Foreign Investor’s Interest and Tax Avoidance: Contingency Perspectives Depending on Coun- try’s Protection Level and Law Systems*

This study investigates the differences in multinational and national firms’ tax avoidance. Furthermore, it investigates the differences between firms’ contingent behaviour, because of a country’s legal system and level of protection for investors. This research takes into account the phenomenon of firms’ tax avoidance. Besides that, it proposes novelties as follows. First, this study highlights that multinational firms tend to avoid taxes more than national ones do. Second, it induces the dividend catering theory related to a country’s level of investor protection. Lastly, it argues that a country’s level of investor protection, and legal system, make firms contingent on their tax avoidance behaviour. This study finds that firms operating in countries with high levels of investor protection and a common law system avoided paying tax more often than firms did in different countries. The findings imply that these firms could grow larger than others. It means that this study suggests there could be economic consequences. One consequence is that countries should increase their levels of investor protection and somehow redefine their legal systems. Therefore, they could enhance their capital markets and subsequently improve their nations’ welfare.


Introduction
Every country encounters obstacles when trying to increase their tax revenues and this can be made more difficult when some companies try to avoid paying tax (Tsakumis, Curatola, and Porcano, 2007). This research extends the work done by Salihu, Annuar, and Obid (2015). Salihu, Annuar, and Obid (2015) suggested that the relationship between the interests of investors and tax avoidance by multinational companies in Malaysia remains unclear, as it does not consider the environmental or state conditions. These two factors influence the level of tax avoidance. This study examines the differences in tax avoidance levels between multinational companies and national companies operating in the same country. This study further examines the differences in corporate behaviour as a result of the different legal systems and investor protection levels between countries. Moreover, this study investigates the confusing results of previous studies by conducting this study in an international setting.
This research is relevant because it has new ideas about the phenomenon of tax avoidance from an international perspective. Tax avoidance is considered to be a double-edged sword. Companies perform tax avoidance to regulate the fluctuations in their cash inflows (Shin and Woo, 2018). Tax avoidance increases corporate liquidity, which creates growth opportunities. Likewise, multinational companies choose to invest in countries with flexible taxation rules (Blonigen and Davies, 2002;Hong and Smart, 2010;Egger, Merlo, and Wamser, 2014). This study proposes three fresh arguments. First, multinational companies tend to avoid taxes. Sikka and Willmott (2010) show that multinational companies can easily avoid taxes by utilising transfer pricing in their subsidiary and parent relationships, which typically exist in developed and developing countries. Multinational companies usually perform tax avoidance based on the understanding that there is no long-term consequence or effect on their reputations (Akhtar et al., 2017). Therefore, this study argues that multinational companies tend to take tax avoidance measures more frequently, in comparison with national firms, because multinational companies aim to maintain capital. From another side, the existence of media re-engineering business activities supports the behaviour of multinational companies.
Second, this study posits the concept of the dividend catering theory, which is about the level of investor protection and a country's legal system. The dividend catering theory formulates that managers strive to always meet investors' demands by paying dividends to maximise the company's stock price (Baker and Wurgler, 2004;Li and Zhao, 2008). Tax avoidance is performed by corporate managers to gain greater corporate liquidity, so that the company can pay dividends. This study argues that a country's environment strongly influences the actions underlying this dividend catering theory. In countries with a high level of investor protection, corporate CEOs bond with investors at a high level. Thus the CEOs must maintain high profit levels and distribute large dividends to the investors. This process then requires the CEOs to perform tax avoidance.
Third, this study includes the contingency of the legal and the country environment that both influence companies' decisions to perform tax avoidance. La Porta et al., (1998) suggested that common law countries have higher levels of protection against creditors and investors in comparison to code law countries. Meanwhile, the legal and country environments indicate there are different levels of protection between countries. This difference in the levels of country protection affects corporate CEOs decisions to avoid taxes. The legal and country environments that protect investors usually give companies the freedom to avoid taxes. The reason is that company CEOs have to maintain capital (capital maintenance concepts). Thus, they tend to provide an adequate return for their investors through good tax planning. This policy agrees with the political cost hypothesis and the tax hypothesis of Watts and Zimmerman (1978).
This study refers to several previous studies and their research constructs' measurements. The parameters used to assess the interests of foreign investors include foreign direct investment (FDI) (Salihu, Annuar, and Obid, 2015;Schwarz, 2009;Blonigen and Davies, 2002). FDI is very beneficial for increasing investment volumes and efficiency, resulting in competition among countries' governments to attract foreign investors (Vua and Noy, 2009;Li and Liu, 2005;Hong and Smart, 2010). This study uses tax avoidance measurements based on four proxies: accounting effective tax rate (ETR accounting), long-term cash effective tax rate (long-term cash ETR), income tax rate ratio in operating cash flow and tax ratio paid on cash flow operations (Salihu, Annuar, and Obid, 2015). This study includes developed and developing countries that embrace common law and code law systems, accompanied by the level of investor protection and the legal environment. The research period is from 2009 to 2016. The countries used in this study each represent the level of investor protection obtained from World Bank data in 2016. The result suggests that firms operating in countries with high levels of investor protection, which also have a common law system, will perform higher tax avoidance. However, they usually have higher company growth in comparison to the others. As a consequence, this study suggests countries create more flexible taxation policies to enable greater company growth.
The structure of this study is as follows: Chapter 2 discusses the previous literature and theoretical basis. Chapter 3 discusses the research methodology. Chapter 4 discusses the results of the study. Chapter 5 discusses the conclusions, limitations and development of future research.

