The Influence of Tax Avoidance on the Ex-Ante Cost of Equity
Master Thesis
ABSTRACT
This thesis investigates the effect of tax avoidance on the ex-ante cost of equity. By analyzing a sample of 32, 386 firm-year observations over the span of time between 2005-2015, I discover that there is no significant relationship between tax avoidance and the ex-ante cost of equity. This is partially consistent with previous studies. In addition, I analyze how CSR and investor protection affect this relationship. My main analysis and robustness check do not give a strong evidence that these variables have a moderating effect on the association between tax avoidance and the ex-ante cost of equity.
Keywords:
Tax avoidance · Ex-ante cost of equity · Corporate Social Responsibility · Investor Protection ·
JEL Classification: G32; H26
Name: Milena Manafova Student number: S3266680
1
I. Introduction
Tax avoidance has received a lot of attention in the last years as more and more companies have started to use different ways to minimize their taxes. In fact, this minimization has been successful for many of the firms. Dyreng et al. (2008) show that approximately a quarter of the publicly traded US companies are capable of maintaining cash effective tax rates at a level less than 20 percent. Following Goh. et al. (2016), I define tax avoidance as all corporate activities that decrease the taxes that a company has to pay. Previous literature has examined different topics related to tax avoidance, where many of the papers have focused on whether it creates value for the firm. For example, Kiesewetter and Manthey (2017) find that tax avoidance does not create value and Desai and Dharmapala (2009) show that value creation in this case is a function of company governance. Nevertheless, there is one topic that has received little attention in the past and this is the effect of tax avoidance on the cost of equity. There are papers that analyze this effect but their number is limited, which makes their evidence also limited (Goh et al., 2016; Hutchens and Rego, 2013; Cook et al., 2017). I would like to extend the knowledge about this topic, which has been received from their work. More precisely, I examine what is the influence of the cash effective tax rate, which is a measure of tax avoidance, on the ex-ante cost of equity. It is essential to explain why I chose the ex-ante measure for the cost of equity. This measure has become part of more and more studies in the past years and all of them give different argumentations why it is better than the ex-post cost of equity, which is based on historical realized returns. The most important reason is the one, which Gupta (2018) states. He argues that ex-ante is a better measure as it is forward-looking and it depends on expected cash flows whereas ex-post returns are a poor proxy of expected returns as they are backward-looking and do not reflect shocks in cash flow or discount rate.
2
cost of equity changes when companies engage in such activities is very limited. It is important to know more about this change as the cost of equity reflects the investors’ perception about risk and a change in perceived risk can have a big impact on the firm’s value: higher risk would cause higher cost of equity, which would decrease the value of the company. Thus, a firm can prevent its undervaluation by making sure that its cost of equity is minimized. Therefore, it is important to know what can create this minimization before defining the firm’s value. Moreover, cost of capital is one of the financial metrics, which investors consider in evaluating whether certain firm could be a good investment. Thus, based on all of these reasons I believe that examining what is the impact of tax avoidance on the ex-ante cost of equity is useful and relevant. In addition, my thesis extends the more general knowledge regarding tax avoidance and its consequences for both the company and its investors. Moreover, I include moderators in my analysis, which shows what kind of variables can or cannot have an impact on the association between tax avoidance and the ex-ante cost of equity.
Applying the model of Lambert et al. (2007) I argue that tax avoidance can have two contradicting impacts on the ex-ante cost of equity. The first one is that due to the cash tax saving, which it produces and which is riskless, tax avoidance can decrease the cost of equity. The reason is that adding a risk-free component to the firm’s future cash flows decreases the return’s risk, which makes investors requiring a lower return (Goh et al., 2016; Cook et al., 2017). This is consistent with the implications of Capital Market Theory. The second impact is that tax avoidance can increase the variance of the firm’s cash flows and their covariance with the cash flows of other companies. Higher variance and covariance of cash flows would make the cost of equity also higher. This increase can be created by aggressive tax strategies such as transfer pricing, which have complicated transactions (Goh et al., 2016). Another way the variance and the covariance to be increased is the higher possibility of having adjustments done by official authorities and the penalties, which they charge, as these damage the reputation of the companies. Destroyed reputation can alter the way how investors evaluate their expected cash flows (Cook et al., 2017; Graham et al., 2014). Finally, variance and covariance of cash flows can increase due to higher chance for managers to extract resources because there is less transparency when a firm is engaged in tax avoidance. This would make investors uncertain about their expected cash flows (Desai and Dharmapala, 2009).
3
identified as potential ways to mitigate the negative effect of tax avoidance in terms of higher possibility for managers to extract resources. The first moderator is engagement in Corporate Social Responsibility (CSR) as it is related to more transparency and information quality (Kim et al., 2012). More transparent business actions would reduce the information asymmetry and would make it harder for managers to extract resources (Lambert et al., 2007). A similar argument is related also to the other moderator, which is investor protection and which is also associated with higher transparency (Bushman at al., 2004).
By using a sample from the period 2005-2015 with companies from all over the world and from different industries a panel analysis was done. The results show that there is no significant relationship between tax avoidance and the ex-ante cost of equity. Moreover, after careful analysis of the main findings and the robustness check it can be concluded that CSR and investor protection do not have an impact on the aforementioned association.
The rest of the thesis is structured in the following way. The next section discusses theoretical and empirical findings from the past and based on them it develops the hypotheses. After that in Section III the data and the methodology, which have been used, are described. Section IV provides the results from the analysis and finally, in Section V, the final discussion and conclusions are presented.
II. Literature Review and Hypothesis Development
The relationship between tax avoidance and the ex-ante cost of equity
The cost of equity is defined as the compensation that the shareholders demand for providing finance and its level depends on time preferences and the risk related to their returns. Before focusing on the association between tax avoidance and the ex-ante cost of equity it is important to explain why the ex-ante measure is used in the analysis. In the financial theories and literature two measures are used - ex-ante cost of equity, which is mostly based on forecasts or expectations, and ex-post, which is based on historical realized returns. My choice for the former one is based on the increase use of this measure in the literature. For example, the ex-ante cost of equity is used in studies about agency costs (Chen et al., 2011), audit quality (Hope et al., 2009), environmental sustainability (Gupta, 2018) and corporate governance (Chen et al., 2009).
4
not reflect shocks in the cash flows or the discount rate. Moreover, Elton (1999) and Fama and French (1997) claim that historical realized returns are poor proxy of cost of equity. In addition, Chava and Purnanandam (2010) argue that long-time series of historical realized returns have to be used in order to define the true association between risk and return. They state that this is not possible to be done in every research and that is why they prefer the ex-ante measure of cost of equity. This is consistent with Mishra and O’Brien (2005), who examine the relationship between risk and the ex-ante cost of equity of emerging markets.
