Tax Avoidance and Financial Constraints: A Simultaneous Equations Analysis

Transcription

1 Tax Avoidance and Financial Constraints: A Simultaneous Equations Analysis Onur Bayar a, Fariz Huseynov b a University of Texas at San Antonio, College of Business, One UTSA Circle, San Antonio, TX onur.bayar@utsa.edu. Tel: b North Dakota State University, College of Business, P.O. Box 6050, Dept. 2410, R. H. Barry Hall 202, Fargo, ND fariz.huseynov@ndsu.edu. Tel: August 12, 2015

2 Tax Avoidance and Financial Constraints: A Simultaneous Equations Analysis Abstract We examine the simultaneous relationship between tax avoidance and financial constraints by controlling for its endogeneity. We find that tax avoidance leads to greater financial constraints and a higher probability of financial distress, while firms with more binding financial constraints are pressured to avoid taxes to a greater extent. Tax avoidance leads to greater financial constraints especially in poor governance firms and in low capital intensity firms with limited future capital investment requirements. We also find that financially constrained firms tend to avoid taxes to a greater extent especially in firms with low levels of earnings management. Keywords: tax avoidance, financial constraints, corporate governance, earnings management, simultaneous equations analysis JEL classification: G30; G33; H25; H26 1. Introduction Corporations can use aggressive tax avoidance strategies to reduce their effective corporate income tax rates, which can help them realize substantial tax savings to increase shareholder value. 1 When investors recognize tax avoidance as a value-enhancing activity, managers acting in the best interests of their shareholders are encouraged to channel greater efforts in tax planning. Further, when firms face binding financial constraints, the marginal value of tax savings can substantially increase, since tax avoidance can be used as an effective means of relaxing the firm s financial constraints. 2 However, in an agency framework, tax avoidance may also be driven by the opportunism of self-interested managers who can divert some of the firm s tax resources for their own benefits, 1 Hanlon and Heitzman (2010) reviews tax avoidance literature. Graham and Tucker (2006), Hanlon and Slemrod (2009) and Armstrong et al. (2011) analyze the implications of tax avoidance on firm value. 2 Edwards et al. (2013) and Graham et al. (2014) find that financially constrained firms are more likely to manage their taxes to obtain cash savings. 1

3 especially in firms with poor governance mechanisms (Crocker and Slemrod, 2005; Desai and Dharmapala, 2006, 2009). In this case, a more active practice of tax avoidance can lead to a greater cost of external financing and a loss in firm value, since outside investors may associate higher tax avoidance with the existence of managerial incentive problems within the firm. 3 Hence, firms facing greater external financing constraints may actually be discouraged to avoid taxes in order not to worsen their existing precarious financial conditions by issuing another negative signal to the financial markets. Although the existing literature analyzes the determinants and consequences of tax avoidance and those of financial constraints separately, no study, to the best of our knowledge, has so far examined the relationship between tax avoidance and financial constraints by controlling for the endogeneity of this relationship. The objective of our paper is to fill this gap in the literature. Specifically, we examine whether greater tax avoidance leads to greater financial constraints and whether firms avoid more taxes when they are financially constrained by using a two-stage least squares (2SLS) estimation method to control for endogeneity. We also explore how the quality of corporate governance along with other firm characteristics, such as earnings management and capital intensity, affects the relationship between tax avoidance and financial constraints. We first examine the impact of a firm s tax avoidance behavior on its financial constraints. To relax financial constraints, the potential benefits of tax savings should outweigh the costs of tax avoidance. Direct costs of tax planning are administrative costs, litigation expenses, and penalties imposed by tax authorities (Wilson, 2009), whereas indirect costs of tax planning include political costs, potential reputation losses, a greater cost of debt capital, and a higher stock price crash risk (Hanlon and Slemrod 2009; Kim et al., 2011; Hasan et al., 2013; Graham et al., 2014). 3 Several studies argue that tax avoidance leads to higher bank loan spreads (Hasan et al., 2014), negative stock market reaction (Hanlon and Slemrod, 2009), and stock price crash risk (Kim et al., 2011). 2

4 In particular, tax avoidance may concur with opaque information environments where the firm s cost of raising external financing is higher due to managerial incentive problems such as the insiders diversion of firm resources for their own private benefits and earnings management (Lev and Nissim, 2004; Hanlon, 2005; Desai and Dharmapala, 2006 and 2009; Hanlon and Heitzman, 2010). According to the information hypothesis, if the costs of tax avoidance outweigh the potential benefits of tax savings that accrue to shareholders, higher tax avoidance can actually increase a firm s cost of external financing and therefore intensify the degree of its financial constraints. On the other hand, prior studies (Desai and Dharmapala, 2006; 2009; Minnick and Noga, 2010) find that tax avoidance benefits shareholders when the interests of managers and shareholders are aligned. Strong governance mechanisms with properly structured managerial incentive schemes can ensure the positive value effect of tax avoidance, especially when shareholders believe that such activities are value-enhancing (Graham and Tucker, 2006). The increase in firm value due to tax savings may be associated with a lower cost of external financing and a lower likelihood of financial distress for the firm. Thus, according to the value hypothesis, tax avoidance will reduce the extent of a firm s financial constraints and its risk of financial distress. The direction of the relationship between tax avoidance and financial constraints can be reverse. A financially constrained firm may be more likely to avoid taxes to increase its internally generated funds and thereby increase its financial flexibility given its external financing constraints (e.g. Edwards et al., 2013 and Graham et al., 2014). Financial flexibility in the form of higher cash holdings is more valuable to constrained firms allowing them to implement their value-increasing investments (Denis and Sibilkov, 2010). Thus, one can expect more financially constrained firms 3

