4.1 Sample
I obtain the main data for this study from several sources. I obtain the audit fee data from the Audit Analytics database, compensation data from ExecuComp, accounting data from Compustat.I also hand-collected detailed data of U.S. CEO compensations from proxy statements (DEF 14A) on the various aspects of stock,option, and cash awards of fiscal years 2010 and 2011.In this thesis, I focus only on performance-contingent equity awards that have absolute performance goals attached to them in order to simplify the stud y.
4.2 Research Designs
I adopt the audit fee model developed by Simunic (1980) and follow prior literatures (e.g. Chen et al., 2013) in selecting control variables.
LAFEE = 𝛽𝛽0+ 𝛽𝛽1ST + 𝛾𝛾1SIZE + 𝛾𝛾2SEG + 𝛾𝛾3FOREIGN + 𝛾𝛾4GROWTH + 𝛾𝛾5LEV +𝛾𝛾6ROA + 𝛾𝛾7VAROA+ 𝛾𝛾8LOSS + 𝛾𝛾9INVREC + 𝛾𝛾10CURRENT
+𝛾𝛾11QUICK + 𝛾𝛾12DA + 𝛾𝛾13RET + 𝛾𝛾14MA + 𝛾𝛾15ZSCORE + 𝛾𝛾16BIG4 + ε (1) Where LAFEE is the natural logarithm of audit fees.ST is an indicator variable equals to 1 if there is any performance-contingent equity award that has performance period shorter than 12 months, and 0 otherwise. The detailed definitions of control variables are shown in Appendix 1.
Followed by Bizjak et al (2013), ST is an indicator variable of interest. It takes the value one if the company has any of the performance-contingent equity award with short performance period, which means the periodwithin 12 months, and zero
26
otherwise.In Panel B of Figure 1 and Figure 2, the performance period is three year for 3M, and one year for Cameron. Therefore, the value equals to zero for 3M, but one for Cameron.
The dependent variable is LAFEE, the natural logarithm of audit fees. Consistent with prior research, I express the dependent variable in log form to mitigate the effects of nonlinear relation (Hay et al., 2006).As prior literatures indicated, a short-term performance-contingent award may induce executives to manage earnings in order to maximize theirown benefits. I expect that there is a positive relation between audit fee and ST because from the view point of auditor, the risk of misreporting should be higher if there is an award that has longer performance period.
Prior literatures indicate that the audit fees vary with client characteristics, including client size, audit complexity, client risk and audit characteristics. Regarding client size, I use the natural logarithm of total assets (SIZE). As audit effort is expected to increase in the scale of the client, the predicted sign is positive. To capture audit complexity, I include the number of segments (SEG), the percentage of foreign sales relative to total sales (FOREIGN), and investment opportunity measured by the market-to-book ratio (GROWTH). As audit effort is expected to be higher due to more complex or international operation, I expect the coefficients of SEG, FOREIGN and
GROWTH to be positive. Next, I include control variables to capture firm risk,
including leverage ratio measured by long-term debt divided by total assets (LEV). As more leveraged firms face greater financing constraints, the predicted sign is positive.Return on assets (ROA), the variance of ROA (VAROA), and a distress indicator variable capturing negative net income (LOSS), are predicted negative and positive signs respectively. Receivable and inventory intensity measured by receivables and inventory ratio (INVREC), as these may be subject to higher risks of error, the predicted sign is
27
positive. Current ratio (CURRENT) and quick ratio (QUICK) are predicted negative signs. As more discretionary accruals (DA) indicates a higher risk of misreporting, I expect a positive sign.RET is the firm’s stock return. Organizational change is measured by an indicator variable equals one if there is a merger by the firm during the year (MA).
The Altman Z-score (Z SCORE) is a measure of the probability of bankruptcy, the lower value indicates the greater probability of financial distress. I expect that MA and Z
SCORE to be positively and negatively related to audit fees, respectively.
Another group of control variables include audit characteristics. I include Big N (BIG4) to capture the quality or reputational effects of larger audit firms.
Prior studies (Carcello et al., 2002; Abbott et al., 2003) find that companies with stronger corporate governance pay higher audit fees, since better-governed firms care more about financial reporting quality and thus are willing to purchase more audit services. I also control for firms’ corporate governance characteristics and ownership structure as follow. CEO dual chair (DUALITY) is an indicator variable equals one if CEO also serves as the chairman of the board.CEOOWN is CEO ownership, which means the proportion of firm’s outstanding shares held by CEO. BOARDSIZEis the natural logarithm of the number of board size.IND is the proportion of independence directors on the board.AUDEXPERT is the proportion of financial expertise in audit committee.
Further, I substitute the natural logarithm of the number of performance measures of the performance-contingent equity awards that have performance periods in future years (Log(Num_LT)) for the ST indicator variable to find evidence on the influence of audit fees. When it comes to the year that the performance targets are evaluated, it also provides an incentive, as strong as short-term measures, to manage earnings in order to trigger payout of the award or increasing the payout in the incentive zone, which may
28
influence auditors’ risk assessing towards a company (Bizjak et al., 2013). When there are more number of long-term performance measures, it represents that the CEO compensation of the firm based more performance measures overlong horizon. In addition, the centralization on a certain performance measure is much easier for CEO to manipulate financial reports for achieving the performance target. On the contrary, it is more difficult to manage earnings to reach those targets if the performance-contingent equity award is based on different measures. Therefore, I expect the number of the long-term performance measures in CEO’s performance-contingent equity awards has negative association with audit fees.
LAFEE = 𝛽𝛽0+ 𝛽𝛽1Log(Num_LT) + 𝛾𝛾1SIZE + 𝛾𝛾2SEG + 𝛾𝛾3FOREIGN + 𝛾𝛾4GROWTH +𝛾𝛾5LEV + 𝛾𝛾6ROA + 𝛾𝛾7VAROA+ 𝛾𝛾8LOSS + 𝛾𝛾9INVREC + 𝛾𝛾10CURRENT +𝛾𝛾11QUICK + 𝛾𝛾12DA + 𝛾𝛾13RET + 𝛾𝛾14MA + 𝛾𝛾15ZSCORE + 𝛾𝛾16BIG4 + ε
(2) In the Appendix 3-3, for instance, each of three measures are separated into three rowsfor different years. The column of vest high represents the end of the performance period from the grant date. Thereare 3 measures (rows), which have the value of vest high smaller than 12, will be evaluated in current year. Therefore, the value of this variable is log(6) because only 6 measures left are over long performance horizon.
29