Positive Accounting Theory
There are four hypotheses proposed by the positive accounting theory. However, this study focuses on only two of them: the political cost and tax hypotheses. Watts and Zimmerman (1978) constructed the political cost hypothesis, suggesting that companies lobby governments when legal or accounting standards reduce their profits. The companies bears contract costs because they perform political processes, such as the costs incurred to protect them from and to avoid government regulations (Watts and Zimmerman, 1990). Jensen and Meckling (1976) explained that agents (managers) would not act to maximise the interests of the principals (investors). When managers act it affects their bonding costs. Meanwhile, the political costs align with the size hypothesis, which proposes that the more the big sized firms spend, the higher their political costs are. It means that big companies tend to lower their profits Zim-merman, 1978, 1990). This research posits to what Watts and Zimmerman (1978) suggested. However, the agency relationship is between companies and governments. This study infers that companies do not attempt to bond with governments as the principals.

Tax Hypothesis
A company's management makes discretionary policies to select accounting procedures that could increase or decrease its tax payment (Watts and Zimmerman, 1978). Management make efforts to maximise the company's book value, but this can impact on the corporate earnings. Thus, it also means that earnings affect the company's tax policy (Watts and Zimmerman, 1990). Watts and Zimmerman (1978);(1990) argued that if a company attempts to manipulate the tax rules so it pays less tax, it can earn higher profits. This study takes into account the logical reasonings of Watts and Zimmerman (1978);(1990). Companies maximise their profits, part of which is distributed to their investors. However, this is a contingency perspective because the companies' achievement process depended on how the could get quickly. Dyreng, Hanlon, and Maydew (2008) defined tax avoidance as an increase in cash effectiveness through long-term tax rate reductions. Hanlon and Heitzman (2010) developed the definition of tax avoidance from Dyreng, Hanlon, and Maydew (2008) and defined the widespread avoidance of taxes as the actual tax deductions that companies should pay. The bonus-compensation factor encourages managers to be more aggressive and increase corporate value through tax avoidance (Desai and Dharmapala, 2006). Kim and Zhang (2016) explained that where low legal compliance exists, the fines for law violations are relatively low and the market demand for financial transparency is quite low, causing companies to be more aggressive in avoiding taxes. Braga (2017) showed that the avoidance level during the period when IFRS was adopted increased managements' discretion to encourage tax avoidance. Atwood et al., (2012) suggested that companies performed tax avoidance because of various factors, such as company-specific factors (firm size, leverage, operating costs, firm performance, multinational corporation operations) as well as factors across countries which can affect the tax avoidance of firms, such as tax rates, profit volatility, and institutional factors. Morita (2015) stated that companies that make prepaid tax payments have a lower level of tax deviation. This study intends to expand the inter-state factor that influences companies' tax avoidance, namely the contingency of the legal environment and the level of investor protection.