After clearing why ex-ante cost of equity is used it is time to look at the previous studies that are related to the topic of tax avoidance’s effect on the ex-ante cost of equity. From the definition of cost of equity, it becomes clear that the level of risk is its main determinant and the investors base their decision on how certain their future cash flows are. This leads to the conclusion that the association between tax avoidance and the ex-ante cost of equity depends on the risk that tax avoidance brings. The previous literature regarding this association includes both theoretical and empirical findings. The former has built the basics and the latter has tested what the theory had found with the goal to extend the knowledge about the aforementioned relationship. In this section first the theory’s findings are described and then the empirical evidence from the past are included.
- Theoretical background
5
with the Capital Market Theory, which states that adding a risk-free component to firm’s assets reduces their standard deviation, thus their risk. Lower risk of cash flows makes investors requiring a lower return.
Lambert et al. (2007) apply their model in order to access the impact of information quality on the expected return and risk of a firm. However, the implications of their paper can be used to access any company’s actions that affect the future cash flows of the firm and the risk related to these cash flows. Planning strategies that aim to reduce taxes paid are such actions. Tax avoidance may influence the ex-ante cost of equity by two conflicting ways. The first one is that cash savings caused by good tax avoidance strategies can be risk-free as there is no uncertainty regarding their receiving. More precisely, cash tax savings can be accepted like cash flows appropriated by the company from tax authorities and, thus, this appropriation is expected to increase the future cash flows (Goh et al., 2016). Following the implications of Lambert et al (2007) adding a risk-free component to the firm’s cash flows decreases the risk related to them, so the risk of investor’s returns decreases. Thus, the first type of effect of tax avoidance on the ex-ante cost of equity consists of shareholders receiving more in the end while being exposed at less risk. The second effect consists of the fact that tax avoidance can increase the variance of a company’s cash flows and their covariance with the cash flows of the other firms in the market. This would result in a higher risk of return for shareholders. Both effects have been empirically examined in the past and the findings are described below.
- Empirical background
6
investments made by the firm that increase their after-tax cash flows by adding riskless components to these cash flows, so in the end they receive more money brought by lower risk. A closer look at the findings of Goh et al. (2016) is needed as the authors do not take into account only the benefits of tax avoidance in their analysis. They clearly state that the planning activities that have the goal to reduce taxes paid increase the risk of the returns for shareholders and this could lead to a higher ex-ante cost of equity. The authors argue that the benefits of tax avoidance outweigh its disadvantages and the final results of their analysis prove that. However, this cannot be accepted like a certain fact. For example, Hutchens and Rego (2013) examine how tax risk and the ex-ante cost of equity are related. They measure tax risk by different ways but their main result is regarding the measurement connected to tax reserves of companies. The authors find that the level of company’s reserve for income taxes is significantly positive related to the ex-ante cost of equity. Their findings suggest that shareholders accept tax reserves like reflection of two things: the cash tax savings and the risk caused by tax avoidance, which affects future cash flows. Thus, it can be also the case that the disadvantages of tax avoidance are stronger than its benefits in the eyes of the investors and this leads to a higher ex-ante cost of equity. Therefore, the negative effects of tax avoiding activities have to be carefully analyzed. This is done in the next paragraph by reviewing the empirical findings of previous literature regarding these negative effects.
7
8
consistent with Chen et al. (2011), who analyze a sample of 13,140 firm-year observations for the period from 1990 to 2004. They find that when there is information asymmetry (between managers and investors) and agency problems, the weak corporate governance increases company’s exposure to systematic risk and increases ex-ante cost of equity.
In sum, it has become clear that there are contradicting effects that tax avoidance can have on the ex-ante cost of equity. On the one side, shareholders might require lower rate of return as cash tax savings can be risk-free and they would increase the future cash flows while decreasing the risk of these cash flows. On the other side, the ex-ante cost of equity might increase because of higher variance and covariance of cash flows due to more aggressive tax strategies, uncertainty of returns caused by official authorities’ adjustments, and increased chances for managers to extract firms’ resources. These contradicting effects are consistent with the findings of (Cook et al. 2017), who also examine the effect of tax avoidance on the ex-ante cost of equity. They argue that this association depends on investors’ expectations about the investment related to tax avoidance. More precisely, the authors find that ex-ante cost of equity increases with the magnitude of deviation between actual and expected levels of tax planning activities. When these levels are in line, the shareholders do not require a higher return. This shows that it is hard to define one certain direction of the relationship between tax avoidance and the ex-ante cost of equity. Hence, I argue that there is relationship between them and I empirically investigate whether this association is positive or negative:
H1: Tax avoidance is associated with the ex-ante cost of equity.
Moderating effects on the main association
9
Many studies in the past have empirically analyzed the influence of CSR on firms’ transparency and information quality. Sufi (2007) argues that it is important for firms that require more monitoring to have a good reputation. Companies that are tax aggressive are part of those firms. Bebbington et al. (2008) state that in order to enhance the reputation of a company it is imperative for it to show via CSR disclosure that it is managing the ethical, social, and environmental aspects that are parts of its existence. Lopatta et al. (2016) argue that firms, which have high score on CSR performance, build investor confidence and this leads to lower information asymmetry. Moreover, El Ghoul et al. (2017) show that better CSR improves the communication between company and shareholders by providing better transparency. This is consistent with the finding of Kim et al. (2012), who state that companies engaged in more CSR activities provide more transparent reporting. In addition to these empirical findings, the theoretical analysis of Lambert et al. (2007) shows that transparency reduces the information asymmetry between companies and their shareholders. All of these studies build a confidence that the higher information quality associated with CSR engagement can reduce the problem of tax avoidance related to the agency issue. More specifically, with more CSR there would be higher transparency about company’s activities, which would make it harder for managers to extract firm’s resources. Lower information asymmetry and less opportunistic behavior would reduce firm’s risk exposure in the eyes of the shareholders. This would result in a situation where it is less likely that tax avoidance would lead to a higher ex-ante cost of equity as now investors would be more certain about their cash flows in the presence of tax avoidance activities within the company. In fact, an increase in information quality in terms of disclosure is associated with a lower cost of capital (Lambert et al., 2007). More precisely, this effect is achieved in a direct as well as an indirect way. The direct way is that better information quality affects market participants’ assessment about future cash flow distribution. The indirect way is that quality of disclosure affects real decisions that alter the future cash flow distribution (Lambert et al., 2007).
