_𝑖,𝑡+ 𝛽₆𝑀𝑇𝐵_𝑖,𝑡+ 𝛽₇𝑁𝑊𝐶_𝑖,𝑡+ 𝛽₈𝐼𝑛𝑒𝑓𝑓_𝑖,𝑡+ 𝛽₉𝐺𝐷𝑃𝐺_{𝑖,𝑡−1} + ∑ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠_𝑖,𝑡 + ∑ 𝑌𝑒𝑎𝑟𝐷𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀_𝑖,𝑡

(28)

28

4. Results

4.1 Descriptive statistics

Table 2 reports the summary statistics of all variables used in the analysis. Winsorizing is applied to treat the outliers of the variables. SA, Size, Age, CashHold, MTB, NWC, CapEx and Ineff are winsorized at the 1% level. In accordance with previous tax avoidance studies, observations with negative pretax income are deleted. However, as mentioned by Dyreng et al. (2008), when taxes paid are negative or when taxes paid exceed pretax income, non-meaningful CashETRs arise, with negative CashETRs or CashETRs higher than 100 percent. Therefore, in order to achieve meaningful CashETRs, CashETR is winsorized at the 5% level.

Table 2. Descriptive statistics by variable

Variable N Mean Median Minimum Maximum Std. Dev

SA 11,926 -5.302 -4.672 -10.847 -2.702 1.799 Size 13,195 8.378 8.326 3.829 11.589 1.366 Age 11,942 55.644 40 0 195 44.891 DivDum 13,214 0.791 1 0 1 0.407 CashETR 11,926 0.232 0.239 0.028 0.413 0.100 CashHold 11,926 0.137 0.097 0.002 0.636 0.125 MTB 11,926 3.735 2.741 -2.794 22.883 3.701 NWC 11,926 0.055 0.048 -0.325 0.539 0.145 CapEx 11,926 0.050 0.036 0.002 0.287 0.046 Ineff 11,926 0.023 0 0 0.208 0.036 GDPG 11,926 2.036 2.224 -9.132 25.557 2.325 FirmGov 11,926 63.955 70.770 1.410 97.860 24.492 CountryGov 11,926 1.255 1.283 -0.759 1.960 0.433

Note: This table reports the summary statistics for the variables employed in the empirical analyses, which include the dependent, independent, control and moderating variables. Definitions of the variables are provided in Table 1.

(29)

29

70.770 and 24.492, respectively. The country-level moderator CountryGov has a mean of 1.255, a median of 1.283 and a standard deviation of 0.433.

Table 3 reports the descriptive distribution by year and indicates a gradual increase in observations in the period from 2002 till 2011. This is mainly explained by the gradual increase in availability of corporate governance data in Asset4. From 2010 till 2016 the number of observations have been more stable with no particular pattern. In the sample period, the year 2002 has the least observations and 2016 has the most.

Table 3. Descriptive distribution by year

Year N Percentage 2002 194 1.63 2003 206 1.73 2004 332 2.78 2005 482 4.04 2006 707 5.93 2007 796 6.67 2008 801 6.72 2009 867 7.27 2010 1,060 8.89 2011 1,082 9.07 2012 1,055 8.85 2013 1,068 8.96 2014 1,057 8.86 2015 1,103 9.25 2016 1,116 9.36 Total 11,926 100

Note: This table provides information about the distribution of the observations over the sample period of 2002 to 2016.

(30)

30

Table 4. Descriptive statistics by country

SA CashETR GDPG FirmGov CountryGov

Country N Mean Mean Mean Mean Mean

Australia 729 -4.736 0.231 2.745 65.039 1.595 Austria 49 -6.754 0.276 1.490 38.621 1.568 Belgium 73 -6.421 0.237 1.276 49.972 1.323 Brazil 41 -5.327 0.205 2.362 23.284 0.039 Chile 43 -6.655 0.177 3.583 6.865 1.170 China 58 -3.777 0.312 8.232 17.686 -0.519 Denmark 75 -6.813 0.255 0.913 39.248 1.807 Finland 208 -5.058 0.266 0.633 59.190 1.844 France 458 -5.996 0.270 1.067 56.898 1.214 Germany 365 -6.641 0.286 1.653 32.877 1.492 Greece 62 -5.116 0.279 -0.799 21.753 0.502

