최근 들어 세계 각국에서 과도한 비감사보수가 감사품질을 저해하느냐에 대한 연구가 다수 수행되어 왔다. 이는 Enron 사건과 관련된 Arthur Andersen 회계법인의 경우처럼, 회계법인들이 고객기업에게 감사 및 비감사 서비스를 동시에 제공하면서 막대한 보수를 받고 있기 때문에, 이러한 행동이 감사인의 행동에 영향을 미칠 가능성이 존재하기 때문이다. 선행연구들이 대부분 연구의 범위를 비감사보수로 제한한 것에 비해 본 연구에서는 연구대상을 비감사보수와 감사보수로 확대하여, 이들과 고객기업의 재무제표에 반영된 보수주의의 정도로 측정된 감사품질 사이에 어떠한 관계가 존재하는지 분석하였다. 보수주의적 회계처리란 비용(또는 손실)은 발생이 예상될 때 즉시 인식하고, 수익(또는 이익)에 대하여는 보다 높은 수준의 검증가능성을 요구하여 그 인식을 이연시키는 것을 말한다. 만약 과다한 감사보수 또는 비감사보수가 감사인의 감사품질을 저해한다면 고객기업의 보수주의 정도는 감소될 것이다. 본 연구에서는 이러한 예측을 2000년부터 2005년까지의 최대 3,482개 기업-연도 자료를 이용하여 실증적으로 분석하였다. 분석에 사용된 보수주의 모형은 Ball and Shivakumar (2005, 2006)가 최근에 개발하여 사용한 기존의 여러 모형들을 종합한 모형이다. 분석결과 실제 감사보수 또는 비감사보수가 적정 보수를 초과하는 정도로 계산된 비정상 감사보수 또는 비정상 비감사보수가 증가 할수록 고객기업의 보수주의 정도가 감소되었다. 이러한 결과는 여러 다양한 민감도 분석에서도 강건성을 가지고 있었다. 즉 과도한 보수를 지급하는 고객기업들의 경우 감사품질이 저해된다는 발견이다. 또한 과도한 보수 수준만이 아니라 감사보수나 비감사보수 자체도 보수주의 수준과 유의적인 관계를 가지고 있었다. 선행연구인 Ruddock et al.(2006)이 Basu(1997)의 모형을 사용하여 분석한 결과 비감사 보수수준은 고객기업의 보수주의 정도와 유의적인 관련성이 있다는 귀무가설을 기가하지 못한바 있다. 본 연구에서도 이와 동일하게 Basu(1997)의 방법론을 이용해서 분석한 결과, 역시 유의적인 발견을 할 수 없었다. 따라서 본 연구의 주 발견은 Basu(1997)의 방법론보다 더 우수한 보수주의 측정치인 Ball and Shivakumar(2005, 2006)의 방법론을 사용했기 때문으로 판단된다. 이러한 발견은 규제기관이나 실무계, 투자자 및 학계에 많은 시사점을 제공해 주고 있다.
This study examines the association between abnormal audit and non-audit fees and the audit quality measured by the magnitude of the accounting conservatism. Abnormal audit or non-audit fees imply the level of fees in excess of the normal level of fees. If auditors receive excessive level of fees - either audit or non-audit fees, auditors may impair audit quality to maintain close relationship with clients to receive the fees continuously. Since the Enron incident and subsequent collapse of Arthur Andersen which occurred at year 2001, regulators have great concern if the excessive non-audit service impairs audit quality or financial reporting quality. Many studies also have investigated this issue either empirically or theoretically. Empirical studies use various proxy for the audit quality but the results up to now are mixed at best. However, not many studies examines the role of excessive audit fees in this context. This study extends the prior research and examines both audit and non-audit fees with respect to the magnitude of accounting conservatism of the client firms. If excessive fees impair the audit quality, auditors may allow client firms to engage in aggressive accounting practices which result in lesser magnitude of the conservatism. In this study, the conservatism implies the degree that how quickly the financial statements reflect economic losses compared with economic gains. This is the concept that first introduced by Basu(1997) and adopted by various accounting research subsequently. But surprisingly, none of the prior studies try to link the conservatism measures with respect to audit fees. Like excessive non-audit fees impair auditor independence, excessive audit fees could impair the auditor independence in the same way. This study is intended to explore this possibility. In addition, after the passage of Sarbanes-Oxley Act in 2002, the provision of non-audit service to clients by auditors are strictly prohibited or throughly monitored in the U.S. Similarly, Financial Supervisory Service of Korea (equivalent to the Securities and Exchange Commission of the U.S.) also regulate auditors' provision of the non-audit service. Thus, auditors could be very careful when they provide non-audit service to their audit clients. Media and academics also have greate concern on the effect of non-audit service. However, the concern on audit service has not been changed much. We expect that this study fills the void in this area. Moreover, because we do not know on ex ante basis if total fees or excess fees which are defined by the difference between total fees and normal level of fees are related to the auditor’s incntive which could influence the audit quality, we use both measures in our analyses. This study uses 3,483 firm-year observations collected over 2000 to 2005 period. We collect all the non-financial industry observations as long as all the necessary data are available and the firm has December fiscal year-end. In the analyses, the magnitude of the conservatism is measured by Ball and Shivakumar’s (2005, 2006) methods. The measure was developed most recently and known as the superior method to Basu’s (1997) measure of conservatism which has been used mostly frequently in prior studies. Specifically, the conservatism measure developed by Ball and Shivakumar(2005, 2006) is based on three different models: cash model, Dichev and Dechow(2002) model, and Jones(1991) model. We use all of those three models in our analyses in order to show the robustness of our empirical findings. To perform empirical analyses, we follow the following procedures. First, we measure excessive level of audit and non-audit fees by estimating normal level of fees. To do so, we regress the determinants of audit and non-audit fees to find out the coefficients on each determinants. Using the estimated models, we calculate the normal level of the fees by plugging the values of each variables into the model. The abnormal fees are measured by the difference between actual fees minus the estimated normal fees for both audit and non-audit fees, respectively. Second, we add the abnormal fees in the conservatism models to see if the level of conservatism changes depending on the level of abnormal fees. The empirical results reveal that the financial statements become less conservative as both abnormal audit or non-audit fees increase. These results are robust in various sensitivity analyses. In addition, the level of audit and non-audit fees themselves are also negatively associated with the degree of conservatism. These results imply that higher level of both audit and non-audit fees could impair audit quality or financial reporting quality. Because the findings in this study are different from those in Ruddock et al.’s (2006) study, which show that the non-audit fees are not associated with Basu’s (1997) measrue of conservatism, we try to replicate Ruddock et al.’s (2006) finding. We find that the Basu’s (1997) conservatism measure is significantly associated with neither audit nor non- audit fees in our sample firms. These inconsistency suggests that our initial significant results are due to the use of Ball and Shivakumar’s (2005, 2006) measure of conservatism, which is known to be a better measure than Basu’s (1997) measure. These findings provide several valuable insights into the effect of audit and non- audit fees toward regulators, practitioners, investors, as well as academics.