Tax Avoidance in Multinational Companies Versus National Companies
Taxation policies in a country can increase foreign investment, but many countries often use them to constrain multinational companies from conducting tax planning (Blonigen and Davies, 2002;Hong and Smart, 2010). Taylor and Richardson (2012) researched companies operating in Australia and found that the practices of transfer pricing, capitalisation and profit-refining practices were all closely linked to tax avoidance practices. Salihu, Annuar, and Obid (2015) conducted a study in Malaysia, as a developing country, to show that tax avoidance rates by multinational corporations are closely related to the level of foreign interest reflected by FDI. Hong and Smart (2010) further explained that there is a dilemma with foreign investment by multinational corporations, especially when a country sets its tax rates for them. When a country sets too high a tax rate, it will increase firms' tax planning but decrease the foreign investment from multinational companies. Otusanya (2011) found that multinational corporations in Nigeria sought to earn higher profits, but most of them undertook tax avoidance. Then, the Nigerian government intervened by issuing a regulation to overcome its top companies' classes and leads professional institutions. In contrast, Rego (2003) and Atwood et al., (2012) found that domestic companies in the US reported higher effective tax rates than multinational ones classified as high-income companies. Most of the previous literature indicates that multinational companies tend to avoid paying tax more than the national ones do. The multinational companies usually conduct a transfer pricing strategy. This research also argues that multinational companies tend to avoid taxes by employing their transfer pricing strategies. Therefore, this study formulates the first hypothesis as follows: H1: Tax avoidance by multinational companies operating in a country will be higher than that of the national companies.

Investor Protection and Contingency Level
Different countries experience different rates of tax avoidance so it is still a puzzle that needs to be solved (Desai and Dharmapala, 2006). The level of investor protection is the best proxy for showing the differences in the characteristics of different countries (De-Fond, Hung, and Trezevant, 2007). The result of Haidar's (2009) study showed that the level of investor protection becomes the distinguishing factor between countries. Countries with high levels of investor protection will have higher country growth rates than countries with low investor protection levels. The level of investor protection is always related to the protection gained from the implementation of regulations and the law, so that if a country has a reasonable level of investor protection, then the country's capital market will be strong (La Porta et al., 2000a). Larrain, Tapia, and Urzúa (2017) found that firms with high investor protection levels have higher productivity than those with weaker protection, resulting in higher profits for the firms with high levels of investor protection. The level of investor protection is a legal regime that acts as a contract filter and varies between countries (Bergman and Nicolaievsky, 2007). Shin and Woo (2018) explained that companies which avoid tax will find that their investors demand higher returns, due to the risks associated with the quality of the financial statements. Baker and Wurgler (2004) suggested the dividend catering theory which has three main fundamentals: investors demand cash dividend payouts from the company, the fulfilment of their investors' desires influences the stock price, and the decision on cash dividend payments is based on short term and long term cost and benefit considerations. Thus, based on the dividend catering theory, managers seek to pay dividends to the shareholders. The research of Jun, Li, and Yugang (2017) and Baker and Wurgler (2004) suggested that managers have concerns about meeting the investors' demands for cash dividends and found that the managers' would try to meet the investors' expectations. However, the fulfilment of the demands for dividend payments encourages agency issues. Dividend policies affect a firm's value and are closely related to their agency costs (Braouezec and Lehalle, 2010).
This study considers that multinational companies have agency problems with the governments of the countries where the companies operate. The changes to their dividend policies are caused by changes in the taxation rules (Hanlon and Hoopes, 2014). The high level of investor protection encourages better resources allocations, so that the level of investor protection is closely related to the distribution of corporate profits to shareholders. High investor protection supports the distribution of profits (La Porta et al., 2000b). Alzahrani and Lasfer (2012) also suggested that countries with better investor protection levels encourage higher dividend payouts. Harford, Mansi, and Maxwell (2008) and Shin and Woo (2018) explained that firms operating in countries with low investor protection levels tend to hold cash and provide a low-profit distribution to their investors. However, in countries with high investor protection levels, the companies hold less cash. This study argues that multinational companies operating in countries with good investor protection levels have bonding more easily and readily with their investors, so that managers try to distribute more profits to the investors. This study takes into account that tax avoidance has a positive response, especially when the company doing it has low profitability but must ensure a high dividend payment. Therefore, this study formulates the second hypothesis as follows: H2: Tax avoidance by multinational companies operating in countries with high investor protection levels is greater than the avoidance of companies operating in countries that offer less protection.

