1 Introduction

Recent studies have explored the effects of senior managers’ personal characteristics on firm business decisions and the economic consequences (Bantel and Jackson 1989; Bertrand and Schoar 2003; Ge et al. 2011). Bantel and Jackson (1989) find that the education levels of senior managers are positively related to firm performance. Bertrand and Schoar (2003) show that the age of senior managers is positively associated with the conservatism of firms’ investment strategies. Ge et al. (2011) show that senior managers’ demographic characteristics, such as sex, age and education background, have an effect on firms’ financial performance. The findings of previous studies suggest that the work experience and personal characteristics of senior managers may have a profound effect on firms’ various decisions and operating performance. However, few prior studies have examined the effects of personal characteristics of the top executive on corporate tax planning.

Corporate chief financial officers (CFOs) are the key executives responsible for preparing companies’ financial reporting as well as tax returns. Few prior studies have addressed the effect of CFO accounting expertise on corporate effective tax rates (ETRs). Accounting expertise is closely related to corporate taxation in that the determination of taxable income is mainly based on accounting books with the adjustments for book-tax income differences when firms file tax returns. Therefore, accounting expertise helps executives manage their firms’ income taxes while accounting for the effect of tax consequences on financial reporting. For example, the growing book-tax gap suggests that the substantial discretion available in GAAP provides firms the opportunity to manage their book earnings upward without affecting their taxable income (Phillips 2003; Hanlon 2005). Furthermore, prior studies consider the most advantageous type of tax planning one that creates a permanent wedge between financial and taxable incomes—i.e., enabling firms to manage taxable income downward without impacting book income, or vice versa. The increasing trend in permanent book-tax differences since the late 1990s suggests that firms have the opportunity to engage in aggressive financial and tax reporting behaviors in the same reporting period (U.S. Congress Joint Committee on Taxation 1999; Weisbach 2002; McGill and Outslay 2004).

The developments in international taxation and accounting standards also demand competent accounting expertise in dealing with corporate income tax compliance. Companies are now facing increasingly complex international tax compliance issues. For example, base erosion and profit shifting (BEPS) action plans require profit splitting among corporate transfer-pricing transactions to conform to the economic substance of those transactions, which often calls for the analysis of the related parties’ financial statements. Accounting expertise helps to prepare the financial analysis for tax purposes. Furthermore, both tax regulations and accounting standards demand heightened transparency on tax shelter transactions and income tax information. For example, firms are required to file Form 8886 for specified tax shelter transactions and to disclose uncertain tax position according to FIN 48 (or ASC 740-10) in financial statements. Accounting expertise is essential to properly address such complex disclosures. We thus conjecture that an accounting expert CFO may have the advantage of dealing with tax planning over a CFO without accounting expertise.

By constructing a data set that tracks the movement of executives across firms over time, Dyreng et al. (2010) show that individual top executives have incremental effects on their firms’ tax avoidance that cannot be explained by characteristics of the firm. However, they cannot attribute the executive fixed effects to a specific personal characteristic or education background. Furthermore, the executives in their sample include different positions, such as CEOs, CFOs, and other titles, such as presidents and vice-presidents, making it difficult to identify the extent of the CFO effect on corporate tax avoidance. Unlike Dyreng et al. (2010), our study specifically examines the effect of the accounting expertise of CFOs on corporate tax avoidance. In addition, we investigate the moderating effect of compensation design on the relation between CFO with accounting expertise and corporate tax avoidance. Compensation incentives play an essential role in motivating managers to link effort to performance (Core and Guay 1999; Carter et al. 2007; Hoitash et al. 2012). We expect that the accounting expertise of CFOs is positively related to corporate tax avoidance and that the compensation design for CFOs with accounting expertise will motivate them to further explore tax planning opportunities.

By using a sample of firms with neither CEO nor CFO turnovers during 2010–2012 from the ExecuComp database, we are able to ensure a consistent company-wide tax planning policy under the same management within a company. We then manually collect CFOs’ work background and autobiographic information from S&P Capital IQ files to identify CFO’s accounting expertise and conduct the two-stage regression estimation to control for the potential endogeneity problem that arises when firms may self-select their CFOs with accounting expertise. Our results show that, ceteris paribus, CFOs with accounting expertise are associated with lower ETRs, suggesting that CFOs with accounting expertise are more likely to explore tax planning opportunities. Furthermore, we find that the abnormal variable pay of CFO with accounting expertise is negatively associated with corporate ETRs, consistent with our expectations that compensation design, such as variable pay, has an effect on the efforts of CFOs with accounting expertise in exploiting corporate tax planning opportunities.

Our study makes the following contributions. First, it contributes to the existing literature on corporate tax avoidance. Prior studies have addressed the effects of firm characteristics and corporate governance on corporate tax avoidance (Chen et al. 2010; Armstrong et al. 2012; McGuire et al. 2012). However, little is known about whether CFOs’ accounting expertise and compensation design affect corporate tax avoidance. Our study fills this gap in the literature.

Second, our empirical findings complement the literature on managerial compensation designs. Agency theory applies to how to design an efficient compensation scheme to drive managers’ effort to better firm performance. However, efficient compensation schemes may vary with the characteristics of executives in driving efforts toward better results. Our study considers both the incentive design and expertise of executives in examining the relationship between the incentive compensation of accounting expert CFOs and corporate tax avoidance. Our findings help further understand the impact of CFO compensation design interacted with accounting expertise on corporate tax avoidance.

The remainder of this paper proceeds as follows. Section 2 discusses the related literature. Section 3 develops the research hypotheses and describes the research design and data. Section 4 reports our empirical findings. Section 5 discusses robust and additional tests, and Sect. 6 concludes.