Hong Kong SAR. China 270 -4.008 0.178 3.698 35.251 1.448

India 193 -5.191 0.291 7.266 34.118 -0.294 Indonesia 72 -4.874 0.286 5.517 21.262 -0.384 Ireland 56 -4.404 0.157 5.202 67.843 1.490 Israel 53 -5.389 0.194 3.348 39.923 0.619 Italy 87 -5.027 0.327 -0.310 46.755 0.558 Luxembourg 11 -5.626 0.336 2.333 22.553 1.702 Malaysia 57 -4.693 0.257 4.748 37.199 0.332 Mexico 39 -4.644 0.268 2.518 20.094 -0.191 Netherlands 162 -5.933 0.213 1.103 65.283 1.666 New Zealand 40 -4.089 0.236 2.466 52.596 1.803 Norway 90 -4.858 0.242 1.589 55.985 1.717 Philippines 25 -4.445 0.169 6.006 10.690 -0.358 Poland 31 -4.385 0.171 3.258 18.544 0.821 Portugal 40 -6.178 0.216 0.023 48.299 1.011 Russian Federation 40 -5.217 0.221 2.052 33.089 -0.739 Singapore 162 -4.649 0.162 6.078 41.526 1.494 South Africa 183 -5.390 0.267 2.103 67.233 0.240 Spain 105 -5.629 0.258 1.013 48.924 0.883 Sweden 184 -6.259 0.249 2.240 57.210 1.759 Switzerland 281 -5.772 0.211 2.033 52.200 1.734 Thailand 46 -4.875 0.208 3.084 56.232 -0.304 Turkey 47 -4.889 0.175 5.357 22.213 -0.076 United Kingdom 1,535 -5.854 0.217 1.420 74.402 1.425 United States 5,956 -5.117 0.229 1.801 72.708 1.271 Total 11,926 -5.302 0.232 2.036 63.955 1.255

(31)

31

Furthermore, the lowest corporate governance mean is in Chile with a value of about 7 and the highest is in the United Kingdom with an average value of approximately 74. The table provides further details about SA, CashETR and GDPG.

Table 5 provides results of a univariate comparison between financially constrained and unconstrained firms. The sample has been divided based on the four measures of financial constraints. Based on three out of four measures of financial constraints, namely SA, Age and DivDum, there is a significant difference in CashETR between financially constrained and unconstrained firms. The CashETR for constrained firms is between 2.2% and 3% lower than for unconstrained firms, thus indicating that financially constrained firms avoid taxes to a greater extent than unconstrained firms.

Based on three out of four measures of financial constraints, namely SA, Size and Age, unconstrained firms are better internally governed. The results for country governance are mixed. Based on SA and Size as measures of financial constraints, CountryGov is higher for constrained firms. Based on the other two measures there is not enough evidence to indicate a significant difference in CountryGov between constrained and unconstrained firms. These findings are intuitive, as corporate governance is a firm-level variable and investors may associate better corporate governance with less risk and consequently the firm might experience less constraints, and this line of reasoning is not applicable to country-level governance.

(32)

32

Table 5. Univariate comparisons for constrained and unconstrained firms

Measure of financial constraints

SA Size

Constrained Unconstrained Constrained-Unconstrained Constrained Unconstrained Constrained - Unconstrained

CashETR 0.220 0.242 -0.022*** (-12.225) 0.231 0.232 -0.001 (-0.734) N 5,391 6,535 3,772 8,154 FirmGov 63.118 64.646 -1.528*** (-3.393) 57.212 67.074 -9.862*** (-20.816) N 5,391 6,535 3,772 8,154 CountryGov 1.264 1.247 0.017** (2.117) 1.267 1.249 0.018** (2.067) N 5,391 6,535 3,772 8,154 Measure of financial constraints Age DivDum