The contingency of the Legal Environment
La Porta et al., (1998) explained that countries with a civil law legal environment or code law have weaker shareholder protection levels than countries with the common law code. Goyal and Muckley (2013) confirmed the results of Harford, Mansi, and Maxwell (2008), Alzahrani and Lasfer (2012), and La Porta et al., (2000b). They indicated that firms in countries with high investor protection levels and with common law systems pay higher dividends.
La Porta et al., (1998), andMahoney (2001) argue that the legal characteristics of a country means it has different views to other countries; common law countries have stronger capital markets than code law ones do. This study highlights that companies operating in common law countries show high levels of growth, this is due to the availability of high levels of protection for property and contracts. Jaggi and Low (2000) found that corporate finance disclosure levels in countries with common law environments are higher than are found in code law countries. This research argues that companies operating in countries with high investor protection levels and under a common law legal system will have a high growth rate. Moreover, these companies usually avoid paying more tax. This study infers that they could achieve higher growth rates because the countries they operate in gave them more opportunities to manipulate their taxable profits. In other words, they can evade their taxes and use the money saved to increase their earnings. Furthermore, they can avoid paying tax because the countries' legal environments support them. Therefore, this study constructs the third hypothesis as follows: H3: Tax avoidance by multinational companies operating in countries with high levels of investor protection and common law is higher than the avoidance by companies operating in countries that do not have these things.

Research Sample
This study selects its sample by categorising countries with high and low levels of investor protection. The categorisations are according to the level of investor protection obtained from the World Bank data from 2016. Moreover, this study also identified whether the country applied common law or code law -company data and tax avoidance information were derived from the Osiris -Bureau van Dijk Database. The level of protection for investors was categorised as high if the score was equal to or more than six. It was categorised as being low if the score was less than six. This research collected data from 10 countries. The research period was from 2009 to 2016. The study used a purposive sampling method. It designed its criteria with the following steps. First, this study selected and simultaneously identified national and multinational companies which issued financial statements for the period from 2009 to 2016. Second, it eliminated banking and financial companies. Third, it selected companies reporting positive earnings during 2009 to 2016. Finally, this study identified every company with foreign investors who owned at least 5.00% of the company. Table 1 describes the chosen countries' investor protection levels and legal environments.

Research Variable
This research model extends Salihu, Annuar, and Obid (2015). This study used the tax avoidance level as a dependent variable. It measured the company tax avoidance rate (CTA rate) based on four proxies adopted from Salihu, Annuar, and Obid (2015). The first proxy was the accounting effective tax rate (ETR) or CTA1, measured by income tax expenses divided by pre-tax profits (Chen et al., 2010;Dyreng, Hanlon, and Maydew, 2010). The second proxy was the long-term cash effective tax rate or CTA2, measured by the taxes paid for four years divided by the 4-year pre-tax profits (Chen et al., 2010;Dyreng, Hanlon, and Maydew, 2010). The third proxy was CTA3 which was measured by income tax expenses divided by operating cash flows (Lanis and Richardson, 2012). The fourth proxy, CTA4, was measured using the taxes paid divided by operating cash flows (Hanlon and Heitzman, 2010).
This study used four items as the independent variables. The first independent vari- This study measured foreign variables by using three proxies adopted from Salihu, Annuar, and Obid (2015). The second independent variable was the firm operating level, which showed the tax avoidance of multinational and national firms. This study employed a dummy variable that had a value of "1" for a multinational corporation and "0" for a national company. This study defined multinational companies as companies that stand and operate in other countries, and have foreign investment in their equity accounts.
The third independent variable was the level of investor protection offered by each Dichotomy variable where 1 = company with foreign ownership at 5% and 0 = company with foreign ownership less than 5%.

