The executive compensation implications of the tax component of earnings

Abstract

We extend prior research by examining the weight applied to earnings generated by changes in ETRs (i.e., the tax component of earnings) in determining CEO and CFO compensation. We examine both bonus and total compensation and find that the predicted relationships between compensation and the tax component of earnings are largely limited to bonus compensation. This is not surprising since bonus compensation represents an unambiguous link between contemporaneous performance and compensation, while equity compensation is in part determined by agency considerations. Our evidence suggests that both CEOs and CFOs are compensated for the tax component of earnings. We find that CEOs are rewarded equally for the tax component of earnings relative to other components of earnings, while CFOs are rewarded more for the tax component of earnings relative to other components of earnings. Additionally, the weight applied to the tax component of earnings when determining CFO bonus compensation is greater when; (1) the tax component of earnings does not appear to be related to earnings management; (2) ETRs decrease rather than increase, (3) the firm pays bonus based on after-tax earnings rather than pre-tax earnings, and (4) the firm is tax aggressive rather than non-tax aggressive. The variations in the weighting of the tax component of earnings for CFO bonus compensation noted above in combination with evidence that CEO bonus compensation is indifferent to ETR-related earnings versus other components of earnings, suggests that the tax component of earnings is a contractual component of CFO bonus compensation.

Appendices

Appendix 1

See Table 8.

Table 8 Control variable definitions

Appendix 2: tax-aggressiveness measures

To identify firms that have a propensity to engage in tax-aggressive behavior, we utilize three different proxies for tax aggressiveness. We discuss each of our measures below.

2.1 Permanent book-tax differences (PBTD)

Our first proxy is permanent book-tax differences (PBTD). PBTD measures aggressive tax reporting by focusing on permanent book-tax differences. Temporary book-tax differences frequently reflect book-tax differences that result from true differences in tax and financial reporting rules, rather than from actual tax planning (e.g., depreciation differences). Temporary book-tax differences may also reflect earnings management activity through the use of accruals, which is unrelated to tax planning. By removing these temporary differences PBTD produces a more accurate measure of true tax aggressiveness. We calculate PBTD as total book-tax differences less deferred tax expense grossed-up by the applicable federal statutory tax rate (Rego and Wilson 2008).

2.2 Discretionary permanent book-tax differences (DTAX)

Our second proxy for tax reporting aggressiveness is a measure of discretionary permanent book-tax differences (DTAX). Similar to PBTD, DTAX measures tax aggressiveness by focusing on permanent book-tax differences. DTAX improves upon the PBTD measure by including only discretionary permanent book-tax differences to measure tax aggressiveness. We calculate DTAX based on the work of Frank et al. (2009) who regress permanent book-tax differences on nondiscretionary items unrelated to tax planning that are known to cause permanent differences (see Eq. 8 below). We use the residuals (ε _it ) from this equation as our estimate of discretionary permanent differences:

$${\text{PERMDIFF}}_{it} = \, \beta_{0} + \, \beta_{1} {\text{INTANG}}_{it} + \, \beta_{2} {\text{UNCON}}_{it} + \, \beta_{3} {\text{MI}}_{it} + \, \beta_{4} {\text{CSTE}}_{it} + \, \beta_{5} \Delta {\text{NOL}}_{it} + \, \beta_{6} {\text{LAGPERM}}_{it} + \, \varepsilon_{it}$$

(8)

where PERMDIFF_i = total book-tax differences less temporary book-tax differences for firm i in year t, {BI _it − [(CFTE _it _CFOR _it )/STR _it ]} − (DTE _it /STR _it ); BI _it  = pre-tax book income (#170) for firm i in year t; CFTE _it  = current federal tax expense (#63) for firm i in year t; CFOR _it  = current foreign tax expense (#64) for firm i in year t; DTE _it  = deferred tax expense (#50) for firm i in year t; STR _t  = statutory tax rate in year t; INTANG _it  = goodwill and other intangibles (#33) for firm i in year t; UNCON _it  = income (loss) reported under the equity method (#55) for firm i in year t; MI _it  = income (loss) attributable to minority interest (#49) for firm i in year t; CSTE _it  = current state income tax expense (#173) for firm i in year t; ∆NOL _it  = change in net operating loss carryforwards (#52) for firm i in year t; LAGPERM _it  = one-year lagged PERMDIFF for firm i in year t; and ε _it  = discretionary permanent difference (DTAX _it ) for firm i in year t.¹¹

All variables above are as defined in Frank et al. (2009).

2.3 Predicted tax shelter firms (SHELTER)

Our final measure of tax aggressiveness is SHELTER, a measure which identifies likely tax shelter participants. Wilson (2009) developed the predictive model using a sample of identified tax shelter participants. We use SHELTER to predict which of our sample firms are currently engaged in tax shelter activity (see Eq. 9). A firm’s tax shelter score (SCORE) is obtained by inserting firm specific variables into the model and using the estimated coefficients from Table 5 in Frank et al. (2009) as follows:

$${\text{SCORE }} = - 4.30 \, + \, 6.63 \, \times {\text{BTD }} + \, - 1.72 \, \times {\text{ LEV }} + \, 0.66 \, \times {\text{ SIZE }} + \, 2.26 \, \times {\text{ ROA }} + \, 1.62 \, \times {\text{ FOR\_INCOME }} + \, 1.56 \, \times {\text{ R}}\& {\text{D}}$$

(9)

All variables are as defined in Wilson (2009). SIZE is the log of total assets (data 6). LEV is long-term debt (data 9) divided by total assets. ROA is pre-tax earnings (data 170) divided by total assets. FOR_INCOME is (data 273) divided by lagged total assets. R&D is (data 46) divided by lagged total assets. BTD is book income less taxable income scaled by lagged total assets. Book income is pre-tax income (data 170). Taxable income is calculated by grossing up the sum of the current federal tax expense (data 63) and the current foreign tax expense (data 64) and subtracting the change in NOL Carryforward (data 52). If the current federal tax expense is missing, then total current tax expense is calculated by subtracting deferred taxes (data 50), state income taxes (data 173) and other income taxes (data 211) from total income taxes (data 16). The predicted probability of tax sheltering is then calculated as follows:

$${\text{P}}\left( {\text{SHELTER}} \right) = \, e^{{({\text{SCORE}})}} /(1 \, + e^{{({\text{SCORE}})}} )$$