Constrained Unconstrained Constrained-Unconstrained Constrained Unconstrained Constrained - Unconstrained

CashETR 0.219 0.243 -0.024*** (-13.070) 0.209 0.238 -0.030*** (-13.476) N 5,379 6,547 2,591 9,335 FirmGov 63.238 64.544 -1.306*** (-2.898) 66.866 63.147 3.719*** (6.852) N 5,379 6,547 2,591 9,335 CountryGov 1.260 1.251 0.009 (1.131) 1.255 1.255 0 (-0.009) N 5,379 6,547 2,591 9,335

(33)

33

Table 6. Correlation matrix

SA Size Age DivDum CashETR CashHold MTB NWC CapEx Ineff GDPG FirmGov CountryGov

SA 1 Size -0.3367* 1 Age -0.9948* 0.2861* 1 DivDum -0.2939* 0.1982* 0.2954* 1 CashETR -0.0977* 0.0529* 0.1007* 0.1309* 1 CashHold 0.2176* -0.1971* -0.2006* -0.1727* -0.1043* 1 MTB 0.0294* -0.0939* -0.0224* -0.0283* 0.0306* 0.1570* 1 NWC 0.0207* -0.1154* -0.0201* -0.0548* 0.0304* 0.1381* -0.0959* 1 CapEx 0.1076* -0.0986* -0.0985* -0.0521* -0.0421* -0.0886* 0.0339* -0.1589* 1 Ineff 0.0112 0.0506* -0.0115 -0.0250* -0.0353* -0.1311* -0.1677* -0.0612* 0.4203* 1 GDPG 0.1137* -0.1162* -0.1095* 0.0702* -0.0575* 0.0547* 0.0798* -0.0232* 0.1205* 0.0145 1 FirmGov -0.0856* 0.1877* 0.0854* -0.0886* -0.0332* -0.0952* 0.0307* -0.0155 -0.0699* 0.0627* -0.1727* 1 CountryGov -0.0770* -0.0306* 0.0852* -0.0792* -0.0435* -0.0318* -0.0325* 0.0293* -0.1044* 0.0056 -0.3890* 0.2906* 1

(34)

34

Furthermore, CashHold, MTB, NWC, CapEx and GDPG are control variables that are significantly and positively related to SA. Lastly, the correlation analysis indicates that FirmGov and CountryGov are negatively related to SA, suggesting that better corporate governance and country governance are associated with less financial constraints.

4.2 The effect of tax avoidance on financial constraints

(35)

35

is in line with the expectations based on theory. These models provide initial evidence supporting hypothesis 1a.

Table 7. Regression analysis financial constraints and tax avoidance

OLS

Model 1 Model 2 Model 3 Model 4

b/se b/se b/se b/se

CashETR -0.805** -0.877** -0.891** -0.937** [0.376] [0.374] [0.381] [0.379] CashHold 2.634*** 2.522*** 2.577*** 2.489*** [0.303] [0.301] [0.305] [0.302] MTB -0.003 -0.001 -0.004 -0.002 [0.011] [0.011] [0.011] [0.011] NWC 0.615* 0.588* 0.625** 0.598* [0.317] [0.317] [0.317] [0.317] CapEx 3.860*** 3.502*** 3.772*** 3.468*** [0.908] [0.907] [0.909] [0.907] Ineff -0.522 -0.156 -0.524 -0.191 [0.931] [0.923] [0.929] [0.923] GDPG 0.067*** 0.052*** 0.051*** 0.041*** [0.014] [0.014] [0.014] [0.014] FirmGov -0.005*** -0.005*** [0.002] [0.002] CountryGov -0.221** -0.170* [0.089] [0.092] Constant -5.209*** -4.874*** -4.907*** -4.673*** [0.272] [0.268] [0.337] [0.316]

Industry FE Yes Yes Yes Yes

Year FE Yes Yes Yes Yes

R-squared 0.159 0.164 0.162 0.165

Observations 11,926 11,926 11,926 11,926

Note: This table provides the estimated coefficients from regressing the independent variable and the other firm- and country-level variables on SA. Definitions of the variables are provided in Table 1. The sample period is 2002 – 2016. Standard errors provided in the brackets are robust and clustered at the firm-level. ***, ** and * indicate statistical significance at the 1%, 5% and 10% level respectively.

