臺大管理論叢 NTU Management Review VOL.30 NO.2

Do Government Initiatives and Reasons for Auditor-Client Realignment Influence Audit Fees and Audit Quality? Evidence from China 226 3.2 Research Methodology 3.2.1 Reasons to Disclose Auditor-Client Realignment Decisions This study first explores why certain firms decide to disclose their reasons for auditor-client realignment, while others choose not to do so. For this analysis, the study postulates that state ownership and the issuance of the government initiative would influence such a disclosure decision. According to Research Question 1, the variables of interest are STATE , POST_2007 , and STATE × POST_2007 . To examine this question, we follow Sankaraguruswamy and Whisenant (2004) and develop the following probit regression model: REASON_DISCLOSURE = β 0 + β 1 STATE + β 2 POST_2007 + β 3 STATE × POST_2007 + β 4 LNTA + β 5 D_LNTA + β 6 CFOTA + β 7 NEWFINANCING + β 8 TIE + β 9 DA + β 10 LEGALFORMLLP + β 11 LEGALFORMLLC + β 12 TOP10 + β 13 LOCAL + β 14 CMI + β 15 GDI + β 16 LEI + β 17 ISSUEB + β 18 ISSUEH + EXCHANGE + INDUSTRY + ε . (1) As indicated in Model (1), the dependent variable, REASON_DISCLOSURE , is an indicator variable. This variable is coded “1” if the firm discloses a reason for the auditor change and “0” otherwise. As shown in Model (1), the tested variables are STATE , POST_2007 , and STATE × POST_2007 . Like the dependent variable, STATE is also an indicator variable. It is coded “1” for firms controlled and operated by the state and “0” otherwise. Because we are interested in exploring the effects of the initiative to expand and strengthen accounting firm practices on the disclosure over the reasons for auditor- client realignment and because this initiative was launched in 2007, POST_2007 is treated as an indicator variable in Equation (1). This variable is coded “1” if observations are obtained from 2007 to 2014 and “0” if observations are obtained from 2002 to 2006. For the purpose of analysis, STATE × POST_2007 represents the interactive effect of state ownership and the initiative on the decisions to disclose the reasons for realignments. As for the control variables, we follow Sankaraguruswamy and Whisenant (2004). Moreover, we include several control variables in the regression models according to the findings by Huang et al. (2015) to explicitly take specific characteristics of Chinese-listed f i rms into account . As shown in Equa t ion (1) , these cont rol var i abl es are LEGALFORMLLP , LEGALFORMLLC , TOP10 , LOCAL , CMI , GDI , LEI , ISSUEB ,