5 to engage in a relatively higher extent of tax avoidance (the pressure hypothesis) until the marginal cost of tax avoidance exceeds the marginal benefit of financial flexibility gained through additional tax savings. Further, this effect could be stronger in more capital-intensive firms, which have greater financing requirements to fund their positive-npv growth opportunities. Alternatively, tighter financial constraints may discourage a firm from adopting an aggressive tax avoidance policy, especially when outside investors consider tax avoidance as complementary to managerial rent seeking and the diversion of resources by firm insiders. The severity of its financial constraints may impose a monitoring effect on a firm to limit its tax avoidance, as the deterioration in the market s perception about the firm s agency problems can lead to an even higher cost of external capital. Thus, according to the monitoring hypothesis, financially constrained firms will restrain their tax avoidance. We conduct a two-staged least-squares (2SLS) analysis to examine the abovementioned effects between tax avoidance and financial constraints. 4 We measure the degree of a firm s financial constraints by using two proxies: the Altman Z-Score (Altman, 1968) and the WUscore (Whited and Wu, 2006). Following Dyreng et al. (2008), we define tax avoidance as the reduction in the firm s taxes relative to its pretax accounting income. We use the cash effective tax rate, Cash ETR and income effective tax rate, GAAP ETR as our measures of tax avoidance. 5 We calculate both the short-run ETRs that allow us to observe the firms short-term adjustments of their tax obligations due to their financial needs and the long-run ETRs that more closely track the firm s 4 For the econometric method used in our study, see also O Brien and Bhushan (1990), Alford and Berger (1999), and Hong et al. (2014). We also conduct Hausman tests for endogeneity and the results reject the null hypothesis that there is no simultaneity. 5 Dyreng et al. (2008) suggest that Cash ETR directly measures a firm s cash tax burden and is the preferred measure of tax avoidance. Cash ETR also measures the impact of financial constraints on the firms incentives to save cash through tax planning and tax sheltering. However, recent studies (Armstrong et al., 2011; Graham et al., 2013) show that GAAP ETR is also essential to tax management and other corporate decisions. Specifically, in their survey of nearly 600 corporate tax executives Graham et al. (2013) find that 84% of care at least as much about the GAAP ETR as they do about cash taxes paid. 4

6 tax avoidance behavior over multiple years to avoid spikes in year-to-year volatility of our tax burden measures. We assume that a lower (higher) effective tax rate indicates higher (lower) tax avoidance. We find that firms that avoid taxes more (less) are likely to have greater (lower) financial constraints. Specifically, each of our proxies for tax avoidance (i.e. ETRs) has a significantly positive effect on the firm s Altman Zscore (the higher the firm s Zscore, the lower its bankruptcy risk), and a negative effect on the firm-specific index of financial constraints, WUscore (the lower the firm s WUscore, the less binding are its financial constraints). These results are strongly consistent with the information hypothesis which implies that higher tax avoidance is associated with firms in which the extent of managerial moral hazard (e.g., diversion, earnings management) and the degree of information asymmetry between outside investors and firm insiders are greater. We also find that in the presence of more stringent financial constraints, firms are likely to avoid taxes more to obtain a greater degree of financial flexibility. This result is consistent with the pressure hypothesis according to which financially constrained firms experience a greater pressure to reduce their tax cash outflows in order to bypass their external financing constraints by increasing their financial slack, allowing them to fund their value-increasing investments. We also examine the impact of firm characteristics such as corporate governance quality, capital intensity, and earnings management on the relationship between tax avoidance and financial constraints. First, we find that higher tax avoidance leads to an increase in financial constraints in both strong governance and poor governance firms. However, this effect is stronger in firms with poorer corporate governance in which the potential for managerial diversion and the risk of adverse selection for outside investors are greater and the value-efficiency of tax avoidance is lower (Crocker and Slemrod, 2005; Desai and Dharmapala, 2006; 2009; Minnick and Noga, 5

7 2010). This is consistent with the notion that opaque information environments facing the investors of poorly governed firms are likely to deteriorate further as a result of greater tax avoidance. Second, we find that tax avoidance leads to a greater degree of financial constraints in both high- and low-capital intensity firms. However, this effect is stronger for firms with lower capital investment needs. Since firms with higher capital intensity need to save cash for future investment requirements, the negative effects of information problems associated with tax avoidance may be partially offset by the value-increasing effects of achieving greater financial slack through tax avoidance. On the other hand, in low-capital intensity firms, investors may infer that the extent of information problems associated with tax avoidance is more severe, since the managers of such firms are more likely to engage in tax avoidance for their self-interests rather than to obtain financial flexibility for future capital investment needs. Finally, we examine the impact of earnings management since prior literature suggests that tax avoidance can be utilized to manipulate earnings and meet earnings forecasts, hence weakening the quality of the information environment for the firm s investors (Hanlon and Heitzman, 2010). We find that the negative impact of tax avoidance on a firm s financial constraints is not affected by the level of its earnings management. This suggests that tax avoidance independently acts as a strong signal of the deterioration in the information environment the firm provides to its outside investors. We also investigate the role of firm characteristics on the impact of financial constraints on tax avoidance. More financially constrained firms are likely to avoid taxes to a greater extent especially in the subsample of firms with low earnings management in which the information environment between the firm and its investors is more transparent. This is consistent with several studies (Wang, 2010; Balakrishnan, Blouin, and Guay, 2011) finding a trade-off behavior between 6