(36)

36

Table 8. 2SLS analysis financial constraints and tax avoidance

2SLS Stage 1 Stage 2 Stage 1 Stage 2 Stage 1 Stage 2 Stage 1 Stage 2 Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

b/se b/se b/se b/se b/se b/se b/se b/se

CashETR -1.439** -1.585** -1.650** -1.727*** [0.642] [0.637] [0.657] [0.653] CashHold -0.071*** 3.233*** -0.072*** 3.150*** -0.071*** 3.171*** -0.072*** 3.112*** [0.015] [0.291] [0.015] [0.288] [0.015] [0.293] [0.015] [0.290] MTB 0.000 -0.008 0.001 -0.006 0.000 -0.008 0.001 -0.006 [0.000] [0.012] [0.000] [0.012] [0.000] [0.012] [0.000] [0.012] NWC 0.018 0.142 0.018 0.137 0.018 0.165 0.018 0.155 [0.011] [0.299] [0.011] [0.298] [0.011] [0.300] [0.011] [0.299] CapEx 0.006 4.431*** 0.002 4.173*** 0.005 4.297*** 0.002 4.099*** [0.040] [0.790] [0.040] [0.787] [0.040] [0.787] [0.040] [0.784] Ineff -0.086** 0.043 -0.081* 0.379 -0.086** 0.037 -0.081* 0.341 [0.043] [0.958] [0.043] [0.949] [0.043] [0.957] [0.043] [0.951] GDPG -0.001 0.029*** -0.001* 0.021** -0.001* 0.021** -0.001** 0.016* [0.000] [0.009] [0.000] [0.009] [0.000] [0.009] [0.000] [0.009] FirmGov 0.000 -0.004*** 0.000 -0.004** [0.000] [0.002] [0.000] [0.002] CountryGov -0.001 -0.179** 0.000 -0.133 [0.003] [0.082] [0.003] [0.087] IndMedCashETR 0.754*** 0.753*** 0.753*** 0.752*** [0.019] [0.019] [0.019] [0.019] Constant 0.062*** -5.674*** 0.067*** -5.332*** 0.064*** -5.370*** 0.067*** -5.141*** [0.006] [0.183] [0.007] [0.207] [0.007] [0.227] [0.008] [0.230] R-squared 0.344 0.344 0.344 0.344 Partial R-squared for excluded instruments 0.333 0.330 0.326 0.325 F-statistic for excluded instruments 1,659.03*** 1,637.95*** 1,648.76*** 1,636.74*** Observations 11,806 11,806 11,806 11,806 11,806 11,806 11,806 11,806

Note: This table provides the first stage and second stage of the 2SLS analysis. Model 1, 3, 5 and 7 provide the results of the first stage of the analysis, Model 2, 4, 6 and 8 provide the results of the second stage of the analysis, which are the estimated coefficients from regressing the instrumental variable, and the other firm- and country-level variables on SA. Definitions of the variables are provided in Table 1. The sample period is 2002 – 2016. Standard errors provided in the brackets are robust and clustered at the firm-level. ***, ** and * indicate statistical significance at the 1%, 5% and 10% level respectively.

(37)

37

validity for the instrument. The second stage of the analysis confirms the results of the OLS analysis that tax avoidance leads to an increase in financial constraints. Thus together, the OLS and 2SLS analyses provide evidence confirming hypothesis 1a.