Therefore, I argue that because CSR has been related to higher information quality, which is associated with a lower cost of capital in general, firms that use tax avoiding and that are CSR oriented would have a lower ex-ante cost of equity. Therefore, my next hypotheses are the following:
10
H2b: CSR engagement makes the effect of tax avoidance in decreasing the ex-ante cost of equity more pronounced.
As it is mentioned above, better transparency and communication with investors can reduce the possibilities of managers to divert tax savings of the company for their own consumption. Bushman et al. (2004) argue that countries with good investor protection have higher financial transparency. Moreover, investor protection is associated with more corporate risk-taking. The reason is that when shareholders are protected, the managers are able to extract less cash and benefits for their private consumptions (John et al., 2008). In addition, Chen at al. (2010) show that family firms are less tax aggressive compared to their non-family counterparts. However, tax avoidance used by family firms increases when there is better outside monitoring. The reason is that effective monitoring reduces the aforementioned rent extraction possibilities related to tax avoidance, which have been the main concern of family firms about engaging in tax planning activities. In fact, only countries with strong corporate governance are associated with higher tax revenues while the corporate tax rates increase compared to their previous levels according to the empirical analysis of Desai et al. (2007). Moreover, Goh et al., (2016) claim that when there is better outside monitoring and firms have higher information quality, it is more likely that tax avoidance leads to a lower ex-ante cost of equity. This is consistent with the general findings of Chen et al. (2011), who analyze the impact of shareholder rights at firm level on the cost of equity and the moderating effect of the agency problem from free cash flow on this relationship. They find that strong shareholder rights lead to a lower cost of equity and that this effect is stronger for firms that face severe agency issues from free cash flows. Even though Chen et al. (2011) examine shareholder rights at a firm level and not within a country, their findings show that more protected investors in general would lead to less agency problems and a lower cost of equity as shareholders are exposed to less risk. In consistency with these results Hail and Leuz (2009), who empirically analyze a sample of different firms from 45 countries over the period 1990-2005, reach an important conclusion. More precisely, they show that good investor protection can lead to a lower cost of equity and that effective legal rules make it difficult for managers to extract private benefits from firm’s resources.
11
H3a: Investor protection makes the effect of tax avoidance in increasing the ex-ante cost of equity less pronounced.
H3b: Investor protection makes the effect of tax avoidance in decreasing the ex-ante cost of equity more pronounced.
III. Data and Methodology
MethodologyAfter preparing my data, I tested the hypotheses by the use of panel analysis. As a first step only the main hypothesis is tested, so the basic regression includes only the ex-ante cost of equity as the depended variable and tax avoidance as the independent one together with control variables. In the next steps, I add each moderator and the interactions, so their effect on the relationship between tax avoidance and the ex-ante cost of equity can be seen. The complete regression model is shown in Equation (1):
𝑅𝑖𝑡 = 𝛼 + 𝛽𝐶𝐸𝑇𝑅𝑖𝑡−1+ 𝜓 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 + 𝛾𝐶𝑆𝑅𝑖𝑡−1+ 𝛿𝐶𝑆𝑅𝑖𝑡−1∗ 𝑇𝐴𝑋𝑖𝑡−1+
𝜂𝐼𝑁𝑉𝑃𝑅𝑂𝑖𝑡−1+ 𝜅𝐼𝑁𝑉𝑃𝑅𝑂𝑖𝑡−1∗ 𝑇𝐴𝑋𝑖𝑡−1+ 𝑌𝑒𝑎𝑟𝐹𝐸 + 𝐹𝑖𝑟𝑚𝐹𝐸 + 𝜀𝑖𝑡−1 (1)
where 𝑅𝑖𝑡 is the measure of the ex-ante cost of equity capital; 𝐶𝐸𝑇𝑅𝑖𝑡−1 is tax avoidance measure; 𝐶𝑆𝑅𝑖𝑡−1 is the measure for CSR; 𝐶𝑆𝑅𝑖𝑡−1∗ 𝐶𝐸𝑇𝑅𝑖𝑡−1 refers to the moderating effect of CSR; 𝐼𝑁𝑉𝑃𝑅𝑂𝑖𝑡−1 refers to the measure of investor protection; 𝐼𝑁𝑉𝑃𝑅𝑂𝑖𝑡−1∗ 𝐶𝐸𝑇𝑅𝑖𝑡−1 is
the moderating effect of investor protection; 𝑌𝑒𝑎𝑟𝐹𝐸 and 𝐹𝑖𝑟𝑚𝐹𝐸 are time and firm fixed effects. I follow the procedure of Goh et al. (2016) and lag the tax avoidance variable, the moderators, the moderating effects, and all the controls with 1 year, so time 𝑡 − 1, to ensure that shareholders have information about companies’ tax avoidance activities before defining the expected returns. The reason is that at least a few months are required in order the investor to be able to process the information and evaluate properly what their expected return should be. In my analysis I use a period of around four months after the end of the previous fiscal year to measure the ex-ante cost of equity.
Sample selection
12
each country is obtained from the Economic Freedom of the World: 2017 Annual report. Finally, the data for control variables is taken from Compustat and Datastream. Following Goh et al. (2016) and Cook et al. (2017) I exclude firms in the financial industries (SIC codes 6000 – 6999). The reason is that the business nature of such firms differs and estimating their performance can be problematic. The sample selection captures all companies in Datastream ASSET4 ESG as only there CSR information is available. Thus, in order to be able to test all hypotheses the final sample is based on matching Compustat and I/B/E/S with the company universal list of Datastream ASSET4 ESG. There are 7,181 firms in this Datastream list, so the analysis depends on what kind of information for these firms can be found in the other two data sources. The final sample size used in the regression analysis consists of 32, 386 firm-year observations for the whole period of 2005-2015. It contains companies from all over the world and the countries that have the biggest number of representatives are Australia, United Kingdom, and Japan. Because matching Datastream list with Compustat North America was limited for this amount of time, I managed to include US and Canadian companies but their number is not very high compared to firms from other countries: only 759 US and 77 Canadian firms. Moreover, firms included in the sample are mostly from the manufacturing (SIC code 2000 – 3999) and the transportation, communications, electric, gas and sanitary industry (SIC code 4000 – 4999).