8 tax avoidance and financial reporting quality such that firms with greater financial constraints may engage in tax avoidance in exchange for providing more transparent financial statements. Our results for the role of governance and capital intensity are somewhat mixed. When we use WUscore as our proxy of financial constraints, we find that in both subsamples of corporate governance quality, firms with financial constraints are likely to avoid taxes more (consistent with the pressure hypothesis) and the magnitude of this effect is greater in firms with stronger governance. This is consistent with the view that stronger governance mechanisms reduce the extent of agency problems associated with tax avoidance in financially constrained firms. In contrast, when we examine the impact of Zscore on Cash ETR, we find that, consistent with the monitoring hypothesis, only in firms with weaker corporate governance, financially constrained firms tend to avoid taxes less, since increasing tax avoidance may further exacerbate financial constraints or increase the risk of financial distress. We also find that greater financial distress risk (lower Zscore) leads to lower cash tax avoidance in less capital intensive firms (consistent with the monitoring hypothesis) and to greater income tax avoidance in both capital intensity groups (consistent with the pressure hypothesis). When we use WUscore as a proxy for firm-specific financial constraints, our findings indicate that more financially constrained firms avoid taxes (Cash ETRs) to a greater extent, which is consistent with the pressure hypothesis, and this impact is greater in the subsample of high capital intensity firms. 6 Our study sheds further light on the under-sheltering puzzle (Weisbach, 2002) that questions why all firms do not exhaust tax avoidance opportunities. Our findings suggest that, in the presence of managerial incentive problems and the information asymmetry between insiders 6 We also examined the interaction of corporate governance and capital intensity (untabulated). We find that greater tax avoidance results in greater financial constraints, especially in poor governance firms with lower capital investment needs. Managers may privately benefit from tax avoidance at the expense of the shareholders in poorly governed firms with lower growth opportunities (lack of capital investment requirements). 7

9 and outsiders, the potential increase in financial constraints may discourage firms from exploiting their tax avoidance strategies to the fullest extent. We also extend our understanding of the accounting factors and their implications on financial constraints, an area that is relatively unexplored (Hanlon and Heitzman, 2010). Our paper also contributes to the corporate finance literature examining the effect of costly external financing on a firm s financial and investment policies. Capital market imperfections substantially increase the cost of external financing relative to internally generated funds (see, e.g. Myers and Majluf, 1984). Faulkender and Wang (2006), Pinkowitz and Williamson (2006), and Denis and Sibilkov (2010) report evidence consistent with the view that cash holdings are more valuable for constrained firms than for unconstrained firms. One of the ways in which firms can mitigate the adverse consequences of external financing is to adjust their tax policies to obtain corporate tax savings. Our findings suggest that the costs associated with tax management activities may outweigh the financial benefits of tax savings and exacerbate financial constraints, especially when the quality of corporate governance is low. The remainder of this paper is organized as follows: Section 2 develops our main hypotheses. Section 3 discusses data sources, the sample selection, and the research methodology. In Section 4 we present and discuss our empirical results, and we conduct some robustness tests. Section 5 concludes. 2. Hypothesis Development We first build a set of hypotheses predicting the impact of tax avoidance activity on a firm s financial constraints. This impact depends on the direct and indirect costs of tax planning activities and on the benefits achieved as a result of avoiding taxes (Hanlon and Slemrod, 2009; Wilson, 8

10 2009; Graham et al., 2014; Hasan et al., 2013). In an extensive survey of managerial tax planning and avoidance, Graham et al. (2014) find that 69 percent of responding executives rate reputation as the second important factor explaining why firms do not adopt a potential tax planning strategy. Firms avoiding taxes more are likely to operate in opaque information environments where the firms cost of raising external capital is higher due to the presence of moral hazard (e.g., diversion, earnings management) and adverse selection problems (Lev and Nissim, 2004; Hanlon, 2005; Desai and Dharmapala, 2006 and 2009; Hanlon and Heitzman, 2010). Desai and Dharmapala (2006) show that even though tax management strategies can be deployed to maximize shareholder value, these strategies can also be complementary to managerial rent diversion. Francis et al. (2014) find a significantly negative relation between managerial ability and tax avoidance. Since tax planning activities can be motivated by the managers propensity to seek rents and/or hoard bad news, tax avoidance can increase the equity market s uncertainty about the valuations of firms with more aggressive tax sheltering, exposing them to higher price crash risks (Kim et al., 2011). This, in turn, can have adverse consequences on the firm s financial condition and its cost of obtaining external funds in public markets. Therefore, our information hypothesis posits that a greater level of tax avoidance leads to a deterioration of a firm s financial constraints and an increase in its risk of financial distress. Further, we predict that this negative effect is stronger in firms with poorer corporate governance quality. Tax avoidance may also create value in the presence of properly structured incentive mechanisms that constrain the rent-seeking behavior of managers. Graham and Tucker (2006) propose that tax planning is a value-enhancing activity and find that shareholders hold that belief. 7 7 Mills et al. (1998) find that an additional $1 investment in tax planning results in a $4 reduction in tax liabilities. Scholes et al. (2009) conclude that potential tax savings from aggressive avoidance strategies could be economically large. Dyreng et al. (2013) find that the use of Delaware subsidies as domestic tax havens, on average, increases net income by %. 9