2 Related literature

2.1 Effects of personal characteristics of top managers on corporate decisions

The upper echelons theory (Hambrick and Mason 1984) suggests that the experiences, values, and personalities of executives greatly influence their interpretations of the situations they face and, in turn, affect their corporate decision making. Prior research has documented the evidence that various firm policies and decisions are associated with top managers’ personal characteristics, such as demographic characteristics (Bantel and Jackson 1989; Bertrand and Schoar 2003; Barua et al. 2010; Francis et al. 2013; Cotei and Farhat 2017). Bantel and Jackson (1989) show that top manager education levels and diversity in function areas of expertise are positively associated with banks’ innovative ability. Bertrand and Schoar (2003) also show that company-wide policies and performance are affected by the turnovers of top managers with different ages and educational backgrounds. Barua et al. (2010) and Francis et al. (2013) indicate that female CFOs tend to be more conservative in corporate financial reporting, and thus, companies with female CFOs have lower discretionary accruals and lower accrual estimation errors than their counterparts with male CFOs. Cotei and Farhat (2017) show that owners’ characteristics, such as experience, education, age, gender, and race, have a significant impact on the decision to lease assets for small startup firms.

In addition to the effects of demographic characteristics, the accounting professional experience of top managers may influence firms’ financial reporting. Prior studies show that companies with accounting expert CFOs tend to be more conservative with external financial reporting and precise in financial disclosure styles and thus are negatively associated with a probability of financial statement restatements or material errors in financial statements (Aier et al. 2005; Bamber et al. 2010; Ge et al. 2011; Hoitash et al. 2016). Rakhman (2009) also shows that companies with accounting expert CFOs tend to have better earnings persistence and earnings informativeness.

Recent studies investigate the effect of top mangers’ characteristics on corporate tax avoidance (Dyreng et al. 2010; Francis et al. 2015; Law and Mills 2017). Francis et al. (2015) find that firms with female CFOs are associated with less tax avoidance. Law and Mills (2017) indicate that managers with military experience tend to avoid conducting aggressive tax planning strategies and believe that avoiding taxes would be unethical. Dyreng et al. (2010) investigate whether top executives have an incremental effect on their firms’ tax avoidance that cannot be explained by firm characteristics. Their results indicate that the individual fixed effects of the executives between the top and bottom quartiles account for approximately 11 percent of GAAP ETRs, suggesting that individual executives appear to be an important determinant in firms’ tax avoidance. The results of Dyreng et al. (2010), however, do not separate the effect of CFOs on corporate tax avoidance from that of other top executives and cannot attribute the tone of executive fixed effects to a specific type of expertise.

The trend in global anti-tax-avoidance cooperation and the demand for greater transparency in tax positions highlight the necessity of accounting and tax expertise in dealing with corporate tax compliance. CFOs are the key executives responsible for preparing company financial reports and tax returns. The prior studies, however, have not addressed the effect of CFOs’ accounting expertise on corporate tax avoidance. We thus seek to fill the research gap by investigating the relationship between CFOs’ accounting expertise and corporate tax avoidance.

2.2 Effects of compensation design of top managers on performance and tax avoidance

Efficient managerial compensation design has long been a focal topic of corporate governance regarding its potential to reduce agency costs and enhance firm performance (Core and Guay 1999; Carter et al. 2007; Hoitash et al. 2012). Prior literature shows the importance of incentive compensation design for the improvement of individual manager and firm performance Hoitash(Gerhart and Milkovich 1990; Mehran 1995; Banker et al. 1996, 2000; Hayes and Schaefer 2000).

Incentive compensation, such as variable pay, provides individual managers with an economic incentive to improve firm performance. Using large-scale, longitudinal data on approximately 14,000 top- and middle-level managers, Gerhart and Milkovich (1990) show that variable pay, such as bonus and long-term incentives, is positively associated with firms’ financial performance, whereas the association between fixed pay and financial performance is not significant. Mehran (1995) and Banker et al. (1996) also find a positive relation between the percent of incentive compensation of executive compensation and firm performance. Additionally, Hayes and Schaefer (2000) show that the abnormal compensation of top managers is positively associated with the future performance of a firm.

In addition to the effect of compensation on firm performance, prior studies have examined the effect of incentive compensation on corporate tax avoidance. Phillips (2003), Desai and Dharmapala (2006), Robinson et al. (2010), and Armstrong et al. (2012) document that incentive compensation is positively related to corporate tax avoidance because greater incentive compensation helps align the incentives of agents and principals, resulting in lower corporate ETRs. In addition, Schmittdiel (2014) finds that companies with greater tax planning opportunities are more likely to link CEOs’ bonuses to corporate income taxes. Hansen et al. (2017) find that both CEOs and CFOs are compensated for the earnings generated by changes in ETRs (i.e., the tax component of earnings) when the firms pay bonuses based on after-tax earnings.

Previous studies suggest that agency theory applies to link the incentive design of executives to corporate tax avoidance. Our paper extends prior studies by incorporating compensation design as a moderating factor in examining the effect of CFO accounting expertise on exploring corporate tax planning opportunities.

3 Research methods

3.1 Research hypothesis

Previous studies find that female managers and managers with military experience are less aggressive in tax avoidance (Francis et al. 2015; Law and Mills 2017). Dyreng et al. (2010) also show that the intangible tone of top executives is associated with corporate tax avoidance. Our subject of interest is the effect of accounting expert CFOs on corporate tax avoidance.

Accounting expertise is closely related to corporate financial reporting and tax planning because taxable income is determined mainly based on financial income with the adjustments for book-tax income differences. Frank et al. (2009) indicate that corporate financial reporting strategies are closely related to tax planning strategies. The most advantageous type of tax planning enables firms to manage taxable income downward without impacting book income, or vice versa (U.S. Congress Joint Committee on Taxation 1999; Weisbach 2002; McGill and Outslay 2004). The growing trend in the book-tax gap suggests that accounting expertise can help executives better manage their firms’ income taxes by taking advantage of the substantial discretions available in GAAP (Phillips et al. 2003; Hanlon 2005).