Main variables
- Dependent variable
I measure ex-ante cost of equity based on the well-known Dividend Discount Model, which is related to one-year holding period. The choice arises completely from book and theoretical implications, which are the basis for valuation. However, as in this analysis the ex-ante return is used, I adapt the formula in a way that it does not reflect the historical but the future returns. More precisely, I define the current ex-ante cost of equity at time 𝑡 as the realized return at time 𝑡 + 1. The way of finding the realized return at time 𝑡 + 1 is derived from Equation (2):
𝑃𝑡= 𝐷𝑡+1+ 𝑃𝑡+1 (1 + 𝑘𝑡+1)
13
𝑘𝑡+1= 𝐷𝑡+1+ 𝑃𝑡+1− 𝑃𝑡 𝑃𝑡
(3) The data here is taken from Datastream in order to obtain all variables needed to calculate the realized return.
Thus, the ex-ante cost of equity, 𝑅, at time 𝑡 is the realized return at time 𝑡 + 1: 𝑅𝑡 = 𝑘𝑡+1
(4) It is important to clarify one thing related to 𝑃𝑡 used in computing the realized return at time 𝑡 + 1. As it is mentioned above, investors need time to process the information related to tax avoidance and that is why all the variables and moderating effects on the right hand side in Equation (1) are lagged. From here it becomes clear that 𝑃𝑡 should already reflect the information connected to firm’s tax avoidance in period 𝑡 − 1.
For robustness of the results two alternative measures of ex-ante cost of equity are applied. Regarding the first one I follow Harris et al. (2003) and I measure the ex-ante cost of equity based on the constant dividend growth model:
𝑅_𝐶𝐷𝐺𝑀𝑡 = 𝐷𝑡+1 𝑃𝑡 + 𝑔
(5) where 𝑅_𝐶𝐷𝐺𝑀𝑡 is the ex-ante cost of equity, 𝐷𝑡+1 is the dividend per share that is expected to
be received next year at time 𝑡 + 1 , 𝑃𝑡 is the current share price, and 𝑔 is the dividends’ expected long-term growth rate for each firm. I include “CDGM” in the name of the ex-ante cost of equity’s variable in order to be clear that it is a different measure and for a better understanding of the analysis. The data regarding future values is taken from I/B/E/S based on analysts’ forecasts. The current price is obtained from Datastream.
The second robustness check related to the measurement of ex-ante cost of equity is based on the familiar earnings price ratio (EPR). It is similar to the model developed by Easton (2004) except for its assumption that the earnings growth is zero:
𝑅_𝐸𝑃𝑅𝑡 =
14
where 𝑅_𝐸𝑃𝑅𝑡 is the ex-ante cost of equity for company and it is equal to the firm’s earning
price ration, 𝐸𝑃𝑆𝑡+1 is analysts’ forecast of earnings per share for one-year ahead, and 𝑃𝑡 is the current share price. I/B/E/S and Datastream are the sources of data used here. I include “EPR” in the name of the ex-ante cost of equity’s variable again for better understanding. The observations with negative EPS are excluded as they would give negative ex-ante cost of equity, which won’t be realistic and useful for the analysis.
An initial main restriction regarding all three measure of the ex-ante cost of equity is that only values between 0 and 1 are included (or in percentage 0% and 100%). In addition, I winsorized the variable at 1%. The reason is that in this way more realistic results would be achieved and their interpretation could be more relevant. Otherwise, there would be observations with ex-ante cost of equity that is negative or that is too big. For example, when the main measure of the depended variable is used, the final sample of 32, 386 observations together with the excluded observations have ex-ante cost of equity with a mean of 0.196, a minimum of -0.715, and maximum of 2.583. This shows that their exclusion is a right decision; otherwise, outliers and unrealistic values of the ex-ante cost of equity would make the results very unreliable.
- Tax avoidance
In order to measure tax avoidance, I am going to use an approach, which is similar to the one used by Dyreng et al. (2008). The authors measure tax avoidance only by cash effective tax rates (CETR). The GAAP ETRs are not used because of certain limitations that they impose. First, Dyreng et al. (2008) criticize that GAAP ETRs reflect only annual data. Second, tax expense consists of current tax expense and deferred tax expense. One of the characteristics of tax avoidance is that it includes accelerating deductions and deferring income for tax reasons relative to book reasons. This results in reduced current tax and increased future tax. As GAAP ETRs consists of both current and deferred taxes, they cannot account for such tax avoidance forms. Moreover, Dyreng et al. (2008) claim that using the current tax expense in the numerator, which is the case in GAAP ETRs, can be problematic.
15
initial list of companies is from ASSET4 ESG, as it is mentioned above. This leads to less firms that have non-missing values for all parts of the cash effective tax rate measure for every year. For instance, there are only 6,737 available observations of Cash ETR (CETR) for different firms of the Datastream list for all years. Thus, in order to have more observations of each variable used in this research and to be able to test all hypotheses, the long-term measure cannot be used. This makes obvious the fact that the tax avoidance measure is not perfect; however, it is inappropriate to use the long-term variation here. Cook et al. (2017) also used the same procedure in their study about tax avoidance and the ex-ante cost of equity. This shows that using this kind of measure is not problematic. The second change recommended by Dyreng et al. (2008) is to use cash taxes paid rather than GAAP tax expenses. The cash type includes tax benefits of employee stock options and the GAAP type does not. Moreover, GAAP effective tax rate is affected by changes in estimate like valuation allowance whereas cash effective tax rate is not affected by such changes. This change is taken into account here and the cash taxes are used.
The annual cash effective tax rate for firm 𝑖 can be found in the following way:
𝐶𝑎𝑠ℎ 𝐸𝑇𝑅𝑖𝑡 = 𝐶𝑎𝑠ℎ 𝑡𝑎𝑥 𝑝𝑎𝑖𝑑𝑖𝑡
𝑃𝑟𝑒𝑡𝑎𝑥 𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑡− 𝑆𝑝𝑒𝑐𝑖𝑎𝑙 𝑖𝑡𝑒𝑚𝑠𝑖𝑡
The data for all parts is taken from Compustat. Following Cook et al. (2017) all observations with negative cash tax paid and negative pretax income are excluded. The reason is that in this way only meaningful 𝐶𝑎𝑠ℎ 𝐸𝑇𝑅𝑖𝑡 (CETR) can be derived - no rates that reflect negative tax
16
Moderators
17
property rights taken again from the Global Competitiveness Report shows how clearly defined and well protected the property rights are. Forth component is the one of military interference in rule of law and politics. The data for it is based on the International Country Risk Guide and it indicates the involvement of the military in politics. Fifth, integrity of the legal system, which is also based on the International Country Risk Guide, reflects both the strength of the legal system and observance of the law. Sixth component is legal enforcement of contracts, which is taken from the World Bank’s Doing Business. Last, the component of regulatory costs of the sale of real property indicates the time costs and the monetary costs of transferring ownership of property. The data for this is taken also from the World Bank’s Doing Business. In the annual report of Economic Freedom of the World for 2017 two additional components are included for the determination of the final index but they are not related with investor protection (INVPRO), so they are not going to be explained here in details.