11 Desai and Dharmapala (2009) show a positive relationship between firm value and tax avoidance only for a subsample of well-governed public firms. If the marginal cost of tax avoidance is sufficiently small, additional tax savings realized through tax avoidance may actually alleviate and even eliminate the firm s financial constraints and reduce the firm s risk of financial distress. Thus, the value hypothesis posits that tax avoidance behavior decreases a firm s financial constraints and its risk of financial distress. Next we build our hypotheses for the impact of a firm s financial constraints on its tax avoidance behavior. Prior finance literature has extensively investigated the effects of financial constraints on firm behavior. 8 We argue that as external financing becomes more expensive for financially constrained firms, such firms are likely to search for strategies that will help them relax or alleviate these constraints internally. Denis and Sibilkov (2010) provide empirical evidence consistent with the view that cash holdings are more valuable to constrained firms, allowing them to implement their value-increasing investments. One way to increase a firm s cash holdings and thereby relax its financial constraints is to realize tax savings by engaging in tax avoidance and tax planning (e.g. Edwards et al., 2013 and Graham et al., 2014). If the cost of implementing tax avoidance strategies is below a certain threshold, we predict that financially constrained firms will increase their tax avoidance as access to external funds becomes more expensive and/or limited to fund the firm s investment needs. Therefore, according to the pressure hypothesis, we posit that more financially constrained firms will engage in tax avoidance to a greater extent, and that this effect is going to be stronger in more capital-intensive firms with greater investment opportunity sets. 8 See, for example, Fazzari, Hubbard, and Petersen (1988), Whited (1992), Froot, Scharfstein, and Stein (1993), Kaplan and Zingales (1997), Gomes (2001), Almeida, Campello, and Weisbach (2004), Whited and Wu (2006), Almeida and Campello (2007), Hennessy and Whited (2007), Duchin, Ozbas, and Sensoy (2010), Faulkender and Wang (2006), Denis and Sibilkov (2010), and, Giroud and Mueller (2013). 10

12 Alternatively, the marginal cost of tax avoidance can be significantly greater than the shortrun benefit of tax savings in relaxing the firm s financial constraints. For instance, if a firm is already operating in an opaque information environment characterized by poor corporate governance and agency problems, tax avoidance and managerial rent diversion can be complementary to each other as first suggested by Desai and Dharmapala (2006) and Kim et al. (2011). Thus, increasing tax avoidance during financial distress can actually increase the firm s cost of external financing, thereby worsening the firm s existing financial condition and reducing the firm s equity market valuation. Therefore, our monitoring hypothesis posits that firms with more binding financial constraints will resort to tax avoidance to a lesser extent. In other words, the severity of financial constraints will have a negative impact on tax avoidance. 3. Data, Variables, and Research Design The data used in this study come from two publicly available data sources. We obtain the financial statement data from Compustat and the Governance index from Andrew Metrick s webpage 9. The index is calculated as described in Gompers et al. (2003) based on the listings produced by the Investor Responsibility Research Center of the takeover defenses adopted by a large sample of firms. We delete firm-year observations with missing values for total assets and with negative or zero values for pretax income because they are treated differently for tax purposes. We also delete firms from the financial and utility industries. Our final dataset includes over 34,000 firm-year observations from Financial Constraints and Tax Avoidance Measures

13 When testing the simultaneous effects between tax avoidance and financial constraints, we use two measures of financial constraints and four measures of tax avoidance suggested by the prior literature. The first measure of financial constraints is the firm s Altman (1968) Zscore in year t. This measure is widely used by prior literature for assessing the financial distress of corporations (see, e.g., Tykvová and Borell, 2013; Altman, 2013; and Avramov, Chordia, Jostova, and Philipov, 2013). Our rationale for using Zscore is that financial distress will be highly correlated with the extent to which a firm faces financial constraints. Lower values of Zscore indicate higher levels of financial distress and therefore, represent more stringent financial constraints. Our second measure of financial constraints is the WUscore, a firm-year specific financial constraint index developed by Whited and Wu (2006) and widely used by previous studies (e.g. Duchin, 2010; Li, 2011; Panousi and Papanikolaou, 2012). Note that different from Zscore, WUscore is coded such that higher values represent tighter financial constraints. Following Dyreng, Hanlon, and Maydew, (2008) we define tax avoidance broadly to encompass anything that reduces the firm s taxes relative to its pretax accounting income. Our first two measures capture the firm s cash effective tax rate, Cash ETR. Since Cash ETR is the most direct measure of a firm s cash tax burden, using this measure allows us to capture the impact of financial constraints on a firm s incentive to save cash by using tax planning and tax shelters. We use both short-run and long-run Cash ETR. The short-run Cash ETR is defined as the ratio of the firm s annual cash taxes paid to its pre-tax accounting income adjusted for special items in year. The long-run Cash ETR is estimated as the centered moving sum of cash paid for income taxes over five years scaled by the sum of pre-tax income (net of special items) over the same period (see Dyreng, Hanlon, and Maydew, 2008). This generates an effective cash tax rate that more closely tracks the firm s tax costs over the long run, and it avoids year-to-year volatility in annual 12