4.3 The moderating role of governance in the impact of tax avoidance on

financial constraints

4.3.1 Firm-level and country-level governance

(38)

38

In Table 9, Model 3, the potential effects of both firm-level and country-level governance are added. As indicated by the insignificant estimated coefficient of CashETR in Model 3, there is no significant effect of tax avoidance on financial constraints for firms with low governance in weakly governed countries. This provides indications that the main relationship between tax avoidance and financial constraints in the full sample is mainly driven by countries with strong governance. In Model 3, the estimated coefficient of CashETR is not significant but the interaction effect FirmGov * CashETR is. The marginally statistically significant coefficient of FirmGov * CashETR of 0.024 along with the coefficient of CashETR of -1.305 indicates that tax avoidance leads to a smaller increase in financial constraints for firms with high firm-level governance. This provides further evidence that strong firm-level governance plays a role in monitoring the firm’s management to efficiently utilize taxes. In both, Model 1 and Model 3,

the estimated coefficient of the interaction term FirmGov * CashETR is statistically significant; and therefore there is a statistical difference in the effect of tax avoidance on financial constraints between low firm governance and high firm governance. This provides support for hypothesis 2a. The analysis in section 4.3.2 will identify the relationship between firm-level and country-level governance together.

(39)

39

or controlling the firm’s management in efficiently utilizing the avoided taxes. Together, Model

2 and Model 3 lead to the conclusion that hypotheses 3a and 3b are not supported.

Table 9. Regression analysis of the effects of firm- and country-level governance

Model 1 Model 2 Model 3

b/se b/se b/se

CashETR -2.197*** -0.227 -1.305 [0.746] [0.763] [0.864] CashHold 2.506*** 2.486*** 2.503*** [0.302] [0.302] [0.302] MTB -0.002 -0.002 -0.002 [0.011] [0.011] [0.011] NWC 0.583* 0.608* 0.597* [0.316] [0.318] [0.317] CapEx 3.449*** 3.476*** 3.459*** [0.907] [0.907] [0.906] Ineff -0.141 -0.191 -0.132 [0.924] [0.923] [0.924] GDPG 0.040*** 0.040*** 0.038*** [0.014] [0.014] [0.014] FirmGov -0.010*** -0.005*** -0.011*** [0.003] [0.002] [0.003] CountryGov -0.182** -0.025 0.043 [0.093] [0.172] [0.176] FirmGov * CashETR 0.020* 0.024* [0.012] [0.012] CountryGov * CashETR -0.576 -0.906 [0.619] [0.652] Constant -4.389*** -4.852*** -4.621*** [0.346] [0.362] [0.370]

Industry FE Yes Yes Yes

Year FE Yes Yes Yes

R-squared 0.166 0.166 0.167

Observations 11,926 11,926 11,926

Note: This table provides the estimated coefficients from regressing the independent variable, the other firm- and country-level variables and the firm- and country-level moderators on SA. Definitions of the variables are provided in Table 1. The sample period is 2002-2016. Standard errors provided in the brackets are robust and clustered at the firm-level. ***, ** and * indicate statistical significance at the 1%, 5% and 10% level respectively.

(40)

40

FirmGov. The negative and significant estimated coefficients of CashETR in Table 10, Model 1 and 2, demonstrate that for firms with low firm-level governance, tax avoidance leads to an increase in financial constraints. The insignificant estimated coefficients of CashETR in Model 3 and 4 indicate that for firms with high firm-level governance, tax avoidance does not impact financial constraints. This provides some additional support for the monitoring role of firm-level governance.

Table 10. Regression analysis of subsamples of good and weakly governed firms

Low firm-level governance sample High firm-level governance sample

b/se b/se b/se b/se

CashETR -1.320*** -1.389*** -0.339 -0.365 [0.427] [0.435] [0.506] [0.507] CashHold 2.408*** 2.388*** 2.774*** 2.727*** [0.329] [0.331] [0.431] [0.432] MTB 0.016 0.015 -0.013 -0.013 [0.012] [0.012] [0.016] [0.016] NWC 0.376 0.39 0.674 0.663 [0.336] [0.336] [0.451] [0.451] CapEx 3.465*** 3.413*** 3.566*** 3.566*** [0.989] [0.990] [1.279] [1.278] Ineff -0.537 -0.558 0.404 0.366 [1.062] [1.063] [1.212] [1.210] GDPG 0.058*** 0.048*** 0.068*** 0.063*** [0.016] [0.017] [0.020] [0.020] CountryGov -0.111 -0.288 [0.089] [0.177] Constant -4.989*** -4.820*** -5.669*** -5.474*** [0.323] [0.373] [0.256] [0.291]