Control variables
Following the paper of Goh et al. (2016) there are two sets of control variables. Datastream provides data for all controls that are needed. The first set is related to the ex-ante cost of equity. The size of the firm should be included and it is measured by the natural log of total assets. The choice for its measure is based on Wasiuzzamn (2015). Thus, in the main analysis this measurement is included and it is represented by the variable lnSize (lnSIZE). Additionally, the size itself measured just by total assets is included as a control for a robustness check of the regression including the main measure of the ex-ante cost of equity. In the main analysis I also include the book-to-market ratio (BM). It is measured by dividing total assets by the sum of market capitalization and total debt. The leverage ratio (LEV) is also part of the controls, measured by total debt-to-asset ratio based on book values. I expect a positive relationship between the ex-ante cost of equity and book-to-market ratio (BM) and between the ex-ante cost of equity and leverage ratio (LEV). I expect a negative relationship between the ex-ante cost of equity and firm’s natural logarithm of size (lnSIZE).
18
In fact, all of the controls except the one related to SG&A are used not only by Goh et al. (2016) but also in the analysis of Cook et al. (2017). Moreover, all controls are measured in consistency with the measurements used in these papers except for size, where they use market capitalization, and for book-to-market ratio, where they use its natural logarithm. However, for my analysis the measures, which I chose, are more appropriate. Additionally, all of them are winsorized at 1 percent, so the effect of outliers is limited. I decided to follow different a procedure with the control variables and only winsorize them without excluding certain observation in advance as there are no specific limits or ranges, which should be followed regarding the controls’ values.
IV. Results
Descriptive statistics
In Table 1 the descriptive statistics on the regression variables can be found. The mean cost of equity (R) is 32.5%, which is comparable to the results of Hutchens and Rego (2013). The mean firm in my sample has cash effective tax rate (CETR) of -19.9%, which is close to the mean found in the analysis of Cook et al. (2017). The average natural logarithm of firm (lnSIZE) is 16.461, the result for leverage (LEV) is 0.236, and the mean of book-to-market ratio (BM) is 0.995. The mean of capital expenditures (CAPEX) scaled by lagged total assets is 0.064. The mean of research and development expenses (R&D) scaled by lagged total assets is 0.032. The mean of selling, general, and administrative expenses (SGA) scaled by lagged total assets is 0.203.
19
Table 1: Descriptive Statistics
Notes: Reported are the descriptive statistics for the sample firms. The sample contains 32, 386 firm-year observations (7181 firms) over the period of 2005–2015. The descriptive statistics for all variables are based on the main measurement of the ex-ante cost of equity. Description of variables can be found in the section of data and methodology. The values of the ex-ante cost of equity and tax avoidance are restricted to be between 0 and 1. Additionally, the ex-ante cost of equity and all control variables are winsorized at 1 percent.
Variables Observations Mean Std. Dev. Minimum Maximum
Dependent Variable R 14,683 0.325 0.245 0.003 0.960 Explanatory Variables CETR 6,737 -0.199 0.160 -0.988 0 Moderators CSR 19,334 1.015 0.623 0.107 22.55 INVPRO 32,375 7.096 1.246 3.5 8.9 Control Variables lnSIZE 29,248 16.461 3.173 9.415 23.960 LEV 29,237 0.236 0.180 0 0.757 BM 27,840 0.995 0.582 0.116 3.165 CAPEX 28,380 0.064 0.066 0.001 0.396 R&D 14,058 0.032 0.054 0 0.343 SGA 23,025 0.203 0.208 0.005 1.175 Empirical findings
20
effect between Cash ETR (CETR) and investor protection is added in Model 2. More precisely, the coefficients of both the interaction term and Cash ETR (CETR) are insignificant. A separate model that shows what are the results when only investor protection is included, so without interaction, is not possible as this variable is omitted. In Models 3 and 4 I followed the same procedure like the one with investor protection but now the CSR is taken into account. However, CSR is not omitted like the investor protection and it is reported in the table. In Model 3, where it is only the variable of CSR itself included, the coefficient of Cash ETR (CETR) still stays insignificant. This is the case also when the interaction between CSR and Cash ETR (CETR) is included in Model 4.
Model 5 represents the full regression analysis that was made for this research. The cash effective tax rate (CETR) has a negative but insignificant coefficient of -0.504. Additionally, no control variable is significant. This is valid also for the interactions between Cash ETR (CETR) and CSR and between Cash ETR (CETR) and investor protection (INVPRO). Finally, the adjusted R-square is equal to 0.222.
Table 2: Correlation matrix
21
Table 3: Regression outputs
Notes: The table contains five OLS regression models. The coefficients for each variable are included together with the standard errors, which are inside the parentheses below the coefficients. The depended variable is the ex-ante cost of equity (R). Model 1 represents the results of the main hypothesis’s test including the controls. Model 2 shows what the influence of interaction with investor protection is. A separate model only with the investor protection (INVPRO) is not possible as it is omitted. Model 3&4 include the other moderator, CSR, and the interaction with it. Model 5 is the full model, where each variable and interaction are tested. Recall that all explanatory variables including controls and moderators are lagged one year relative to the ex-ante cost of equity. ***, ** and *, indicate significance at 1%, 5% and 10% respectively.