Foreign3
The proportion of foreign directors in companies with potential influence Nationality Level (NATLEVEL)

Capint
Fixed asset ÷ total asset. country (ANTIDIR Index) as measured by a dummy variable with the value "1" for countries whose investor protection level is equal to and more than six, or "0" for those that are less than six. The fourth independent variable was the legal environment, which was also measured by a dummy variable. The value "1" was for the common law environment, while the value "0" was for code law states. This study used five control variables. The first one was political connections, measured by the number of shares held by government institutions and government-controlled entities divided by the total share ownership of the company (Adhikari, Derashid, and Zhang, 2006). The other four control variables were firm size and leverage, adopted from Atwood et al., (2012) and Chen et al., (2014), along with profit and capital intensity adopted from Chen et al., (2014). Table  2 contains the operational definitions of all the variables along with their formulations.

Statistical Analysis
This study used cross-sectional data analysis. It used regression analysis by ordinary least-squares (OLS). Because of OLS, it performs BLUE (best-linearity unbiased error) standard. It controls the linearity by conducting normality, heteroscedasticity, and multi-collinearity tests. Moreover, this study picked one out of several methods until the requirements were fulfilled so that there were no unbiased errors for each test. This study examined each hypothesis using four different models. This research examined Hypothesis H1 using model 1, and hypotheses H2 and H3 consecutively.

Information:
CTAn it : Tax avoidance by company i in year t. n: Type of CTA measurement proxy from 1 to 4. CTAnit-1: Tax avoidance by company i in previous period (t-1). Foreign investors interest as measured by proxy one for company i in period t. Foreign2 it : Foreign investors interest as measured by proxy two for company i in period t. Foreign3 it : Foreign investors interest as measured by proxy three for company i in period t. Natlevel it : Classification of company i as multinational or national company in year t. ANTIDIR it : Investor protection index for company i in year t, low if <6, high if ≥ 6. LAW it : Dummy law environment, classified as common law or code law. Polcon it : Political connections of company i in year t.

Statistical Results and Discussions
The method of sampling used purposive sampling. The total number of compa-83 nies observed was 150 companies. This study used a cross-sectional analysis unit, so there were 1,050 firm-years available for observation. This research succeeded in collecting the final sample of 919 firm-years, which were used as the analysis unit. It analysed the CTA2 variable using 528 firm-years only. This was because this study avoids the year 2008 as a firm-year due to the economic recession. The Foreign3 variable used 129 firm-years for its analysis since there were limited data to investigate. Table 3 contains the descriptive statistics of the independent and dependent research variables. Table 4 contains the descriptive statistic of the dummy variables.
The average score of CTA1 was 0.63538. It meant that the companies from six countries have 63.538% of their income estimated as their tax obligations. The average of CTA2 was 0.13868. It indicated that 13.868% of the total expenses were paid to the tax office. The average of CTA3 was The variable of law was categorised into common law and code law. The proportion of companies operating in common law countries was 73.9% and the rest (26.1%), operated in code law countries. The Natlevel variable consisted of multinational and national companies. Of the companies analysed, 63.4% were multinational companies and 36.6% national companies. The ANTIDIR variable was classified into countries with high or low levels of investor protection. There were 77% of the companies operating in countries with a high level of investor protection and the other 33% operated in countries with a low level of investor protection. The average of the policy variable was 0.14644, which showed that 14.744% of the companies' equity was owned by the governments of six countries. The average profit was 8.58714. Meanwhile the leverage variable was 0.15001. It indicated that 15% of the total assets of the companies were funded by long-term debt. The average of the Capint variable was 0.48817. The average variable was 6.49311.
This study met the standards for heteroscedasticity and multicollinearity. However, it did not meet the standard for the normality test. This study posits the central limit theorem (CLT), which justifies the normality assumption. This study believes that the size of the sample utilised in this study, which was more than 200, was sufficient; therefore, the data were not considered normal, so that the normality assumption could be neglected (Gujarati, 2003).