Industry FE Yes Yes Yes Yes

Year FE Yes Yes Yes Yes

R-squared 0.181 0.183 0.186 0.188

Observations 4,949 4,949 6,977 6,977

(41)

41

This additional analysis is also performed to examine the effect of country-level governance on the relationship between tax avoidance and financial constraints, and is provided in Table 11, Model 1, 2, 4 and 5. The sample is divided into a good country governance sample and a weak country governance sample based on the median of CountryGov. In the strong country governance sample in Model 1 and Model 2, CashETR is negatively and significantly related to financial constraints. This means that in strongly governed countries, tax avoidance leads to an increase in financial constraints. In the weak country governance sample in Model 4 and Model 5, CashETR is not significantly related to financial constraints. This means that in weakly governed countries, tax avoidance does not influence financial constraints. This provides further evidence that country-governance does not contribute to monitoring or controlling the firm’s management in efficiently utilizing the avoided taxes.

4.3.2 The relationship between firm-level and country-level governance

(42)

42

Table 11. Regression analysis of the interplay between firm- and country-level governance

Strong country governance sample Weak country governance sample Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

b/se b/se b/se b/se b/se b/se

CashETR -1.355*** -1.385*** -4.005*** -0.533 -0.626 -0.195 [0.508] [0.504] [1.141] [0.458] [0.460] [0.875] CashHold 2.526*** 2.350*** 2.366*** 2.551*** 2.469*** 2.457*** [0.404] [0.401] [0.400] [0.340] [0.337] [0.338] MTB -0.024 -0.021 -0.021 0.007 0.01 0.01 [0.015] [0.015] [0.015] [0.013] [0.013] [0.013] NWC 0.425 0.357 0.344 0.643* 0.615* 0.625* [0.438] [0.438] [0.436] [0.370] [0.368] [0.367] CapEx 4.230*** 3.840*** 3.863*** 3.905*** 3.497*** 3.519*** [1.244] [1.247] [1.242] [1.101] [1.083] [1.082] Ineff -1.782 -1.55 -1.457 0.423 1.009 0.986 [1.238] [1.230] [1.230] [1.115] [1.109] [1.106] GDPG 0.105*** 0.094*** 0.090*** 0.024 -0.001 -0.001 [0.020] [0.019] [0.019] [0.018] [0.019] [0.019] FirmGov -0.006*** -0.016*** -0.007*** -0.005 [0.002] [0.004] [0.002] [0.004] FirmGov * CashETR 0.043** -0.007 [0.017] [0.014] Constant -4.651*** -4.325*** -3.823*** -5.454*** -5.253*** -5.382*** [0.348] [0.376] [0.443] [0.771] [0.752] [0.750]

Industry FE Yes Yes Yes Yes Yes Yes

Year FE Yes Yes Yes Yes Yes Yes

R-squared 0.194 0.199 0.202 0.179 0.187 0.187

Observations 6,051 6,051 6,051 5,875 5,875 5,875

Note: This table provides the estimated coefficients from regressing the independent variable, the other firm- and country-level variables and the firm-level moderator on SA in two sample groups. The sample is split based on the median of CountryGov. Definitions of the variables are provided in Table 1. The sample period is 2002-2016. Standard errors provided in the brackets are robust and clustered at the firm-level. ***, ** and * indicate statistical significance at the 1%, 5% and 10% level respectively.

(43)

43

that in the sample of good country governance, a substitution effect is found. Adding good firm governance partially reduces the negative effect of tax avoidance on financial constraints in countries with strong governance.

4.4 Robustness analysis

In order to test the robustness of the results, additional analyses are performed. Table 12 provides an analysis of the main relationship in different subsamples.