14 effective tax rates. We also measure tax avoidance using the firm s effective tax rate as defined under GAAP (hereafter, GAAP ETR). Although, a firm is more likely to achieve greater tax savings through lowering its Cash ETR, recent studies (Armstrong, Blouin, Larcker, 2011; Graham, Hanlon, Shevlin, and Shroff, 2013) show that GAAP ETR is essential to tax management and other corporate decisions. Specifically, Graham et al. (2013) survey responses from nearly 600 corporate tax executives and find that 84% of care at least as much about the GAAP ETR as they do about cash taxes paid. Similar to Cash ETR, we calculate both short-run and long-run GAAP ETRs. The short-run GAAP ETR is equal to the ratio of the firm s annual total GAAP tax expense (current tax expense plus deferred tax expense) to the firm s pre-tax accounting income adjusted for special items in year t. The long-run GAAP ETR is estimated as the centered moving sum of annual total GAAP tax expenses over five years scaled by the sum of pre-tax income (net of special items) over the same period. We assume that firms that avoid taxes more (less) will have lower (higher) effective tax rates Determinants of Financial Constraints The key determinants of financial constraints in our study are four measures of tax avoidance, namely the long-run and the short-run Cash ETR and GAAP ETR. According to the information hypothesis, tax avoidance may lead to a tightening financial constraints and the ETRs should have a positive (negative) impact on Zscore (WUscore). On the other hand, according to the value hypothesis, tax avoidance leads to less stringent financial constraints and the ETRs have a negative (positive) impact on Zscore (WUscore). Since there exist two competing hypotheses, the predicted impact of tax avoidance on financial constraints is ambiguous and the question needs to be resolved empirically. 13

15 We control for some firm-specific characteristics, such as firm size, ROA, financial leverage, firm performance, capital expenditures, and dividend policy that are associated with tax avoidance but also affect firms external financial constraints. Since smaller firms may be more sensitive to capital market imperfections (Denis and Sibilkov, 2010), we control for firm size measured as the natural logarithm of total assets, Log of Assets. Since more profitable firms and firms with better equity market performance are less likely to be financially constrained, we also control for firm profitability measured by Return of Assets (the ratio of firm s pretax income to total assets) and the firm s Tobin s Q ratio. The Tobin s Q ratio is calculated as the book value of assets less the book value of equity plus the market value of the equity, divided by the book value of the assets. Since firms with higher financial leverage are more likely to have exhausted their debt capacity and therefore face tighter financial constraints, we control for Leverage, the ratio of long-term debt to total assets. We also control for the sales growth rate of the firm, Sales Growth, and its industry, Industry Sales Growth, as some of the other factors that determine the firm s level of financial constraints (Whited and Wu, 2006). Prior literature (e.g. Faulkender and Wang, 2006; Denis and Sibilkov, 2010) finds a positive association between a firm s cash holdings and its financial constraints. Financially constrained firms may hold more cash to decrease the need for costly external financing, especially when the future cash flows of the firm are expected to be insufficient to satisfy the firm s future capital investment requirements. We control for this effect by using cash holdings scaled by total assets, CashTA, as another control variable. We expect that firms with greater growth opportunities and higher capital expenditures will be more financially constrained because of their greater needs for external funding (Korajczyk and Levy, 2003). We use three variables, Price-to-Book ratio, calculated as the ratio of market value 14

16 of equity to the book value of equity, Capital Expenditures, defined as the ratio of capital expenditures to total assets, and Net Working Capital calculated as working capital, less cash, divided by total assets, to control for the effects of firms investments needs. We also control for the possibility of investment inefficiency in a firm due to managerial moral hazard. Firms with higher investment inefficiency are more likely to decrease their financial flexibility and therefore, they are likely to face greater financial constraints (Subramaniam et al., 2011). Following Subramaniam, Tang, Yue, and Zhou (2011) we define Inefficiency as the interaction between capital expenditures scaled by total assets and a dummy variable that is one, if a firm s Tobin s Q is lower than the industry s median, and zero otherwise. An opaque information environment and the lack of transparency in financial reporting generally increase the cost of external financing (Francis, Khurana, and Pereira, 2005). Higher accounting quality also enhances the firm s investment efficiency by reducing the information asymmetry between firm insiders (management) and the firm s outside investors (Biddle and Hillary, 2006). The negative impact of disclosure level on the cost of capital is more likely to be driven by higher earnings quality. Francis, Nanda and Olsson (2008) find that good earnings quality has a greater impact on the costs of external financing than other forms of financial disclosure. Following prior studies, such as Hong, Huseynov, and Zhang (2014), we estimate the absolute value of discretionary accruals, AbsDAccruals, using the performance controlled crosssectional modified Jones (1991) model to control for aggressive financial reporting practices (see, e.g., Kothari et al. (2005)). Larger values of AbsDAccruals signal higher earnings management and lower earnings quality. We also use Bonddummy to control for the effect of firm opacity and information asymmetry based on prior studies finding that firms with debt ratings are less financially constrained (Denis and Sibilkov, 2011). Following an approach similar to Denis and 15