Variables/Models Model 1 Model 2 Model 3 Model 4 Model 5
CETR 0.068 (0.068) -0.556 (0.500) 0.076 (0.069) 0.071 (0.187) -0.504 (0.523) Control variables LEV -0.161 (0.098) -0.162* (0.098) -0.142 (0.102) -0.142 (0.102) -0.142 (0.102) lnSIZE -0.002 (0.016) -0.003 (0.016) -0.001 (0.017) -0.001 (0.017) -0.002 (0.017) BM -0.013 (0.032) -0.015 (0.032) -0.015 (0.033) -0.015 (0.033) -0.016 (0.032) CAPEX -0.222 (0.236) -0.214 (0.236) -0.280 (0.236) -0.280 (0.237) -0.272 (0.237) R&D 0.018 (0.383) -0.003 (0.385) 0.024 (0.388) 0.023 (0.393) 0.004 (0.395) SGA 0.073 (0.100) 0.072 (0.100) 0.041 (0.109) 0.041 (0.109) 0.042 (0.109) Moderators CSR 0.004 (0.028) 0.005 (0.037) 0.006 (0.037) Interactions INVPRO X CETR 0.083 (0.068) 0.075 (0.068) CSR X CETR 0.005 (0.140) 0.011 (0.140) Constant 0.393 (0.273) 0.403 (0.272) 0.388 (0.290) 0.388 (0.291) 0.400 (0.290)
Year Fixed Effects Yes Yes Yes Yes Yes
Firm Fixed Effects Yes Yes Yes Yes Yes
Adjusted R-squared 0.224 0.225 0.222 0.221 0.222
Observations 1,791 1,791 1,742 1,742 1,742
Robustness check
22
(R_CDGM) increases by 0.299 percentage points. In addition, the interaction between Cash ETR (CETR) and CSR is insignificant. However, the interaction between Cash ETR (CETR) and investor protection (INVPRO) is significant at 10% and its coefficient is -0.037. The negative coefficient here means that when there is more investor protection, the increase in the ex-ante cost of equity caused by tax avoidance is with lower magnitude. More precisely, with more investor protection in presence 1 percentage point increase in tax avoidance would lead to 0.262 percentage points increase in the ex-ante cost of equity, which is lower than the 0.299 percentage points found above. 0.262 is calculated by summing the coefficients of Cash ETR (CETR) and the interaction term, so 0.299 and -0.037 respectively. Thus, the overall results found with the main measure of the ex-ante cost of equity are not supported with the use of this alternative measure. Here, there is significantly positive relationship between tax avoidance and the ex-ante cost of equity and one of the moderators have significant impact on this relationship. As it is described above, the results with the main measure show no significant coefficients. Table 5 shows the results related to the other alternative measure – the earning-price ratio (EPR). There, the coefficient of Cash ETR (CETR) in the full model is negative and insignificant, which is consistent with the main finding. Moreover, Table 5 shows that the interaction of CSR and Cash ETR (CETR) is insignificant, which is also consistent with the findings of the main analysis. However, here the interaction between Cash ETR (CETR) and investor protection (INVPRO) is significant at 5% and its coefficient is 0.014. This differs from the results found in Table 3.
23
of equity are applied. Although the analysis with the main measure contains this problem, the coefficient of Cash ETR (CETR) is not significant in this analysis anyway. Thus, this big change in the coefficient should not be taken into account for the final conclusions.
Table 4: Regression outputs – measure based on Constant Dividend Growth Model
Notes: The table contains five OLS regression models. The coefficients for each variable are included together with the standard errors, which are inside the parentheses below the coefficients. The depended variable is the ex-ante cost of equity (R_CDGM). Model 1 represents the results of the main hypothesis’s test including the controls. Model 2 shows what the influence of interaction with investor protection is. A separate model only with the investor protection (INVPRO) is not possible as it is omitted. Model 3&4 include the other moderator, CSR, and the interaction with it. Model 5 is the full model, where each variable and interaction are tested. Recall that all explanatory variables including controls and moderators are lagged one year relative to the ex-ante cost of equity. ***, ** and *, indicate significance at 1%, 5% and 10% respectively.
Variables/Models Model 1 Model 2 Model 3 Model 4 Model 5
CETR -0.031 (0.025) 0.256 (0.164) -0.034 (0.026) 0.027 (0.056) 0.299* (0.168) Control variables LEV -0.028 (0.033) -0.028 (0.033) -0.027 (0.034) -0.031 (0.034) -0.031 (0.034) lnSIZE -0.005 (0.005) -0.005 (0.005) -0.004 (0.005) -0.004 (0.005) -0.003 (0.005) BM -0.025*** (0.010) -0.025*** (0.010) -0.027*** (0.010) -0.027*** (0.010) -0.026*** (0.010) CAPEX -0.012 (0.093) -0.022 (0.094) -0.026 (0.094) -0.023 (0.093) -0.032 (0.094) R&D -0.029 (0.154) -0.028 (0.154) -0.020 (0.155) -0.021 (0.155) -0.020 (0.155) SGA 0.064 (0.047) 0.065 (0.047) 0.062 (0.050) 0.060 (0.050) 0.061 (0.050) Moderators CSR 0.010 (0.009) 0.002 (0.011) 0.002 (0.011) Interactions INVPRO X CETR -0.038* (0.022) -0.037* (0.021) CSR X CETR -0.054 (0.044) -0.050 (0.042) Constant 0.247*** (0.093) 0.240*** (0.093) 0.221** (0.090) 0.225** (0.090) 0.218** (0.090) Year Fixed Effects
Yes Yes Yes Yes Yes
Firm Fixed Effects
Yes Yes Yes Yes Yes
Adjusted R-squared
0.326 0.327 0.324 0.325 0.326
24
Table 5: Regression outputs – measure based EPR ratio
Notes: The table contains five OLS regression models. The coefficients for each variable are included together with the standard errors, which are inside the parentheses below the coefficients. The depended variable is the ex-ante cost of equity (R_ERP). Model 1 represents the results of the main hypothesis’s test including the controls. Model 2 shows what the influence of interaction with investor protection is. A separate model only with the investor protection (INVPRO) is not possible as it is omitted. Model 3&4 include the other moderator, CSR, and the interaction with it. Model 5 is the full model, where each variable and interaction are tested. Recall that all explanatory variables including controls and moderators are lagged one year relative to the ex-ante cost of equity. ***, ** and *, indicate significance at 1%, 5% and 10% respectively.