Tax Avoidance by Multinational Companies versus National Companies
This study analysed four dependent variables (CTA1, CTA2, CITA 2, CTA4), each one was regressed separately. Each dependent variable had three measurements explained in each sub-model. Sub-model 1A utilised Foreign1 as the control variable. Sub-model 1B used Foreign2 as the control variable and sub-model 1C used Foreign3 as the control variable. The Natlevel variable was the focus for the test of the first hypothesis. Although the Natlevel variable was the main focus, the Foreign1, Foreign2 and For-eign3 variables show the metamorphosis of the Natlevel variable. If the analysis shows an insignificant result for the Natlevel variable (unable to read the Natlevel), then this study concludes with these variables. Table 5 presents the examination result of model 1.
The result of the Natlevel variable was significant at the level of 1% in the CTA1 model (sub-models 1A and 1B). The result of CTA1 model sub-model1C showed that the Foreign3 variable was significant at the level of 5%. Besides that, in sub-Model 1B, the Foreign2 variable was significant at the level of 1%. The result showed that the first hy- pothesis was supported (Ha1 supported) by using the various measurement models. The best model to test the first hypothesis was the CTA1 model. The Lev variable was positive and significant at the level 1% in sub-models 1A and 1B. Meanwhile, sub-model 1C was also positive and significant at the level of 10%. This finding is in line with the previous research done by Dyreng, Hanlon, and Maydew (2008). The Profit variable had a negative correlation and was significant towards those three models at the level of 1%. It indicates that when a company's profit is low, the company is likely to be more aggressive regarding tax avoidance. It corresponds to the research of Shin and Woo (2018). They argued that there is a positive response towards tax avoidance by a company when the company is undergoing a decline in its profits.
Multinational companies have more significant opportunities for tax avoidance. They are usually supported by creating the special purpose vehicles used to transfer the price mechanism, the system of market capitalisation, and the ability of profit smoothing (Taylor and Richardson, 2012). The transfer price mechanism can be a strategy for the multinational firms' tax planning. Multinational companies become the mediators of profit shifting towards their parent companies (Schwarz, 2009). The absence of non-tax fees guaranteed by the multinational companies contributes to tax avoidance (Akhtar et al., 2017). This study found that national companies tend to be less ag-gressive regarding tax avoidance. They have a different focus in comparison with the multinational companies; they do not prioritise their investors. Consequently, national companies do not always intend to provide dividend payments or keep the profits they make. This result contradicts Rego (2003) and Atwood et al., (2012). They stated that the involvement of multinational companies in tax avoidance is lower than national ones.

The Contingency of Investor Protection Level
A separate examination of each dependent variable, along with the four measurements was conducted. This study explained the statistical results for each measurement in the four sub-models. Table 6 shows the summary of the test result for model 2. The Note: ***, **, * is significant at 1%, 5%, 10% examination of Foreign1, Foreign2, Foreign3, and Natlevel variables was split due to multicollinearity. The result shows that CTA2 in sub-models 2A and 2C, the Foreign1, Foreign3 and ANTIDIR variables had a negative correlation and significance at the 5% level. The moderations of the Foreign1*ANTIDIR and Foreign2*ANTIDIR variables show positive correlations and were significant at the level of 1%. The best model to predict the second hypothesis was the model CTA2 (sub-models of 2A and 2C). The result indicates that the second hypothesis was supported (Ha2 supported) by utilising the various measurements.
The Polcon (political connection) control variable showed a positive correlation and was significant at the level of 1%. The result is similar to the research done by Kim and Zhang (2016). This finding contradicts Salihu, Annuar, and Obid (2015). They found that there was a negative correlation between political connections and companies' aggressiveness in their tax avoidance. The Lev variable had a positive correlation and was significant in the CTA2 model 2A sub-model, at the level of 1%. The other control variables -size and profit-had a negative correlation and were significant at the level of 5% towards both sub-models. The findings agree with the findings of Rego (2003) and Atwood et al., (2012). They believed that the bigger the company was, the less aggressive they were in avoiding taxes. The result of the profit variable was similar to the result found by Shin and Woo (2018). The variable of Capint had a negative correlation and was significant at the level of 1% in sub-models 2A and 2C. This result showed that if the capital intensity was low, the tax avoidance would be more aggressive. The Capint variable was related to a depreciation budget which can affect the amount of tax payable.
This study found that multinational companies, which were related to foreign investors and operating in countries with a high level of investor protection, were more aggressive in avoiding taxes. The level of investor protection ensured the investors would receive an income. Harford, Mansi, and Maxwell (2008) stated that companies Note: ***, **, * is significant at 1%, 5%, 10% consecutively. $ is the LAW variable that is omitted and automatically deleted by the SPSS statistical program due to multicollinearity.
operating in countries with a high level of investor protection would have their high dividend payments affected. The companies would attempt to pay the dividend as long as they received a positive response from the markets (Baker and Wurgler, 2004). This study concluded that the investors in companies that operate in countries with a high level of investor protection feel secure and protected. Therefore, they have no problem with the tax avoidance issue as long as the company keeps paying them their dividends. This fact has caused the companies to be more aggressive in avoiding taxes in order to make the companies' cash more flexible.

