초도감사의 감사보수 할인 현상에 대한 실증연구

Abstract

Initial audit fee discounting, frequently termed “low-balling”, means that auditors lowball the fees for initial-year audits. The practice of setting audit fees below costs on initial audit engagements has been the subject of a number of studies because of its potential impact on auditor independence. The Korean Institute of Certified Public Accountants(KICPA) Code of Professional Conduct says that the practice of setting initial audit fees significantly less than predecessor or other auditors' fees might be considered impairment of auditor independence. The investigation of initial-year audit fee cutting not only can be quite helpful for facilitating future studies on the effects of low-balling, but also can provide regulatory agencies with validating the regulations on audit fees. However, because most of Korean prior studies examined the initial audit fee discounting using the level variable model, they didn't get around the omitted variable bias. In this respect, using the difference variable model in order to resolve the problem of correlated omitted variable bias, we need to investigate whether initial audit fees are discounted by audit firms. Furthermore, after controlling the effect of auditor change on audit quality, we can afford to examine the fee cutting on initial audit engagements. To address our research questions, we empirically test whether there exists a significant reduction in audit fees on initial audit engagements in Korean audit market, and then examine whether the extent of discounting initial audit fees varies with types of auditor switches. In addition, we investigate the effect of initial audit contracts on audit time and managers' earnings management. Our final data consist of 1,436 Korean listed firm-year observations from the year 2000 to 2006 which are identified using sample selection criteria such as non-financing firms, fiscal year and data availability. We identify 159 firm-year observations which represent those switching auditors during the sampling period. Using the difference variable regression model, we investigate the fee cutting on initial audit engagements. In our model, the dependent variable is defined as the difference in audit fees between current audit fees and previous year's audit fees divided by the prior year's audit fees. The independent variable is a dummy variable indicating whether each observation is on the initial audit engagement or not; 1 for the firms if observation is on the initial audit engagements, 0 otherwise. In order to control the effects of variables that may affect the change in audit fees, we regard the changes of total assets, account receivables, inventory, foreign sales, return on investment and debt ratio as control variables. We also introduce industry and year dummy variables to control the difference in audit fees due to the type of industry in which firms are listed and the year in which observations are. We empirically test our first hypothesis and find that a significant reduction in audit fees for initial audit engagements exists, compared with audit fees of continuing audit engagements. Using a subgroup sample of clients switching their auditors, we classify types of auditor switches into the following four categories: auditor switches from Big to Big, from Big to non-Big, from non-Big to Big, and from non-Big to non-Big auditors. For firms switching between auditors of the same type such as auditor switches from Big to Big and from non-Big to non-Big auditors, we find that there is a significant fee discount on initial audit contracts. For firms switching from Big to non-Big auditors, we find that there exist significant fee discounts on initial audit engagements and the extent of fee discounts is significantly larger than that of firms switching between auditors of the same type. However, we cannot find a significant fee reduction on initial audit contracts for firms switching from non-Big to Big auditors. As robustness tests, we use the audit fees per hour and change in audit time as dependent variables instead of the difference in audit fees. The dummy variable which represents initial audit engagements has a significant negative association with audit fees per hour but a significant positive relationship with the difference in audit time. These results imply that although auditors increase audit time, auditors lowball the fees for initial-year audits. As additional robustness check, we find that the dummy variable which represents initial audit contracts has no significant associations with managers' earnings management. These robustness checks confirm our hypotheses and test results that there exists a significant reduction in fees on initial audit engagements in Korean audit market and the extent of discounting initial audit fees varies with types of auditor switches. While previous Korean literatures investigated the initial audit fee discounting using the level variable model, this study examines low-balling using the difference variable model, in result our study expands previous literatures investigating initial-year audit fee cutting in Korean audit market. Our findings also can help regulatory agencies validate the audit fee regulations.