코스닥시장에서 감사인 규모선택에 영향을 미치는 피감사기업의 특성 및 감사품질에 관한 연구 - 성향점수 대응표본 설계에 의한 재검증

Many empirical researches studies whether Big4 auditors provide higher level of audit quality. Early researches concluded that the audit quality of Big 4 auditors is generally higher than Non-Big4. Because Big 4 auditors are more independent to their client and have better knowledge and greater reputation to protect. However, recent evidences are not clearly supporting the argument that auditor size is a proxy for audit quality. Also, recent studies concern that the client characteristics of Big 4 and non-Big4 are different and empirical findings of prior researches reflect them not auditor characteristics. Practically, non-Big4 auditor argues that they can provide comparable audit quality, especially in local markets for up to medium size client. Thus this study reexamine the relation between auditor size and its quality in KOSDAQ market. I restrict analyses to KOSDAQ market because it allows relatively greater degree of dispersion in auditor choice between Big 4 and non-Big 4. For the analyses, data are selected from 2005 to 2010. There are 4,049 firm-year observations in the full sample, of which 45.6% and 54.4% reflect Big 4 and non-Big 4 clients, respectively. The descriptive statistics for the full sample indicate that Big 4 and non-Big 4 auditors are significantly different. Big 4 clients are larger, more profitable, less financed, more invested by foreign investor and shows high incidence of IPO. I match the Big 4 and non-Big 4 samples with propensity-score reflected multi-client characteristics. I use propensity-score matching models to control client characteristics between Big 4 and non-Big 4 auditor by estimating the probability of selecting a Big4 with logit model. The results show that the client’s size, , debt ratio, foreign investor ownership and IPO variables are significant in logit model. I impose a caliper distance of 3 percent to match samples and obtain 2,442 firm-years for the propensity score matched sample. Mean difference test shows that Client characteristics are not different between two auditor groups. To examine whether the Big 4 effect can be attributed to specific client characteristics, I perform the regression analyses controlling for an extensive list of client and auditor variables. I measure performance-adjusted discretionary accruals as audit-quality proxy by the modified Jones model(Kothari et al, 2010). The results show that the Big4 auditors provide higher level of audit quality in full sample but superior quality of Big 4 vanishes in matched sample. The results show that the absolute value of discretionary accruals of Big 4 are not significantly different from those of Non-Big 4 in matched sample. I supports the concern that the result of prior research are influenced by client characteristics not auditor’s. The treatment effects of Big 4 auditors are insignificantly different from those of non-Big 4 auditors in our results. For the sensitivity test, I divide Non-Big 4 into middle and small Non-Big 4 groups. Middle Non-Big 4 includes the auditors that employ more that 50 CPAs and the small Non-Big 4 includes other Non-Big 4. There are no differences between Big 4 and middle Non-Big 4 and also between Big 4 and small Non Big 4. In the analysis between the individual Big 4 auditor and Non Big 4, I find consistent results. These results suggest that the findings of prior researchers which is not employing the methodologies correcting the self selection bias in their research design, are largely affected by client characteristics not auditor’s size. Practically, robust evidences that the audit quality of Non Big 4 is not different from that of Big 4, can be used usefully in selecting audit size for clients and in making regulation for the supervisory body. However I caution that these results are limited to KOSDAQ market and must be interpreted with regard to the limitation of propensity score matching model. The finding of this study should not be generalized for the clients excluded in the matching process.