17 Sibilkov (2010), we set BondDummy equal to one if the firm has its long-term debt rated by Standard & Poor s Long-term Senior Debt Rating (available on Compustat) and its debt is not in default (rating of D or SD ), and zero otherwise. Several studies relate financial constraints to firms dividend policy and argue that unconstrained firms are more likely to have higher payout ratios than constrained firms (Fazzari, Hubbard, and Petersen, 1988; Denis and Sibilkov, 2011). Our dummy variable, Divdummy, controls for the negative association between a firm s dividend payout rate and its financial constraints. Note also that, by construction, our measures of financial constraints, Zscore and WUscore, are calculated using several firm-specific variables. Hence, for each of these measures we have to exclude their components from control variables used in the regression models. Therefore, we exclude net working capital from our models for Zscore, and leverage, firm size, and the dummy variable for dividend payout from our control variables for WUscore Determinants of Tax Avoidance In this study, we consider two measures of financial constraints as the key determinants of a firm s tax avoidance behavior. Lower values of Zscore and higher values of WUscore indicate that a firm is more financially constrained. According to the monitoring hypothesis, the presence of tighter financial constraints imposes some discipline over the firm s managers and discourages tax avoidance. If this is true, the impact of Zscore (WUscore) on ETRs should be negative (positive). On the other hand, the pressure hypothesis predicts that the presence of tighter financial constraints may encourage managers to avoid taxes more in order to achieve higher financial flexibility. This suggests that the impact of Zscore (WUscore) on ETR should be positive (negative). 16

18 In addition to our key explanatory variables, we include several control variables in our regression model. We control for firm size, Log of Assets, since we expect that it is negatively related to effective tax rates. It is well known that larger firms are more sophisticated and can better implement tax planning strategies with the help of tax advisors (Mills, Erickson, and Maydew, 1998; Hanlon, Mills, and Slemrod, 2007). The firm s return on assets, ROA, is another control variable since more profitable firms exhibit higher effective tax rates as firms with higher ROAs are subject to higher marginal tax rates and have less of their taxes offset by fixed tax shields (see, e.g., Wilkie, 1988 and Shevlin and Porter, 1992). Prior literature suggests that capital structure decisions and the level of financial leverage have important tax implications, since interest expenses on debt provide an important tax shield. 10 Graham and Harvey (2001) find that tax benefits of interest deductibility is one of the most important factors that affect how CFOs choose the appropriate amount of debt for their firms. Newberry and Dhaliwal (2001) argue that multinationals can place their debt in high-tax locations and thus reduce their effective tax rates. Firms can also structure off-balance-sheet financing to maximize interest deductions without decreasing book income (Mills and Newberry, 2004) or they can structure their debt to use foreign tax credits (Newberry, 1998). Collectively, these studies suggest that debt is negatively associated with tax rates. We use Leverage to control for the impact of financial leverage on effective tax rates. Since firms with higher capital expenditures can obtain greater tax incentives for new investments, capital-intensive firms exhibit lower tax burdens (Gupta and Newberry, 1997; Huseynov and Klamm, 2012), higher book-tax differences (Mills and Newberry, 2001; Wilson, 2009; Lisowsky, 2010), and higher IRS deficiencies (Rice, 1992; Mills, 1998). Therefore, we also 10 Graham (1996) reviews this literature. 17

19 include Capital Expenditures, and Net Working Capital, to control for this effect on effective tax rates. In their review of tax avoidance literature, Hanlon and Heitzman (2010) discuss how some tax avoidance practices may be related to earnings management. If firms that employ earnings management practices are more tax-aggressive as suggested by Frank, Lynch, and Rego (2009), then we would expect a positive association between earnings management and tax avoidance. Thus, we use AbsDaccruals to control for the effect of earnings management on tax avoidance. Prior studies (for example, Dhaliwal et al., 2011) find that tax avoidance may be negatively related to a firm s cash holdings if its managers are likely to divert some of the firm s tax savings. We control for this effect using Cash Holdings (cash holdings scaled by total assets). In addition, we use a measure of investment inefficiency that arises due managerial moral hazard. Managers of firms with inefficient investment behavior may also utilize tax avoidance for diversion or rentseeking purposes. We use the variable Inefficiency to control for this effect. We also control for other firm specific variables such as dividend payout and the firm s future growth opportunities. Several studies (for example, Desai and Dharmapala, 2008; Minnick and Noga, 2010; Huseynov and Klamm, 2012) argue that a firm s dividend payout policy may have an impact on its income tax structure. To control for the potential relationship between dividend policy and tax avoidance we use a dummy variable, DivDummy, that takes the value of one if the firm pays dividends, zero otherwise. Dyreng et al. (2008) find that firms with higher growth opportunities have higher effective tax rates. We use the market to book ratio, Price-to-book, to control for the firm s growth potential. Finally, prior studies suggest that the opaqueness of a firm s financial reporting environment may have an impact on its tax avoidance behavior. While Kerr (2012) finds that international firms 18