Furthermore, the recent development in accounting standards and increasing awareness of global anti-tax avoidance demand competent accounting expertise in dealing with corporate income tax compliance. For example, ASC 740-10 (or FIN 48) requires firms to reflect uncertain tax benefits in accounting for income taxes. US taxpayers are also required to file Form 8886 when participating in the specified tax shelter transactions. The demand for greater transparency in tax position highlights the importance of accounting expertise in dealing with increasingly complex tax compliance. CFOs are those primarily responsible for corporate financial reporting and tax-return preparation. Therefore, we conjecture that CFOs with accounting expertise are more likely to explore tax planning opportunities while accounting for the effect of tax consequences on financial reporting. Accordingly, we propose our first hypothesis as follows.

H1

Ceteris paribus, the accounting expertise of chief financial officers is positively associated with the level of corporate tax avoidance.

Previous research shows that management compensation schemes have a profound effect on managerial behavior and corporate operating performance and that incentive compensation, such as variable pay, provides managers with an economic incentive to improve firm performance (Gerhart and Milkovich 1990; Mehran 1995; Banker et al. 1996, 2000). Hayes and Schaefer (2000) and Combs and Skill (2003) also find that abnormal compensation is positively associated with firm performance.

In addition, prior research documents the positive effect of incentive compensation on corporate tax avoidance. Phillips (2003), Desai and Dharmapala (2006), Robinson et al. (2010), Armstrong et al. (2012) and Hansen et al. (2017) find that greater incentive compensation helps motivate executives to exploit aggressive tax planning opportunities, resulting in lower corporate ETRs. As agency theory suggests that compensation incentives play an important role in linking managers’ effort to performance, we thus incorporate compensation design as a moderating factor in examining the effect of CFO accounting expertise on corporate tax avoidance. We expect that greater compensation incentives will better incentivize CFOs with accounting expertise to utilize this expertise to explore tax planning opportunities, resulting in a lower corporate tax. Hence, we state our second hypothesis as follows.

H2

Ceteris paribus, the abnormal compensation of chief financial officers with accounting expertise is positively associated with the level of corporate tax avoidance.

3.2 Empirical models and variable definitions

3.2.1 Effect of CFO with accounting expertise on tax aggressiveness

H1 tests whether the expertise of a CFO has an effect on corporate tax avoidance. To address the potential self-selection problem of a CFO with/without accounting expertise within the companies, we adopt Heckman’s (1979) two-stage regression estimation. We first estimate the inverse Mills ratio (MILLS) in Model (1) by running a probit regression on whether CFO has accounting expertise and then include MILLS in Model (2) as a control variable to control for the self-selection problem of firms choosing CFOs with/without accounting expertise. The two-stage regression models are as follows.

Stage 1:

Probit estimation of firms’ selection of a CFO with accounting expertise

$$ CFO\_ACC_{\rm it} = \alpha_{0} + \alpha_{1} SIZE_{\rm it} + \alpha_{2} NOL_{\rm it} + \alpha_{3} FOREIGN_{\rm it} + \alpha_{4} BTM_{\rm it} + \alpha_{5} MBA_{\rm it} + \alpha_{6} GENDER_{\rm it} + \alpha_{7} AGE_{\rm it} + \varepsilon_{\rm it} $$
(1)

The subscripts i and t denote firm and year, respectively. The dependent variable CFO_ACC is a dummy variable that equals 1 if the firm has a CFO with accounting expertise and 0 otherwise. We identify a CFO as having accounting expertise if he or she meets one of following criteria: (a) has been a partner of an accounting firm, (b) self-identifies as possessing accounting expertise in his or her autobiography, or (c) has been a CFO of another company and has an accountant degree. The following are brief definitions of the independent variables.

SIZE:

firm size, measured as the natural log value of total assets;

NOL:

a dummy variable that equals 1 if the firm has net operating loss carryforward and 0 otherwise;

FOREIGN:

foreign earnings, measured as foreign earnings ÷ total assets at year t − 1;

BTM:

book to market ratio, measured as book value of shareholder equity divided by market value of shareholder equity;

MBA:

a dummy variable that equals 1 if the firm’s CFO has an MBA degree and 0 otherwise;

GENDER:

a dummy variable that equals 1 if the firm’s CFO is male and 0 otherwise; and

AGE:

age of CFO, measured by the value of the CFO’s age

Independent variables


Omer et al. (2006) and Bernard et al. (2015) show that firms with more complex operations are more likely to hire a CFO with accounting expertise to address complex accounting and tax issues. Therefore, we include SIZE, NOL and FOREIGN to control for the complexity of firms’ operation in Model (1). BTM is used to control for the growth opportunities of firms. McGuire et al. (2012) and Klassen et al. (2013) indicate that firms with more growth opportunities tend to appoint accounting firms to address tax planning. Following previous studies, we also include the education background (MBA), gender (GENDER), and age (AGE) of CFOs to control for the effects of personal demographic characteristics in Model (1).