The Contingency of the Legal Environment
The testing scheme of model 3 was similar to model 2. Table 7 summarises the result of model 3. The result showed significance in the CTA2 model of sub-models 3A and 3C. The Foreign1 variable of sub-model 3A had a negative correlation and was significant at the level of 1%. Meanwhile, the Foreign3 variable of sub-model 3C had a negative correlation and was significant at the level of 5%. The ANTIDIR variable had a negative correlation and was significant at the level of 1% in sub-model 3C. The variables of Foreign1*LAW*AN-TIDIR and Foreign3*LAW*ANTIDIR had positive correlations and were significant at the level of 1%. The best model to predict the third hypothesis was the CTA2 model. The result indicates that the third hypoth-esis was supported (Ha3 supported) by using the various models of measurements.
The Polcon control variable had a positive correlation and was significant at the level of 1% in both sub-models. This result corresponds with the research done by Kim and Zhang (2016). The Leverage variable had a positive correlation and was significant in CTA2 model sub-model 3. Size, as a control variable, had a negative correlation and was significant at the level of 10% (sub-model 3A) and 5% (sub-model 3C). The Profit variable had a negative correlation and was significant at the level of 5%. Lastly, the variable of Capint had a negative correlation and was significant at the level of 1% in both sub-models. The findings for the control variables above were similar to the previous results explained in the findings for model 2.
Corporations which operate in common law countries gave their CEOs more opportunities to avoid taxes. The flexibility given to the CEOs allowed them to provide the payment and keep the profits for the investors as well as the findings from Alzahrani and Lasfer (2012), Harford, Mansi, and Maxwell (2008), and La Porta et al., (2000b). This study suggested that companies operating in common law countries with high levels of investor protection pay higher dividends to their investors. The availability of the protection makes the investors believe that the countries, as well as the companies, will not harm them. The investors' sense of security encourages the tax avoidance of the companies. This study found that corporations which have strong relationships with their foreign investors, while operating in common law countries that have a high level of investor protection will be more aggressive in avoiding paying taxes.