Table 12. Regression analysis of tax avoidance and financial constraints in subsamples

All All but US All but US & UK

Model 1 Model 2 Model 3

b/se b/se b/se

CashETR -0.937** -0.938* -1.954*** [0.379] [0.541] [0.607] CashHold 2.489*** 2.306*** 2.013*** [0.302] [0.483] [0.550] MTB -0.002 -0.009 0.02 [0.011] [0.014] [0.017] NWC 0.598* 0.105 0.109 [0.317] [0.448] [0.491] CapEx 3.468*** 3.023** 1.509 [0.907] [1.256] [1.248] Ineff -0.191 -0.379 1.318 [0.923] [1.309] [1.439] GDPG 0.041*** 0.068*** 0.059*** [0.014] [0.016] [0.017] FirmGov -0.005*** -0.008*** -0.003 [0.002] [0.002] [0.002] CountryGov -0.170* -0.099 -0.093 [0.092] [0.096] [0.095] Constant -4.673*** -3.842*** -3.688*** [0.316] [0.755] [0.785]

Industry FE Yes Yes Yes

Year FE Yes Yes Yes

R-squared 0.165 0.193 0.185

Observations 11,926 5,970 4,435

(44)

44

Together, the US and the UK make up more than 60% of the sample. Therefore in Table 12, the analysis is conducted with different subsamples. Model 2 and Model 3 demonstrate that the estimated coefficient of CashETR is still significant and negative when the US or the US and the UK are removed from the sample. These results are consistent with previous evidence.

Furthermore, an additional analysis of the main relationship between tax avoidance and financial constraints is conducted using several alternative measures of financial constraints.

Table 13. Logistic regression with alternative measures of financial constraints

SA Size Age DivDum

b/se b/se b/se b/se

CashETR -1.672*** -0.079 -1.856*** -2.086*** [0.501] [0.425] [0.502] [0.485] CashHold 3.797*** 1.960*** 3.507*** 3.278*** [0.442] [0.377] [0.439] [0.401] MTB 0.007 0.061*** 0.005 0.017 [0.012] [0.011] [0.011] [0.012] NWC -0.087 1.685*** -0.137 1.073*** [0.385] [0.355] [0.384] [0.401] CapEx 3.804*** 6.659*** 3.768*** 1.845 [1.270] [1.145] [1.274] [1.357] Ineff -0.65 -2.941*** -0.519 2.089* [1.276] [1.123] [1.266] [1.260] GDPG 0.035* 0.016 0.033 -0.140*** [0.021] [0.017] [0.021] [0.022] FirmGov -0.003 -0.018*** -0.002 0.006*** [0.002] [0.002] [0.002] [0.002] CountryGov 0.205 0.436*** 0.144 -0.118 [0.134] [0.098] [0.133] [0.101] Constant -1.113 -1.953** -1.062 -0.708 [0.987] [0.980] [0.928] [0.435] Pseudo R-squared 0.101 0.130 0.098 0.129 Observations 11,873 13,169 11,873 13,076

(45)

45

Firms are classified as either unconstrained or constrained based on the median of the measures DivDum, Size, Age and SA. These classifications are used as binary dependent variables in the analysis. The binary variables are computed such that a value of 0 represents firms that are financially unconstrained and a value of 1 represents firms that are financially constrained. Table 13 provides the results of the logistic regressions. The estimated coefficients of CashETR in Model 1, Model 3 and Model 4, in which SA, Age and DivDum are used as measures for financial constraints, are statistically significant and negative. Model 2 indicates a negative but insignificant relationship between CashETR and Size. Overall, these results suggest that an increase in tax avoidance will increase the log odds of a firm being financially constraint. This is consistent with previously found results.

(46)

46

5. Conclusion and discussion

Although tax avoidance is extensively examined, prior research has given more attention to determinants rather than consequences (Hanlon et al., 2017). Financial constraint is not only a determinant of tax avoidance, but also a consequence. This thesis focuses on this consequence perspective and examines the relationship between tax avoidance and financial constraints. In addition, this thesis complements Bayar et al. (2018) by exploring the relationship in an international sample and examining both the effects of country- and firm-level governance, and a potential interplay of these governance constructs.