Stage 2:

Effects of CFO with accounting expertise on tax aggressiveness

We next construct Model (2) to examine the effect of CFOs with accounting expertise on tax avoidance. We control for the potential endogeneity between CFO_ACC and ETR by including the inverse Mills ratio (MILLS) obtained from Model (1) in Model (2). Following prior studies (Chen et al. 2001; Rego 2003; Cook et al. 2008; Dyreng et al. 2010; Hanlon and Heitzman 2010; Robinson et al. 2010; McGuire et al. 2012; Hansen et al. 2017; Cordis and Kirby 2018), we use corporate ETRs (ETRs) as the proxy variable to measure the levels of corporate tax avoidance in that the ETR summarizes the overall tax burden of a company in a statistic. ETR is defined as income tax expenses divided by pretax income. Model (2) is as follows.

$$ ETR_{\rm it} = \beta_{0} + \beta_{1} CFO\_ACC_{\rm it} + \beta_{2} SIZE_{\rm it} + \beta_{3} LEV_{\rm it} + \beta_{4} ROA_{\rm it} + \beta_{5} FOREIGN_{\rm it} + \beta_{6} BTM_{\rm it} + \beta_{7} NOL_{\rm it} + \beta_{8} DEP_{\rm it} + \beta_{9} EQINC_{\rm it} + \beta_{10} MILLS_{\rm it} + IND + \varepsilon_{\rm it} $$
(2)

where LEV debt ratio, measured as long-term liabilities ÷ total assets; ROA return on assets, measured as net income ÷ average total assets; DEP depreciation and amortization expense, measured as (depreciation expense + amortization expense) ÷ total assets at the beginning of year t; EQINC investment income of equity method investments, measured as investment income under the equity method ÷ total assets at the beginning of year t; and IND dummy variables for firms’ industry membership.

In H1, we hypothesize that the accounting expertise of CFOs is positively associated with the level of corporate tax avoidance, and we therefore expect the coefficient on CFO_ACC to be negative in Model (2).

Following prior ETR studies (Slemrod and Blumenthal 1996; Mills et al. 1998; Omer et al. 2006; Cook et al. 2008; Frank et al. 2009; Lassila et al. 2010; Krishnan and Visvanathan 2011; McGuire et al. 2012; Klassen et al. 2013), we also control for the factors associated with ETR, such as tax shields (NOL, DEP), financing policy (LEV), tax complexity (FOREIGN, EQINC), profitability (ROA), growth opportunities (BTM), firm size (SIZE), and industry membership (IND) in Model (2).

As for the instrumental variables used for estimating Model (2), we select proxy variables for tax planning opportunities (measured by the number of segments and the ratio of property, plant and equipment to total assets), audit quality (measured by discretionary accruals and the dummy variable for corporate financial statements being audited by Big Four accounting firms), and non-audit services (measured by the non-audit fees scaled by total assets) to control for their potential indirect effects on ETR through the independent variables of Model (2). Following the exclusion restrictions noted in Lennox et al. (2012, 596), we exclude the independent variables MBA, GENDER, and AGE in the probit estimation from Model (2) to satisfy exclusion restrictions and conduct the Sargan test of the overidentifying restrictions to examine whether the selected instrumental variables for Model (2) are valid.

3.2.2 Effect of CFO compensation on tax avoidance

H2 examines the effect of the compensation scheme of CFOs with accounting expertise on tax avoidance. We separate the compensation of CFOs into fixed and variable pay and construct Model (3) to test the effects of abnormal fixed and variable pay of CFO compensation on corporate tax avoidance. Model (3) is as follows.

$$ ETR_{\rm it} = \eta_{0} + \eta_{1} CFO\_ACC_{\rm it} + \eta_{2} AB\_FIXED_{\rm it} + \eta_{3} AB\_VAR_{\rm it} + \eta_{4} CFO\_ACC \times_{{}} AB\_FIXED_{\rm it} + \eta_{5} CFO\_ACC \times_{{}} AB\_VAR_{\rm it} + \eta_{6} SIZE_{\rm it} + \eta_{7} LEV_{\rm it} + \eta_{8} ROA_{\rm it} + \eta_{9} FOREIGN_{\rm it} + \eta_{10} BTM_{\rm it} + \eta_{11} NOL_{\rm it} + \eta_{12} DEP_{\rm it} + \eta_{13} EQINC_{\rm it} + \eta_{14} MILLS_{\rm it} + IND + \varepsilon_{\rm it} $$
(3)

where AB_FIXED abnormal fixed compensation of CFO, measured by taking the residuals from Model (4); and AB_VAR abnormal variable compensation of CFO, measured by taking the residuals from Model (5).

To calculate abnormal fixed and variable pay of CFO compensation, we construct Models (4) and (5) to estimate the expected fixed (FIXED_PAY) and variable (VARIABLE_PAY) components of CFO compensation, respectively. Models (4) and (5), separately, regress fixed and variable pay of CFOs on factors associated with management compensation, including firm size (SIZE), financial performance (ROE), financial leverage (LEV), book to market ratio (BTM), variability of profitability (ROA_STD), financial constraint (CASH_CONSTR), equity interest (CFO_OWNED) and CFO demographic characteristics variables, such as GENDER, AGE, CPA, and MBA (Hogan and McPheters 1980; Core and Guay 1999,2001; Carter et al. 2007). Models (4) and (5) are as follows.

$$ FIXED\_PAY_{\rm it} = \gamma_{0} + \gamma_{1} SIZE_{\rm it} + \gamma_{2} ROE_{\rm it} + \gamma_{3} LEV_{\rm it} + \gamma_{4} BTM_{\rm it} + \gamma_{5} ROA\_STD_{\rm it} + \gamma_{6} CASH\_CONSTR_{\rm it} + \gamma_{7} CFO\_OWNED_{\rm it} + \gamma_{8} GENDER_{\rm it} + \gamma_{9} AGE_{\rm it} + \gamma_{10} CPA_{\rm it} + \gamma_{11} MBA_{\rm it} + \varepsilon_{\rm it} $$
(4)
$$ VARIABLE\_PAY_{\rm it} = \delta_{0} + \delta_{1} SIZE_{\rm it} + \delta_{2} ROE_{\rm it} + \delta_{3} LEV_{\rm it} + \delta_{4} BTM_{\rm it} + \delta_{5} ROA\_STD_{\rm it} + \delta_{6} CASH\_CONSTRIt + \delta_{7} CFO\_OWNED_{\rm it} + \delta_{8} GENDER_{\rm it} + \delta_{9} AGE_{\rm it} + \delta_{10} CPA_{\rm it} + \delta_{11} MBA_{\rm it} + \varepsilon_{\rm it} $$
(5)

where FIXED_PAY CFO’s fixed compensation, measured as CFO’s fixed salaries ÷ total assets at the beginning of year t; VARIABLE_PAY CFO’s variable compensation, measured as (CFO’s total compensation—CFO’s fixed salaries) ÷ total assets at the beginning of year t; ROA_STD operating risk, measured by the standard deviation of returns on assets over the preceding 10 years; CASH_CONSTR cash constraint, measured by (cash dividend—cash flow from investing activities—cash flow from operating activities) ÷ total assets at the beginning of year t; CFO_OWNED CFO’s ownership, measured by the percentage of shares owned by CFO; and CPA a dummy variable that equals 1 if a CFO has a CPA license and 0 otherwise.