Variables/Models Model 1 Model 2 Model 3 Model 4 Model 5
CETR 0.018*** (0.007) -0.086 (0.053) 0.018*** (0.007) 0.014 (0.015) -0.086 (0.055) Control variables LEV -0.019 (0.014) -0.019 (0.014) -0.028** (0.013) -0.028** (0.013) -0.028** (0.013) lnSIZE 0.002 (0.004) 0.002 (0.004) 0.006** (0.003) 0.006** (0.003) 0.006** (0.003) BM -0.003 (0.004) -0.003 (0.004) -0.003 (0.004) -0.003 (0.004) -0.003 (0.004) CAPEX -0.101*** (0.031) -0.100*** (0.031) -0.105*** (0.031) -0.106*** (0.031) -0.104*** (0.031) R&D 0.142** (0.057) 0.141** (0.057) 0.131** (0.058) 0.131** (0.057) 0.130** (0.058) SGA 0.037*** (0.014) 0.037*** (0.014) 0.047*** (0.014) 0.047*** (0.014) 0.047*** (0.014) Moderators
INVPRO omitted omitted
CSR -0.003** (0.001) -0.002 (0.002) -0.003 (0.002) Interactions INVPRO X CETR 0.014** (0.007) 0.014** (0.007) CSR X CETR 0.004 (0.010) 0.002 (0.010) Constant 0.073 (0.068) 0.075 (0.068) 0.007 (0.045) 0.007 (0.045) 0.009 (0.045) Year Fixed Effects
Yes Yes Yes Yes Yes
Firm Fixed Effects
Yes Yes Yes Yes Yes
Adjusted R-squared
0,602 0.603 0.614 0.614 0.615
Observations 2,742 2,742 2,693 2,693 2,693
25
V. Discussion and conclusion
The popularity of the topic related to tax avoidance has increased in the last years as many companies have engaged in different kinds of activities in order to reduce their tax burden. There have been different studies conducted with respect to tax avoidance (Dyreng et al., 2008; Desai et al., 2009). However, the relationship between tax avoidance and the cost of equity has received little attention (Goh et al. 2016; Cook et al.,2017; Hutchens and Rego, 2013). With this thesis I want to examine if there is an association between these two variables and what kind of association it is. Moreover, I use ex-ante cost of equity and the choice for that was made based on the findings of previous literature that the ex-ante measure is more appropriate than the ex-post measure. From analyzing the regression outputs from Table 3, where the main measure of ex-ante cost of equity is used, it can be seen that the coefficient of tax avoidance variable is not significant in any of the models. These results indicate that there is no statistically significant relationship between tax avoidance and the ex-ante cost of equity. This is not consistent with neither Goh et al. (2016) or Hutchens and Rego (2013) that have examined the same association. However, Goh et al. (2016) find the existence of negative relationship between both variables and Hutchens and Rego (2013) have results that show the opposite namely, positive association. This means that a possible explanation for my results of having no significant relationship is that the effects of tax avoidance in increasing and decreasing the ex-ante cost of equity can offset each other. More precisely, the riskless cash tax savings can indeed decrease the risk of investors’ return but in the same time more aggressive strategies, reputation costs, and an opportunistic behavior of managers can impact negatively on shareholder’s perceptions. This means that in the end none of these both effects is stronger than the other one as they offset each other and the result is no relationship. This is consistent with Cook et al. (2017), whose analysis indicates that a certain direction of the relationship between tax avoidance and the ex-ante cost of equity cannot be observed. Another possible explanation of having different results than previous studies is that both Goh et al. (2016) and Hutchens and Rego (2013) use different measures for tax avoidance and the ex-ante cost of equity. The way of measurement can have a big impact on the results and their interpretation.
26
meaning. In addition, both interaction effects have insignificant coefficients in the analysis based on upcoming realized returns as ex-ante cost of equity. Thus, the main findings of this thesis strongly indicate for no effect of any of the variables including both the independent and the moderating variables.
Moreover, it is crucial to interpret the result of the robustness check and if they support the main findings. The first robustness check, which is based on measuring the ex-cost of equity with the Constant Dividend Growth Model, shows that there is a significant positive relationship between tax avoidance and the ex-ante cost of equity, which would mean more tax avoidance leads to a higher ex-ante cost of equity. Additionally, this positive relationship is less pronounced when there is more investor protection. This is logical and consistent with the previous studies because when investors feel more protected, their perception regarding the risk related to tax avoidance can change (Chen et al.,2011; Hail and Leuz, 2009). The most important conclusion from this robustness check is that it does not support the results of the main analysis. However, there is one thing related to this measure of the ex-ante cost of equity that makes it not completely reliable. More precisely, if the expected dividend is equal to 0, the ex-ante cost of equity would be equal only to the long-term growth. Such an assumption is hard to be justified and accepted like an absolutely truth. Thus, concluding only on this robustness check whether the main findings can be supported is not an appropriate decision.
27
One additional thing is important to be mentioned. In every of three regression outputs the interaction effect with investor protection has a special effect. In the main analysis, so Table 3, although there is no significance, it makes the coefficient of the tax avoidance variable too different compared to the models without such an interaction. In the regression outputs related to the first robustness check, so Table 4, it has the expected effect on the main association but it makes again the coefficient of the tax avoidance variable too different. In the regression output of the second robustness check, so Table 5, it has a significant and a positive coefficient while the coefficient of tax avoidance is insignificant. Thus, the interaction between tax avoidance and the ex-ante cost of equity have very conflicting results. Nevertheless, they show that future research about this interaction when the ex-ante cost of equity is the dependent variable of interest could be useful.
In addition, the robustness check with including the size itself in the full regression, which is based on the main measure of the ex-ante cost of equity, confirm all of the main findings. Therefore, based on this and the aforementioned interpretations the following overall conclusions can be made. There is no association between tax avoidance and the ex-ante cost of equity and this is partially supported by the robustness check. The only exception is the robustness check related to the Constant Dividend Growth Model. However, the measure of the dependent variable there is not completely reliable, so its different results do not have the power to influence on the overall conclusion. Additionally, there is no evidence of CSR and investor protection having a moderating effect on the relationship between tax avoidance and the ex-ante cost of equity. In order to be concluded that a moderator can affect a certain relationship this relationship has to be significant and this is the not case in my analysis. The results of the robustness check related to the Constant Dividend Growth Model show a different case regarding the interaction with investor protection but again they do not have enough power to change the overall conclusions.
The final results of my thesis show that additional research is needed in the future in order to build stronger evidence regarding the nature of the association between tax avoidance and the ex-ante cost of equity. In addition, there is no evidence of CSR having an important impact. However, as it is mentioned above, more research about investor protection as a moderator can be useful and interesting.
28
29
Reference
Balakrishnan, K., J. Blouin, and W. Guay. (2013). Does Tax Aggressiveness Reduce Financial Reporting Transparency? Working paper, University of Pennsylvania.
Bebbington, J., Larrinaga, C., & Moneva, J. M. (2008). Corporate social reporting and reputation risk management. Accounting, Auditing & Accountability Journal, 21(3), 337-361.
Bushman, R. M., Piotroski, J. D., & Smith, A. J. (2004). What Determines Corporate Transparency? Journal of Accounting Research, 42 (2), 207–52.
Campbell, J. (1991). A variance decomposition for stock returns. Economic Journal, 101, 157–179. Chava, S., & Purnanandam, A. (2010). Is default risk negatively related to stock returns? Review of Financial Studies, 23, 2523–2559.
Chen, S., Chen, X., Cheng, Q., & Shevlin, T. (2010). Are family firms more tax aggressive than non-family firms? Journal of Financial Economics, 95(1), 41-61.
Chen, K. C., Chen, Z., & Wei, K. J. (2009). Legal protection of investors, corporate governance, and the cost of equity capital. Journal of Corporate Finance, 15(3), 273-289.