(47)

47

Overall, tax avoidance increases financial constraints because of the risks associated with tax avoidance behaviour. The results further indicate that governance mechanisms affect the relationship between tax avoidance and financial constraints, because of the existence of agency conflicts between managers and shareholders that can lead to managerial rent extraction (Desai and Dharmapala, 2009). While country-governance does not contribute to monitoring the firm’s management in efficiently utilizing the avoided taxes, good firm-level governance, individually or in combination with good country-level governance, has a role in mitigating the negative effects. But the nature of the relationship between tax avoidance and financial constraints does not change; tax avoidance does not decrease financial constraints. All together these findings indicate that tax avoidance is not a suitable strategy to employ for reducing financial constraints. Though companies could employ certain governance mechanisms to mitigate or weaken the negative effects, the benefits of tax avoidance behaviour do not seem to outweigh the costs and risks, and companies would thus be advised to consider employing different strategies to reduce financial constraints.

(48)

48

employing lagged variables is merely a moderate method to control for reverse causality between tax avoidance and financial constraints. The 2SLS analysis only tackles this problem for hypothesis 1. A final limitation is that, in the employment of these lagged variables, in order to prevent sample size reduction, a replacement method is applied that replaces the lagged value with the current value if the lagged value is unavailable. Although better than reducing the number of observations, this method does provide a limitation.

(49)

49

6. References

Aggarwal, R., Erel, I., Stulz, R., and Williamson, R., 2009. Differences in governance practices between U.S. and foreign firms: measurements, causes and consequences. The Review of Financial Studies 22 (8), 3131-3169.

Almeida, H., Campello, M., and Weisbach, M. S., 2004. The cash flow sensitivity of cash. Journal of Finance 59 (4), 1777-1804.

Bame-Aldred, C. W., Cullen, J. B., Martin, K. D., and Parboteeah, K. P., 2013. National culture and firm-level tax evasion. Journal of Business Research 66, 390-396.

Bayar, O., Huseynov, F., and Sardarli, S., 2018. Corporate governance, tax avoidance and financial constraints. Financial Management 47 (3), 651-677.

Beladi, H., Chao, C. C., and Hu, M., 2018. Does tax avoidance behavior affect bank loan contracts for Chinese listed firms? International Review of Financial Analysis 58, 104-116. Bouckaert, G., and Van de Walle, S., 2003. Comparing measures of citizen trust and user satisfaction as indicators of ‘good governance’: difficulties in linking trust and satisfaction indicators. International Review of Administrative Sciences 69 (3), 329-343.

Cheng, B., Ioannou, I., and Serafeim, G., 2014. Corporate social responsibility and access to finance. Strategic Management Journal 35, 1-23.

Clearly, S., 2006. International corporate investment and the relationships between financial constraint measures. Journal of Banking & Finance 30, 1559-1580.

Cuomo, F., Mallin, C., and Zattoni, A., 2016. Corporate governance codes: a review and research agenda. Corporate Governance: An International Review 24 (3), 222-241.

DeBacker, J., Heim, B. T., and Tran, A., 2015. Importing corruption culture from overseas: evidence from corporate tax evasion in the United States. Journal of Financial Economics 117, 122-138.

Denis, D. J., and Sibilkov, V., 2010. Financial constraints, investment, and the value of cash holdings. The Review of Financial Studies 23 (1), 247-269.

Desai, M. A., and Dharmapala, D., 2006. Corporate tax avoidance and high-powered incentives. Journal of Financial Economics 79, 145-179.

Desai, M. A., and Dharmapala D., 2008. Taxation and corporate governance: an economic approach. In: Schön, W. (Ed.), Tax and Corporate Governance. Berlin, pp. 13-30.

Desai, M. A., and Dharmapala, D., 2009. Corporate tax avoidance and firm value. The Review of Economics and Statistics 91 (3), 537-546.

Tax Avoidance and Financial Constraints: The influence of Firm- and Country-level Governance