3.2.3 Data and sample selection

Table 1 outlines the sample selection procedures. Our sample is selected from the ExecuComp database. We start with firms having neither CEO nor CFO turnovers from 2010–2012 to ensure a consistent company-wide tax planning policy with the same management. We delete firms in the financial industry (SIC codes 6000–6999) and utilities industry (SIC codes 4900–4999) because of their particular regulatory constraints. CFO’s fixed salaries, variable bonuses and payments, and total compensation are collected from ExecuComp files. We also hand collect CFOs’ background, education, and autobiographic information from S&P Capital IQ files to identify CFOs’ accounting expertise. Financial statement variables are from the Compustat files. Our final sample consists of 969 firm-year observations.

Table 1 Sample selection procedures
Full size table

4 Empirical results

4.1 Descriptive statistics

Table 2 profiles the descriptive statistics of our sample firms for the selected variables. The mean value of ETR is 0.283, which is less than the current US statutory tax rate of 35%. The mean value of CFO_ACC is 0.25, indicating that approximately one-fourth of the companies in our sample have CFOs with accounting expertise. The mean values for GENDER, AGE, and MBA are 0.94, 52.71 and 0.46, respectively. The statistics on CFO demographic characteristics in our sample show that men dominate in the CFO position and that nearly half of CFOs received an MBA degree.

Table 2 Descriptive statistics (N = 969)
Full size table
ETR:

effective tax rate, measured as current income tax expense divided by income before tax;

CFO_ACC:

CFO with accounting expertise, a dummy variable that equals 1 if a firm has a CFO with accounting expertise, and 0 otherwise;

AB_FIXED:

abnormal fixed compensation of CFO, measured by taking the residuals from Model (4);

AB_VAR:

abnormal variable compensation of CFO, measured by taking the residuals from Model (5);

SIZE:

firm size, measured as the natural log value of total assets at year t;

NOL:

net operating loss carryforward, a dummy variable that equals 1 if a firm has net operating loss carryforward, and 0 otherwise;

FOREIGN:

foreign earnings, measured as foreign earnings ÷ total assets at year t − 1;

BTM:

book-to-market ratio, measured as book value of shareholder equity ÷ market value of shareholder equity;

LEV:

debt ratio, measured as long-term liabilities ÷ total assets at year t;

ROA:

pre-tax return on assets, measured as income before tax divided by average assets;

DEP:

depreciation and amortization expense, measured as (depreciation expense + amortization expense) ÷ total assets at year t −  1;

EQINC:

investment income, measured as investment income under the equity method ÷ total assets at year t −  1;

FIXED_PAY:

CFO’s fixed compensation, measured as CFO’s fixed salaries ÷ total assets at year t −  1;

VARIABLE_PAY:

CFO’s variable compensation, measured as (CFO’s total compensation – CFO’s fixed salaries) ÷ total assets at year t −  1;

ROE:

pre-tax return on equity, measured as income before tax divided by average shareholder equity;

ROA_STD:

operating risk, measured by the standard deviation of returns on assets over the preceding 10 years;

CASH_CONSTR:

cash constraint, measured as (cash dividend—cash flow from investing activities- cash flow from operating activities) ÷ total assets at year t −  1;

CFO_OWNED:

CFO’s ownership, measured by the percentage of shares owned by CFO;

CPA:

CPA license, a dummy variable that equals 1 if a CFO has a CPA license and 0 otherwise;

MBA:

MBA degree, a dummy variable that equals 1 if a CFO has an MBA degree and zero otherwise;

GENDER:

gender of CFO, a dummy variable that equals 1 if a CFO is male and 0 otherwise; and

AGE:

age of CFO, measured by the value of CFO’s age.

Table 3 presents the correlation coefficients for the dependent and independent variables of Models (1–5). CFO_ACC is positively related to CPA but negatively related to MBA. The results are consistent with the notion that accounting expertise is more related to professional education than to general management education. ETR, however, is insignificantly related to our variable of interest CFO_ACC. The univariate test result does not lend support to our hypotheses. As the univariate relations do not control for the effects of other factors, we conduct further regression estimations of Models (1) to (3) to test our hypotheses.

Table 3 Correlation matrix
Full size table

4.2 Empirical results and analysis

Table 4 presents the probit estimation results of Model (1), and Table 5 presents the regression results of Model (2). The p-values of the χ2-statistic of Model (1) and the F-statistic of Model (2) are significant at 0.0001, suggesting that both models have an overall satisfactory goodness of fit. We conduct the Sargan test of the overidentifying restrictions to examine whether the selected instrumental variables for Model (2) are valid. The joint null hypothesis of the test is that the instruments are valid, i.e., uncorrelated with the error terms in Model (2), and the excluded instruments are correctly excluded from Model (2). The results of the Sargan test show that the p value of the χ2-statistic for rejecting the null hypothesis is 0.24, suggesting that our results satisfy the overidentifying restrictions tests.

