감사 전 재무제표에 대한 근사치로서의 잠정실적에 관한 연구 : 차기연도 감사시간과 감사보수를 중심으로

Approximating the Pre-Audit Financial Statements with the Difference between Preliminary-Actual Income : Based on Audit Hours and Audit Fees

잠정실적은 기업의 회계 프로세스를 통해 도출되는 지표인 만큼, 외부감사의 대상은 아니지만, 기업의 공시 성향, 회계 프로세스의 완성도, 재량적 회계처리 정도에 따라 영향을 받을 수 있다. 본 연구에서는 기업에서 작성하고 보고한 잠정실적과 실제실적의 차이가 차기연도의 감사시간과 감사보수로 측정되는 감사 리스크에 영향을 주는지를 검토하였다. 유가증권시장과 코스닥 시장에 상장된 기업들을 대상으로 하여 수행한 실증분석 결과, 잠정실적으로 보고되는 다섯 가지 이익 지표 중 잠정매출, 잠정 영업이익, 잠정세전이익, 그리고 잠정당기순이익과 실제실적의 차이는 차기연도의 감사시간 및 감사보수를 유의하게 증가시키는 것으로 나타났다. 또한 실적 차이의 부호를 나누어 검증한 결과 잠정실적이 실제실적에 비해 과소평가된 경우에 감사인들은 더 민감하게 반응하는 것으로 나타났다. 2년 연속으로 차이가 발생한 경우 차기 감사시간과 보수가 더 큰 폭으로 증가하였고, 비정상감사시간과 보수를 이용한 분석에서도 영업이익과 세전이익 차이는 동일한 결과를 나타냈다. 본 연구는 외부인에게 공개되지 않는 감사 전 재무제표의 측정치로서 기업의 자발적 공시에 해당하는 잠정실적 발표가 유의미하게 활용될 수 있음을 보였고, 기업 및 산업마다 상이한 영업방식이 잠정실적 정확도와 연관이 깊음을 보임으로써 실질적인 감사위험의 측정에 공헌한다.

Preliminary financial outcomes are produced by firms without external auditors’ assistance; hence they represent the firms’ accounting capabilities and the communication between the subject firms and auditors. In this paper we examine whether the difference between the preliminary outcome of the operation and the actual profit affect the audit risk measured by subsequent year’s audit hours/fees. Difference between preliminary and actual financial statements(PAD, henceforth) shows a disagreement between management and outside auditor. Due to the confidential nature of these, we use quarterly fair disclosure of earnings announcements as proxies of the preliminary reports. In Korea, firms are required to submit (consolidated) preliminary financial statements within 4(6) weeks prior to shareholders’ meeting. Significant portion of the announcement dates of quarterly earnings fair disclosure overlaps with the legal due dates of the preliminary reports, and thus the two results will be highly related. We define PAD in two ways; 1) if the d ifference fair d isclosure and actual outcome is less than 0 .1% of total amount, and 2) the actual difference. Our sample consists of Korean non-financial listed companies in the sample period of 2010 to 2018, and all the data is publicly available. Over the period there is increasing number of firms announcing various operating outcomes by fair disclosure. Among those outcomes there have been notably more positive PADs in sales – that is, actual outcome is smaller than preliminary outcome – suggesting that preliminary sales are frequently inflated by firms and then denied by auditors. The other outcomes show balanced portion of positive and negative PADs. Using five different operational outcome figures – sales, operating income, EBIT, net income, and net income in controlling interest - we find that firms with bigger difference between the preliminary and actual outcome are more likely to have higher audit risk. The average effect of an existence of PAD(sales) was 76 additional hours and 5.85 million KRW. After dichotomizing the sample by the signs of PADs, negative PADs showed significant and positive relations with the subsequent audit hours/fees. The results suggest that auditors perceive undervaluating firms to have more risky audit profiles. Although temporal difference of audit hours/fees prevents us from reverse causality, we apply three robustness tests to relieve potential endogeneity. First, for the firms with consecutive years of PAD the positive relation between PAD and the subsequent audit hours/fees is found to be stronger. Second, substituting the dependent variables with abnormal audit hours/fees yield results not different from the original tests. Third, we did sensitivity tests of PAD cutoff levels at 0.2%, 0.5%, 1%, 2%, and 5%. In every threshold levels the results remain consistent. Therefore we can conclude that the main results are not driven by the PAD cutoff levels. This paper suggests that the preliminary profit – although not observable from the outsiders – can be utilized as proxies for the preliminary reports. The results are robust for firm/industry characteristics, suggesting that preliminary-actual differences (PAD) can be used even after considering all the varying differences that affect the operations.