Implications
First of all, this study differentiated between national and multinational companies in their tax avoidance behaviour. Secondly, it investigated those firms' behaviour, depending upon how willing the country they operated in was to protect investors. It, then thirdly, investigated those firms' behaviour depending on the legal system of the country they operated in. Finally, it combined both the conditional factors that influenced the companies decisions to avoid paying tax. This study infers that if the firms did avoid paying tax, it was because both the investors' protection model and the law system applied by the country affected their decision. This study recommends a country to take economic policy. It means that the country should have to do whatever is necessary to bring greater prosperity to the nation. This study infers that two policies are needed to tune a country's capital market. The first policy concerns protection for investors. When a country wants to attract foreign investors to trade in its funds, it should create a legal basis that offers them a great deal of protection. The high level of investor protection has consequences, since it gives the firms opportunities to manage their tax planning. It means that the firms could freely conduct tax avoidance schemes. From another perspective, a country has to establish its level of investor protection through regulations.
The second policy concerns the law system applied by a country. The firms feel that they would be more safe and could operate in a more predictable manner in coun-tries that have a common law system. This study infers that the firms could manipulate their accounting practices to determine their net accounting income due to the common law system. Nevertheless, the firms could not justify the accounting earnings caused by the rigidity of the rules under code law. It means that the firms have fewer opportunities to conduct tax planning. From another perspective, the countries that have adopted a code law system control all the firms operating in their countries with a strict set of rules. Consequently the firms are unable to manage their net accounting income as well as tax planning efficiently and generously.
The third policy is the combination of investor protection and the legal system that a country could have. This study highlighted that most firms probably like to operate in countries with high investor protection and common law systems. In these countries, the firms can manage their accounting earnings and tax planning. In other words, the firms could apply the tax hypothesis that Watts and Zimmerman (1978);(1990) suggested. The firms, then, could acquire greater benefits from their bonding with both local and foreign investors. Moreover, the countries could contract with the foreign investors to invest more of their funds so they create higher future potential cash inflows than countries that have low investor protection and the code law system.
Finally, this study marked that it is two countries that are Singapore and Indonesia. Both countries have flags that are almost similar. Singapore offers high levels of protection for investors and uses the common law legal system. Meanwhile, Indonesia has low investor protection and its legal basis is in code law. Moreover, this study defines Singapore as a developed country and its capital market is advanced. Indonesia is still a developing country, and its capital market is not yet an advanced one. This study concludes that these facts are what caused the differences. It, therefore, constructed the idea that a country has three tools to adjust the advancement of its capital market and national prosperity, which are the level of investor protection, the legal system and the combination of them. What a country such as Indonesia could gain by following this idea is an increase in its gross domestic product, low unemployment, economic stability, a stronger currency and other benefits, because local and foreign investors would feel safe and thus increase their investment. This study argues that the rate of tax avoidance between countries with common law and code law is nearly the same. It emphasises that tax avoidance is easier to do in countries with a common law system, where many companies would generaly choose to increase their invested capital in the legal environment. In other words, this study proposes that closely related to the company's improved its liquidity and growths opportunity. It means that the company's decision to avoid tax in order to maintain its liquidity is considered to have a minimum risk compared to tax evasion.

Conclusion
There are two main objectives of this study. First, this study examines the level of tax avoidance by multinational and national companies. Furthermore, it analyses the companies' behaviour as the result of differences in the levels of investor protection and the legal environment. By utilising four measurements of tax avoidance, this study gains some empirical evidence. First, multinational companies are aggressive in avoiding taxes compared to the national ones operating in the same country. Internally, the importance of foreign investors is reflected through foreign direct investment causing higher tax avoidance by the multinational companies.
Secondly, the high level of investor protection in a country gives a sense of security to the investors. Multinational companies tend to be more aggressive in avoiding taxes if they operate in a country with a high level of investor protection. The strong bonding towards the investors causes the company to fulfil the investor's needs by providing high dividend payments. Therefore, the company becomes more aggressive regarding its tax avoidance. If foreign ownership is low, the company will not pay attention to its profits and will protect its investors well. Thirdly, based on the empirical evidence found, multinational companies operating in countries with a high level of investor protection and the legal environment of common law will tend to avoid taxes more frequently. The protection towards the investors will cause the companies' management and CEOs to avoid taxes in order to keep profits up.
The findings of this study do the upcoming future research. First, the analysis used in this study did not consider the economic uncertainty of specific countries, and the next study might consider that. The next research would be better if it considered differentiating between countries with high uncertainty in comparison to those with low uncertainty. The second reason is that this study did not consider different legal tightness among countries. The result of this study only takes two aspects into consideration: the level of investor protection and the type of law system.
This study recommended that firms could do tax avoidance depending upon how each country applies its laws. Thirdly, the level of tax avoidance in this research did not consider the different tax rates in the different countries; this can influence companies' tax avoidance decisions. It means that the measurement of tax evidence is at the country level. The future research may seek to discover if the level of the tax rate in each country becomes a determinant factor. Finally, this new future study could induce press-media pressure and how it affects the firms' tax avoidance.