Table 4 Probit estimation of firms’ selection of a CFO with accounting expertise \( CFO\_ACC_{\rm it} = \alpha_{0} + \alpha_{1} SIZE_{\rm it} + \alpha_{2} NOL_{\rm it} + \alpha_{3} FOREIGN_{\rm it} + \alpha_{4} BTM_{\rm it} + \alpha_{5} MBA_{\rm it} + \alpha_{6} GENDER_{\rm it} + \alpha_{7} AGE_{\rm it} + \varepsilon_{\rm it} \quad (1) \)
Full size table
Table 5 The impact of CFO with accounting expertise on tax aggressiveness \( ETR_{\rm it} = \beta_{0} + \beta_{1} CFO\_ACC_{\rm it} + \beta_{2} SIZE_{\rm it} + \beta_{3} LEV_{\rm it} + \beta_{4} ROA_{\rm it} + \beta_{5} FOREIGN_{\rm it} + \beta_{6} BTM_{\rm it} + \beta_{7} NOL_{\rm it} + \beta_{8} DEP_{\rm it} + \beta_{9} EQINC_{\rm it} + \beta_{10} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \quad (2) \)
Full size table

Consistent with our H1, the results of Table 5 show that the coefficient on CFO_ACC is significantly negative (p-value = 0.079), suggesting that companies with accounting expert CFOs are associated with lower ETRs. CFOs with accounting expertise are more likely to exploit tax planning opportunities, resulting in a lower ETR. Regarding the economic significance, the coefficient of CFO_ACC is − 0.194, indicating that, ceteris paribus, the average ETR of firms with accounting expert CFOs is approximately 19.4% lower than that of their counterparts with non-accounting expert CFOs. These results suggest that CFOs with accounting expertise have a noticeable effect on the level of corporate tax avoidance. The coefficient on LEV is significantly negative (p-value = 0.067), consistent with the notion that interest tax shield lowers corporate taxes. In addition, the coefficients on FOREIGN and EQINC are negative and significant (p-values < 0.01), suggesting that tax complexity provides companies with greater opportunities to exploit aggressive tax planning.

Table 6 reports the regression results of Model (3). The p-value of the F-statistic of Model (3) is significant at 0.001, suggesting that the model has an overall satisfactory goodness of fit. In addition, the results of the Sargan test show that the p-value of the χ2-statistic for rejecting the null hypothesis that overidentifying restrictions are valid is 0.22, suggesting that the results satisfy the overidentifying restrictions and that the instrumental variables of Model (3) are valid as well.

Table 6 The impact of compensation of CFO with accounting expertise on tax aggressiveness \( ETR_{\rm it} = \eta_{0} + \eta_{1} CFO\_ACC_{\rm it} + \eta_{2} AB\_FIXED_{\rm it} + \eta_{3} AB\_VAR_{\rm it} + \eta_{4} CFO\_ACC \times_{{}} AB\_FIXED_{\rm it} + \eta_{5} CFO\_ACC \times_{{}} AB\_VAR_{\rm it} + \eta_{6} SIZE_{\rm it} + \eta_{7} LEV_{\rm it} + \eta_{8} ROA_{\rm it} + \eta_{9} FOREIGN_{\rm it} + \eta_{10} BTM_{\rm it} + \eta_{11} NOL_{\rm it} + \eta_{12} DEP_{\rm it} + \eta_{13} EQINC_{\rm it} + \eta_{14} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \quad (3) \)
Full size table

The results of Table 6 show that the coefficient on CFO_ACC × AB_FIXED is positive but nonsignificant, while the coefficient on CFO_ACC × AB_VAR is negative and significant (p-value = 0.047). The results indicate that the abnormal variable pay of CFOs with accounting expertise is negatively related to ETRs, consistent with our H2. Higher variable pay may drive CFOs with accounting expertise to exploit more aggressive tax planning, resulting in a lower ETR. However, the effect is not salient for companies paying higher fixed pay to their CFO with accounting expertise. The difference in the effects between variable and fixed pay interacted with accounting expert CFOs is consistent with the notion that a variable compensation scheme provides a greater incentive than a fixed compensation scheme to motivate executive effort toward the desired results. The coefficients on LEV, FOREIGN and EQINC remain negative and significant (p-value = 0.073, 0.000 and 0.003, respectively), consistent with the notion of the interest tax shield effect and tax complexity effect.

5 Robust and additional tests

5.1 Alternative definition of CFO with accounting expertise

A potential concern with our findings is whether our results are robust to different definitions of CFO with accounting expertise. To address this concern, we follow the definition of CFO with accounting expertise in Bedard et al. (2014), where a CFO with accounting expertise is either (1) a CFO having a CPA license or (2) a CFO who has worked as a CFO for other companies. We reconstruct our variable of interest, CFO_ACC, according to this definition (CFO_ACC_NEW) and conduct regression tests for Models (2) and (3). Tables 7 and 8 present the regression results for Models (2) and (3), respectively, using the alternative measurement of CFO_ACC_NEW.

Table 7 The impact of CFO with accounting expertise on tax aggressiveness-changing the definition of CFO_ACC \( ETR_{\rm it} = \beta_{0} + \beta_{1} CFO\_ACC\_NEW_{\rm it} + \beta_{2} SIZE_{\rm it} + \beta_{3} LEV_{\rm it} + \beta_{4} ROA_{\rm it} + \beta_{5} FOREIGN_{\rm it} + \beta_{6} BTM_{\rm it} + \beta_{7} NOL_{\rm it} + \beta_{8} DEP_{\rm it} + \beta_{9} EQINC_{\rm it} + \beta_{10} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \)
Full size table
Table 8 The impact of compensation of CFO with accounting expertise on tax aggressiveness-changing the definition of CFO_ACC \( ETR_{\rm it} = \eta_{0} + \eta_{1} CFO\_ACC\_NEW_{\rm it} + \eta_{2} AB\_FIXED_{\rm it} + \eta_{3} AB\_VAR_{\rm it} + \eta_{4} CFO\_ACC\_NEW \times_{{}} AB\_FIXED_{\rm it} + \eta_{5} CFO\_ACC\_NEW \times_{{}} AB\_VAR_{\rm it} + \eta_{6} SIZE_{\rm it} + \eta_{7} LEV_{\rm it} + \eta_{8} ROA_{\rm it} + \eta_{9} FOREIGN_{\rm it} + \eta_{10} BTM_{\rm it} + \eta_{11} NOL_{\rm it} + \eta_{12} DEP_{\rm it} + \eta_{13} EQINC_{\rm it} + \eta_{14} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \)
Full size table