20 with more opaque information environments exhibit higher levels of tax avoidance, Wang (2010) and Balakrishnan, Blouin, and Guay (2011) find that there is a trade-off between financial transparency and aggressive tax planning, and that these two factors may substitute each other. We use BondDummy as a proxy for the transparency of the firm s information environment Simultaneous Equations and Instruments Our hypotheses suggest that there exists a simultaneously (or jointly) determined relationship between tax avoidance and financial constraints. Therefore, if we use an ordinary least-squares regression (OLS) model to estimate the relationship between tax avoidance and financial constraints without controlling for its endogeneity, the error terms from the regressions will be correlated and the coefficients of endogenous independent variables will be biased and inconsistent. Following prior studies (see, e.g., O Brien and Bhushan, 1990; Alford and Berger, 1999; and Hong et al. 2014) we use the 2SLS estimation method to jointly estimate the relationship between tax avoidance and financial constraints. In the first stage, we regress each of the jointly determined endogenous variables on a set of instrumental variables and control variables as follows:,,,,,,,,,, (1a),,,,,,, 19

21 ,,, (1b),,,,,,,,,, (2a),,,,,,,,,,, (2b) The fitted values of the endogenous variables obtained from the above first-stage regressions are then used as key explanatory variables in the second stage. In order to obtain consistent estimates in our hypothesis tests and take into account the endogenous simultaneity between tax avoidance and financial constraints, we first select plausible instruments for our proxies of tax avoidance and financial constraints, respectively. Following Desai and Dharmapala (2009) we choose net operating loss carry forwards scaled by total assets, Loss Carry Forward (NOL), as an instrument for tax avoidance. 11 Net operating loss carryforwards can be used by a firm to reduce its tax liabilities. Thus, their availability can substitute or complement the firm s other tax shelters (Graham and Tucker, 2006), and this results in lower effective tax rates (Chen, Chen, Cheng, and Shevlin, 2010) and facilitates tax avoidance (Desai and Dharmapala, 2009). Consistent with these studies, we expect that firms with high historical NOLs will have lower current effective tax rates 11 Desai and Dharmapala (2009) also use leverage as an instrument for tax avoidance. Because leverage is used directly in the calculation of WUscore, we choose only NOL as our only instrument for tax avoidance. 20

22 since they are able to utilize the loss carry-forwards to reduce taxable income and thus cash taxes. We think that it is plausible to argue that while loss carry-forwards from previous years can be directly used as tax shields to reduce ETRs in a current fiscal year, they will not have a direct effect on the firm s current financial constraints. Thus, we posit that historical NOLs affect a firm s current financial constraints only through their effect on the firm s current ETR. We use Tobin s q as an instrument for a firm s Altman Zscore given that firms with better market valuations of their assets are less likely to have financial constraints and less likely to be in financial distress. We think that it is plausible to argue that a firm s market valuation of its assets affects its tax avoidance and its ETRs only through its effect on the firm s financial health and financial constraints. Similarly, we use the firm s own sales change and the industry sales change as instruments for the WUscore. While there is a direct association between these two variables and the WUscore (see Wu and Whited, 2006) by construction, there seems to be no plausible direct effect of sales growth at the firm and industry level to the firm s decisions on tax avoidance. Our second-stage equations are as follows:,,,,,,,,,,, (3a),,,,,,,,,,, (3b) 21

23 ,,,,,,,,,,, (4a),,,,,,,, 8,,,, (4b) 4. Empirical Results 4.1. Summary Statistics and Univariate Test Results We report the summary statistics of the variables in Table 1. The average values of both Cash ETRs are close to 26% which is much lower than the statutory federal tax rate of 35%, whereas those of GAAP ETR are around 33% that is somewhat closer to the statutory tax rate. This suggests that firms are more likely to avoid cash taxes paid than tax expenses. In dollar terms, reducing tax liability even by 2% of pretax income of an average firm can result in savings of about $2.3 million per year. For most profitable firms, tax savings may amount as high as $75 million during a year. Thus, saving cash taxes by a few percentage points can have a significant positive impact on a firm s financial flexibility. The standard deviation of both Cash ETRs are larger than that of GAAP ETRs indicating that some firms avoid cash taxes more than others. While, the average values of Zscore and WUscore suggest that an average firm in our sample is financially unconstrained, the 22

24 magnitudes of their sample standard deviations imply that there exists considerable variation in the financial strengths of firms. Table 2 presents the correlation matrix. We observe a high correlation between both measures of Cash ETR and GAAP ETR suggesting that firms are likely to utilize both measures of tax avoidance simultaneously. We also note that Zscore is positively correlated with Cash ETR measures while WUscore is negatively correlated with both ETR measures. This suggests that cash tax avoidance is more likely to increase a firm s financial distress or financially constrained firms are likely to increase their tax avoidance to save cash. We note that the endogenous variables and our instruments for them are strongly correlated with signs consistent with our expectations. The correlations between Zscore and its instrument, TobinQ, and WUscore and its instruments, Sales Growth and Industry Sales Growth are highly positive, whereas the correlations between these instruments and our measures of tax avoidance are much weaker. Similarly, our instrument for tax avoidance measures, Loss Carry Forwards is strongly and negatively correlated with both ETRs, while weakly correlated with financial constraint measures. In summary, the correlation coefficients seem to justify our choices of instrumental variables statistically and suggest that our 2SLS equations from (2a) to (4b) are properly identified. Before running multivariate regression analysis, we conduct a univariate analysis of the relationship between tax avoidance and financial constraints. For this purpose, we first separate 23