Chen K., Chen Z., and Wei K.C. (2011). Agency Costs of Free Cash Flow and the Effect of Shareholder Rights on the Implied Cost of Equity Capital. The Journal of Financial and Quantitative Analysis, 46 (1), 171-207.
Cook, K. A., Moser, W. J., & Omer, T. C. (2017). Tax avoidance and ex ante cost of capital. Journal of Business Finance & Accounting, 44(7-8), 1109-1136.
Desai, M., and D. Dharmapala. (2006). Corporate tax avoidance and high-powered incentives. Journal of Financial Economics, 79, 145–179.
Desai, M., Dyck, A., & Zingales, L. (2007). Theft and taxes. Journal of Financial Economics, 84(3), 591-623.
Desai, M. A., & Dharmapala, D. (2009). Corporate Tax Avoidance and Firm Value. Review of Economics and Statistics, 91(3), 537-546.
Dyreng, S. D., Hanlon, M., & Maydew, E. L. (2008). Long‐Run Corporate Tax Avoidance. The Accounting Review, 83(1), 61-82.
Easton, P. D. (2004). PE Ratios, PEG Ratios, and Estimating the Implied Expected Rate of Return on Equity Capital. The Accounting Review, 79(1), 73-95.
El Ghoul, S., Guedhami, O., & Kim, Y. (2017). Country-level institutions, firm value, and the role of corporate social responsibility initiatives. Journal of International Business Studies, 48(3), 360-385. Elton, E. (1999). Presidential address: Expected return, realized return, and asset pricing tests. Journal of Finance, 54, 1199–1220.
Fama, E., & French, K. (1997). Industry costs of equity. Journal of Financial Economics, 43, 153–193. Goh, B. W., Lee, J., Lim, C. Y., & Shevlin, T. (2016). The Effect of Corporate Tax Avoidance on the Cost of Equity. The Accounting Review, 91(6), 1647-1670.
Graham, J. R., Hanlon, M., Shevlin, T., & Shroff, N. (2014). Incentives for Tax Planning and Avoidance: Evidence from the Field. The Accounting Review, 89(3), 991-1023.
30
Hail L, Leuz C. (2009) Cost of capital effects and changes in growth expectations around U.S. cross-listings. Journal of Financial Economics, 93(3),428–454.
Hansen, B., Mihail K. M., and Wintoki, M. (2015). Investor Protection and the Role of Firm-Level Financial Transparency in Attracting Foreign Investment. Financial Review ,50(3), 393-434.
Harris, R. S., Marston, F. C., Mishra, D. R., & Obrien, T. J. (2003). Ex Ante Cost of Equity Estimates of S&P 500 Firms: The Choice between Global and Domestic CAPM. Financial Management, 32(3), 51.
Hope, O.-K., Kang, T., Thomas, W. B., & Yoo, Y. K. (2009). Impact of excess auditor remuneration on the cost of equity capital around the world. Journal of Accounting, Auditing and Finance, 24, 177– 210.
Hutchens, M., & Rego, S. (2013). Tax Risk and the Cost of Equity Capital. Working capital.
James Gwartney, Robert Lawson, and Joshua Hall (2017). Economic Freedom Dataset, published in Economic Freedom of the World: 2017 Annual Report. Fraser Institute.
<www.fraserinstitute.org/economic-freedom/dataset>.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.
John, K., Litov, L., & Yeung, B. (2008). Corporate Governance and Risk-Taking. The Journal of Finance, 63(4), 1679-1728.
Kim, Y., Park, M., & Wier, B. (2012). Is earnings quality associated with corporate social responsibility? The Accounting Review 87 (3), 761–796.
Kiesewetter, D., & Manthey, J. (2017). Tax avoidance, value creation and CSR – a European
perspective. Corporate Governance: The International Journal of Business in Society, 17(5), 803-821. Lambert, R., Leuz, C., & Verrecchia, R. E. (2007). Accounting Information, Disclosure, and the Cost of Capital. Journal of Accounting Research, 45(2), 385-420.
Lopatta, K., Buchholz, F., & Kaspereit, T. (2016). Asymmetric information and corporate social responsibility. Business & Society, 55(3), 458-488.
Luo, X., Wang, H., Raithel, S., & Zheng, Q. (2015). Corporate social performance, analyst stock recommendations, and firm future returns. Strategic Management Journal, 36(1), 123-136. Mills, L. F. (1998). Book-Tax Differences and Internal Revenue Service Adjustments. Journal of Accounting Research, 36(2).
Mills, L., M. Erickson, and E. Maydew. (1998). Investments in tax planning. Journal of the American Tax Association, 20 (1), 1–20.
Mishra, D. R., & Obrien, T. J. (2005). Risk and ex ante cost of equity estimates of emerging market firms. Emerging Markets Review, 6(2), 107-120.
Sikka, P. (2010). Smoke and mirrors: Corporate social responsibility and tax avoidance. Accounting Forum, 34: 153–168.
Sufi, A. (2007). Information asymmetry and financing arrangements: Evidence from syndicated loans. The Journal of Finance, 62(2), 629–668.
31
Appendix
Appendix 1: Correlation matrix including interaction effects
Notes: The table reports the correlation between the variables of the ex-ante cost of equity, the two moderators, and the two interaction effects. All significant correlations are marked with the number of stars, which reflects their level of significance. ***, **, * Denote significance at the 1 percent, 5 percent, and 10 percent, respectively.
(1) (2) (3) (4) (5) (6) 1 R 1.00 2 CETR 0.02 1.00 3 INVPRO -0.03*** -0.02* 1.00 4 INVPRO*CETR 0.02 0.96*** -0.24*** 1.00 5 CSR -0.04*** -0.08*** -0.01 -0.10*** 1.00 6 CSR*CETR 0.04* 0.78*** -0.06*** 0.76*** -0.57*** 1.0
Appendix 2: Regression outputs including size itself
Notes: The table contains only the full model, where each variable and interaction are tested. Recall that all explanatory variables including controls and moderators are lagged one year relative to the ex-ante cost of equity. ***, ** and *, indicate significance at 1%, 5% and 10% respectively.
Variables/Models Full model
CETR -0.428 (0.534) Control variables LEV -0.136 (0.100) SIZE -1.31e-11 (8.97e-12) lnSIZE 0.002 (0.017) BM -0.008 (0.031) CAPEX -0.287 (0.233) R&D 0.032 (0.395) SGA 0.021 (0.104) Moderators INVPRO Omitted CSR 0.010 (0.037) Interactions INVPRO X CETR 0.065 (0.069) CSR X CETR 0.019 (0.140) Constant 0.340 (0.289) Year Fixed Effects Yes Firm Fixed Effects Yes Adjusted R-squared 0.224