The results remain consistent with those in Tables 5 and 6. The results of Table 7 show that the coefficient on CFO_ACC_NEW is negative and significant (p-value = 0.07), consistent with our H1 that CFOs with accounting expertise are related to greater corporate tax avoidance. The results of Table 8 show that the coefficient on CFO_ACC_NEW × AB_VAR is negative and significant (p-value = 0.03), supporting our H2 that a higher variable pay scheme interacted with accounting expertise of CFOs results in greater corporate tax avoidance. Hence, our findings are robust to the alternative definition of CFO with accounting expertise.

5.2 Components of variable incentive compensation

The results of Model (3) suggest that variable compensation better motivates accounting expert CFOs to develop more tax planning strategies. To examine the effects of different components of variable compensation interacted with accounting expert CFOs on corporate tax avoidance, we separate the abnormal variable compensation of CFOs (AB_VAR) into abnormal bonusesFootnote 1 (AB_BONUS), abnormal options award (AB_OPTION) and others (AB_OTHERS). The different components of variable pay may have different risks in the nature of incentive compensation. For example, the value of the option award may be contingent on the future stock returns of the issuing companies. Table 9 presents the results of the regression model for the three components of abnormal variable pay. The coefficients on CFO_ACC × AB_BONUS and CFO_ACC × AB_OTHERS are negative and significant (p-values = 0.006 and 0.011, respectively). The coefficient on CFO_ACC × AB_OPTION, however, is negative but nonsignificant, consistent with the notion that executives with accounting backgrounds tend to be more conservative in financial decisions (Bamber et al. 2010; Hoitash et al. 2016) and thus may prefer less risky incentive award instruments, such as bonuses. The results suggest that the risk attribute of different variable compensation instruments may have different effects on motivating accounting expert CFOs to adopt more tax planning strategies.

Table 9 The impact of compensation of CFO with accounting expertise on tax aggressiveness-dividing the components of incentive payment \( ETR_{\rm it} = \eta_{0} + \eta_{1} CFO\_ACC_{\rm it} + \eta_{2} AB\_FIXED_{\rm it} + \eta_{3} AB\_BONUS_{\rm it} + \eta_{4} AB\_OPTION_{\rm it} + \eta_{5} AB\_OTHERS_{\rm it} + \eta_{6} CFO\_ACC \times AB\_FIXED_{\rm it} + \eta_{7} CFO\_ACC \times_{{}} AB\_BONUS_{\rm it} + \eta_{8} CFO\_ACC \times_{{}} AB\_OPTION_{\rm it} + \eta_{9} CFO\_ACC \times_{{}} AB\_OTHERS_{\rm it} + \eta_{10} SIZE_{\rm it} + \eta_{11} LEV_{\rm it} + \eta_{12} ROA_{\rm it} + \eta_{13} FOREIGN_{\rm it} + \eta_{14} BTM_{\rm it} + \eta_{15} NOL_{\rm it} + \eta_{16} DEP_{\rm it} + \eta_{17} EQINC_{\rm it} + \eta_{18} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \)
Full size table

5.3 The effect between accounting expert CFOs from within or outside the company

Prior studies find that the decision to employ top management teams from either inside or outside a company affects future corporate strategies and performance (Kotter 1982; Dalton and Kesner 1983; Wiersema and Bantel 1992). Successors of top executives employed from outside a company are more likely to adopt changes in corporate strategies, whereas top executives promoted from within the company are likely to adopt a maintenance strategy. To examine the effects of accounting expert CFOs from outside or within the company on tax avoidance, we separate accounting expert CFOs into those recruited from outside the company and those promoted from within the company. Our test variable is OUTSIDE, which is set to one if the accounting expert CFO is recruited from outside of the company and zero if internally promoted from within the company. Table 10 presents the regression results for testing the effect of inside/outside succession of CFOs on corporate tax avoidance. The coefficient on OUTSIDE is negative and significant (p-value = 0.069). The result suggests that accounting expert CFOs recruited from outside the company are likely to adopt more tax planning strategies, resulting in lower ETR, consistent with the notion that outside successors of top executives are more likely to adopt changes in corporate strategies.

Table 10 The impact of CFO with accounting expertise on tax aggressiveness-CFO from within/outside of the company \( ETR_{\rm it} = \beta_{0} + \beta_{1} OUTSIDE_{\rm it} + \beta_{2} SIZE_{\rm it} + \beta_{3} LEV_{\rm it} + \beta_{4} ROA_{\rm it} + \beta_{5} FOREIGN_{\rm it} + \beta_{6} BTM_{\rm it} + \beta_{7} NOL_{\rm it} + \beta_{8} DEP_{\rm it} + \beta_{9} EQINC_{\rm it} + \beta_{10} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \)
Full size table

5.4 Controlling for the effect of CEO with accounting expertise

Baker et al. (2018) indicate that the relative powers of CEOs and CFOs may have an impact on the extent of firms’ accruals and real earnings management. In addition, Matsunaga and Yeung (2008) and Jiang et al. (2013) find that companies of CEOs with financial expertise are associated with more conservative earnings reporting and higher earnings quality. CEOs with accounting expertise may have a greater influence in directing corporate tax reporting decisions than their counterparts without accounting expertise. To control for the potential impact of accounting expert CEOs on the relation between CFOs with accounting expertise and corporate tax avoidance, we construct Models (6) and (7) by adding the variable CEO_ACC to Models (2) and (3), respectively. CEO_ACC is defined as one if the CEO of the company is an accounting expert and zero otherwise. Models (6) and (7) are as follows.