25 firms into four portfolios based on the measures of financial constraints, namely Zscore and WUscore and compare the level of tax avoidance across these portfolios. Panel A (Panel B) in Table 3 reports average ETRs for each Zscore (WUscore) portfolio ranking from most constrained to least constrained. We find that the average Cash ETR of firms in least constrained Zscore group is about percentage points higher than that of firms in most constrained group. The differences for average GAAP ETRs are less pronounced. Our results for WUscore portfolios indicate that the average ETRs of least constrained firms are about percentage points higher. Overall, these results suggest that most financially constrained groups are more likely to engage in tax avoidance. Next, we separate firms into four tax avoidance groups based on the level of their long-run Cash and GAAP ETRs and examine whether financial constraint measures vary across tax avoidance portfolios. Panel C in Table 3 shows that firms in low cash tax avoidance group have higher (lower) Zscore (WUscore) on average. However, in Panel D our findings are mixed and firms with low GAAP tax avoidance have slightly lower Zscore and WUscore on average. In summary, we find some evidence that firms that avoid taxes more are more likely to have greater financial constraints which is consistent with our results for financial constraint portfolios in Panel A and Panel B. The results of univariate analysis support our hypothesis of simultaneous effects between tax avoidance and financial constraints that while more tax avoidance may increase firms financial constraints, firms with greater financial constraints also tend to avoid taxes more. Our multivariate analysis further examines these effects. 24

26 4.2. Simultaneous Equation Analysis of Tax Avoidance and Financial Constraints In this section we present the results of our 2SLS estimation analysis of the joint relationship between tax avoidance and financial constraints. 12 First, we summarize the results from the firststage estimation results for Zscore and ETRs in Panel A in Table 4 and those for WUscore and ETRs in Panel B in Table 4. Our results indicate that the coefficients of all instrumental and control variables are statistically significant and, in general, have the predicted signs in relation to the respective endogenous variables. Specifically, we find that four measures of tax avoidance are negatively and significantly associated with tax loss carry forwards, our instrument for tax avoidance, which is consistent with our expectation that the availability of non-debt tax shields increases tax avoidance. The first-stage results for our proxies of financial constraints show that the coefficient of Tobin s Q, the instrumental variable for Zscore, is positive and significant, which is consistent with the intuition that firms with better market valuations are less likely to be in financial distress. We also find that Sales Growth and Industry Sales Growth are statistically significant and have predicted signs for WUscore. Consistent with Whited and Wu (2006), our results show that firms in faster growing industries face greater financial constraints, whereas firms with higher sales growth are less likely to face financial constraints. We use the fitted values of endogenous variables estimated in the first-stage as key explanatory variables in the second-stage. Next we present our second-stage least-squares estimation results in Panel C and Panel D of Table 4. Our results for the impact of tax avoidance on financial constraints (the first four columns 12 For comparison, we first ran OLS regressions to start our analysis of the relationship between financial constraints and tax avoidance without accounting for the endogeneity of this relationship. In general, these regressions also show that firms with greater financial constraints tend to engage in tax avoidance to a greater extent and that firms engaging in tax avoidance more face greater financial constraints. However, our 2SLS coefficient estimates for ETRs have greater magnitudes than the OLS coefficient estimates for ETRs in both Z-Score and WUscore regressions. This suggests that the effect of tax avoidance activities on the firms financial constraints will be underestimated when the OLS method is used. 25

27 in each panel) indicate that Zscore is positively and WUscore is negatively and significantly associated with all four ETRs. Firms with lower (greater) levels of tax avoidance are less (more) likely to be financially constrained. For example, a one-percent decrease in long-term Cash ETR, the sample standard deviation of which is 14.33%, is associated with a decrease of in Zscore (3.02% of the sample standard deviation of ) and an increase of in WUscore (3.45% of the sample standard deviation of ). Thus, consistent with the information hypothesis, firms avoiding taxes more are likely to be financially constrained and face greater costs of external financing. This result is consistent with the notion that outside investors associate tax avoidance with opaque information environments in which managerial moral hazard (e.g., diversion, earnings management) and the risk of adverse selection for investors are greater (Lev and Nissim, 2004; Hanlon, 2005; Desai and Dharmapala, 2006 and 2009). In such information environments, tax avoidance may increase the uncertainty about the valuations of firms with more aggressive tax sheltering practices, exposing them to higher price crash risks (Kim, Li, and Zhang, 2011). The coefficients of our control variables also suggest that firms with lower profitability, higher capital intensity and higher financial leverage are more likely to be financially constrained. Our results for the impact of financial constraints on tax avoidance are presented in the last four columns of Panel C and Panel D of Table 4. We find that the coefficient of Zscore is negative and statistically insignificant for Cash ETR, but positive and significant for GAAP ETRs. Consistent with the pressure hypothesis, firms with a higher (lower) Zscore, i.e., firms with a smaller (greater) risk of financial distress, are less (more) likely to avoid GAAP tax expenses. When we use WUscore as a proxy for financial constraints, we find that the coefficients of WUscore are negative and statistically significant for measures of both Cash ETRs and GAAP ETRs. For example, a one-standard-deviation increase in WUscore is associated with a decrease 26