$$ ETR_{\rm it} = \beta_{0} + \beta_{1} CFO\_ACC_{\rm it} + \beta_{2} SIZE_{\rm it} + \beta_{3} LEV_{\rm it} + \beta_{4} ROA_{\rm it} + \beta_{5} FOREIGN_{\rm it} + \beta_{6} BTM_{\rm it} + \beta_{7} NOL_{\rm it} + \beta_{8} DEP_{\rm it} + \beta_{9} EQINC_{\rm it} + \beta_{10} CEO\_ACC_{\rm it} + \beta_{11} MILLS_{\rm it} + IND + \varepsilon_{\rm it} $$
(6)
$$ ETR_{\rm it} = \eta_{0} + \eta_{1} CFO\_ACC_{\rm it} + \eta_{2} AB\_FIXED_{\rm it} + \eta_{3} AB\_VAR_{\rm it} + \eta_{4} CFO\_ACC \times_{{}} AB\_FIXED_{\rm it} + \eta_{5} CFO\_ACC \times_{{}} AB\_VAR_{\rm it} + \eta_{6} SIZE_{\rm it} + \eta_{7} LEV_{\rm it} + \eta_{8} ROA_{\rm it} + \eta_{9} FOREIGN_{\rm it} + \eta_{10} BTM_{\rm it} + \eta_{11} NOL_{\rm it} + \eta_{12} DEP_{\rm it} + \eta_{13} EQINC_{\rm it} + \eta_{14} CEO\_ACC_{\rm it} + \eta_{15} MILLS_{\rm it} + IND + \varepsilon_{\rm it} $$
(7)

Tables 11 and 12 report the regression results after controlling for the effect of accounting expert CEOs. The results of Tables 11 and 12 show that both of the coefficients on CFO_ACC and CFO_ACC × AB_VAR remain negative and significant (p-values = 0.082 and 0.047, respectively). The results are consistent with those in Tables 5 and 6 after controlling for the effect of the accounting expertise of CEOs.

Table 11 The impact of CFO with accounting expertise on tax aggressiveness-CEO with accounting expertise \( ETR_{\rm it} = \beta_{0} + \beta_{1} CFO\_ACC_{\rm it} + \beta_{2} SIZE_{\rm it} + \beta_{3} LEV_{\rm it} + \beta_{4} ROA_{\rm it} + \beta_{5} FOREIGN_{\rm it} + \beta_{6} BTM_{\rm it} + \beta_{7} NOL_{\rm it} + \beta_{8} DEP_{\rm it} + \beta_{9} EQINC_{\rm it} + \beta_{10} CEO\_ACC_{\rm it} + \beta_{11} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \quad (6) \)
Full size table
Table 12 The impact of compensation of CFO with accounting expertise on tax aggressiveness-CEO with accounting expertise \( ETR_{\rm it} = \eta_{0} + \eta_{1} CFO\_ACC_{\rm it} + \eta_{2} AB\_FIXED_{\rm it} + \eta_{3} AB\_VAR_{\rm it} + \eta_{4} CFO\_ACC \times_{{}} AB\_FIXED_{\rm it} + \eta_{5} CFO\_ACC \times_{{}} AB\_VAR_{\rm it} + \eta_{6} SIZE_{\rm it} + \eta_{7} LEV_{\rm it} + \eta_{8} ROA_{\rm it} + \eta_{9} FOREIGN_{\rm it} + \eta_{10} BTM_{\rm it} + \eta_{11} NOL_{\rm it} + \eta_{12} DEP_{\rm it} + \eta_{13} EQINC_{\rm it} + \eta_{14} CEO\_ACC_{\rm it} + \eta_{15} MILLS_{\rm it} + IND + \varepsilon_{\rm it} \quad (7) \)
Full size table

6 Conclusions

CFOs play an essential role in making a company’s financing decisions and driving operating performance. Corporate tax cost affects various business decisions and financial performance. Hence, CFOs with accounting expertise may contribute to corporate financial performance by better exploiting tax planning alternatives. Furthermore, companies may motivate their CFOs to improve their performance by designing efficient managerial compensation schemes, thereby linking compensation to performance. Prior studies have explored the effects of senior managers’ work experience and demographic characteristics on corporate strategies as well as operating results (Bantel and Jackson 1989; Bertrand and Schoar 2003; Dyreng et al. 2010). However, few prior studies have directly addressed the effect of CFOs’ accounting expertise on firms’ tax planning decisions. CFOs are responsible for a company’s financial reporting as well as tax compliance. Accounting expertise is closely related to tax decisions. The question of whether CFOs with accounting expertise affect a firm’s tax compliance behaviors remains unanswered. We investigate the effect of the accounting expertise of CFOs on corporate tax avoidance. Furthermore, we extend the previous research on the effect of executive incentive compensation on corporate tax avoidance by investigating the interactive effect between incentive compensation schemes and the accounting expertise of CFOs. Our results show that CFOs with accounting expertise are associated with lower ETRs, consistent with the notion that accounting expertise is closely related to corporate tax planning, and thus, CFOs with accounting expertise are better at exploring tax planning opportunities. Furthermore, we find that the abnormal variable pay of CFOs with accounting expertise is negatively associated with corporate ETRs, suggesting that compensation design, such as the use of variable pay, has an effect on motivating CFOs with accounting expertise to exploit more tax planning opportunities. Our results are robust to the alternative definition of CFO with accounting expertise and various additional tests. The findings of this paper extend prior research on the effects of senior managers’ demographic characteristics and compensation designs.