본 연구는 Hutton et al.(2009)에서 제안된 재량적 발생액의 3년간 시계열적 변동성으로 측정된 재무보고의 불투명성 측정치가 이익예측의 정확성과 편의로 측정되는 재무분석가의 이익예측오차에 미치는 영향을 실증적으로 분석하였다. 나아가 앞서의 관계가 시장유형(KOSPI vs. KOSDAQ)에 따라, 또는 감사인 규모로 측정된 감사품질(Big4 vs. non-Big 4 감사인)에 따라, 그리고 재무분석가의 정보환경의 변화가 있었던 IFRS 의무도입 전후기간에 따라 차이가 있는지에 대해서도 살펴보았다. 분석을 위해 본 연구는 Hutton et al.(2009)의 방법에 따라 과거 3년간 연도별 재량적 발생액(DA)의 절대값의 합으로 측정되는 재무보고의 불투명성 측정치(이하 OPAQUE1)와 전규안과 박종일(2017)에서 제안한 과거 3년간 DA의 표준편차로 측정된 재무보고의 불투명성 측정치(OPAQUE2)를 이용하여 재무분석가의 이익예측오차와의 관계를 분석하였다. 이를 분석하는데 있어 비교목적으로 DA의 결과도 살펴보았다. 본 연구는 금융업을 제외한 12월 결산법인 중 분석에 이용가능 했던 상장기업을 대상으로, 분석기간은 관심변수를 기준으로 2001년부터 2015년까지 최종표본 9,677개 기업/연 자료를 분석하였다. 실증결과는 다음과 같다. 첫째, 일정 변수를 통제한 후에도 재무보고의 불투명성 측정치가 높을수록 재무분석가의 이익예측의 정확성이 떨어지고, 또한 낙관적 편의의 성향이 있는 것으로 나타났다. 이러한 결과는 OPAQUE1과 OPAQUE2 모두 일치된 결과로 나타났다. 이와 달리, 선행연구에서 보편적으로 이용되었던 Kothari et al.(2005)의 방법에 따라 측정된 DA는 재무분석가의 이익예측오차와는 통계적으로 유의한 관계가 나타나지 않았다. 둘째, 전체표본을 다시 시장유형에 따라, 감사인 규모에 따라, 그리고 IFRS 의무도입 전후기간에 따라 표본을 나누어 분석하더라도 재무보고의 불투명성 측정치와 이익예측오차 간에 양(+)의 관계는 모두 일관되게 앞서와 일치된 결과로 나타났다. 이상을 종합하면, 본 연구는 3년간 DA의 다기간으로 측정된 재무보고의 불투명성이 높을 때 재무분석가들이 발표하는 이익예측오차가 증가된다는 것을 보여주었다는데 의미가 있다. 특히, 한해 연도로 측정된 DA와 달리, 3년간 DA의 반전효과가 고려된 재무보고의 불투명성 측정치는 재무분석가의 이익예측오차와 양(+)의 관계로 나타난다는 본 연구의 발견은 재무분석가의 이익예측오차를 통해 살펴본 재무보고의 불투명성과 관련한 대용치(surrogate)의 경우 DA보다는 OPAQUE 측정치가 재무분석가들에게 정보의 불확실성을 더 증가시킨다는 것을 보여준다. 따라서 본 연구의 발견은 이익의 질과 재무분석가의 이익예측오차를 다룬 관련연구에 추가적인 새로운 증거를 제공할 것으로 기대된다. 아울러 본 연구결과는 학계뿐만 아니라 재무분석가의 이익예측치를 이용하는 자본시장의 투자자 및 실무계, 또한 재무보고의 질에 관심이 있는 규제기관에게 재무보고의 불투명성이 높을 때 재무분석가의 이익예측치에 어떤 체계적인 영향을 미치고 있는지와 관련한 전반적인 이해에도 유용한 정보를 더불어 재공해 줄 것으로 예상된다.

This study investigates whether the relation between the financial reporting opacity measured by the OPAQUE measure of Hutton et al. (2009) and analyst’s earnings forecast errors measured as forecasting accuracy and bias. In particular, we test a measure of financial reporting opacity for individual firms based on an indicator of earnings management, specifically, in the prior three year’ moving sum of the absolute value of discretionary accruals as developed by Hutton et al. (2009) and standard deviation of discretionary accruals as suggested by Jeon and Park (2016). In additional analysis, we partition the sample into KOSPI versus KOSDAQ listed firms and test whether the relation between financial reporting opacity and analyst’s earnings forecast errors is different for market type (KOSPI versus KOSDAQ samples). Furthermore, we separate the sample into Big 4 auditor versus non-Big 4 auditors samples and test whether the relation between financial reporting opacity and analyst’s earnings forecast errors is different for audit quality (high audit quality versus low audit quality samples). Moreover, we divided the sample into pre-IFRS period versus post-IFRS period and test whether the relation between financial reporting opacity and analyst’s earnings forecast errors has changed since the adoption of IFRS. In the paper, we further consider the empirical link between financial reporting opacity and analyst’s earnings forecast errors. Thus, our approach differs from the existing literature with a one-year value of discretionary accruals to capture the earnings management. This is, we use a three-year moving sum to capture the multi-year effects of earnings management and because the moving sum is more likely to reflect an underlying policy of the firm to manage earnings (Hutton et al. 2009; Jeon and Park 2017). In the study, we expect analysts’ earnings forecasts inaccurate and to be optimistically biased when financial reporting opacity is increasing. For analysis, our empirical tests employ two firm-level proxies of financial reporting opacity, that is, accrual-based measure of opacity. The first proxy of this study is the measure of opacity in financial reports (hereafter OPAQUE1) as developed by Hutton et al. (2009). We use OPAQUE1 is the measure of financial reporting opacity of Hutton et al. (2009) based on earnings management, calculated as the prior three years’ moving sum of the absolute value of discretionary accruals. Following Hutton et al. (2009), discretionary accruals are measured using the modified Jones model by Dechow et al. (1995). The second proxy of this study is the measure of financial reports opacity (hereafter OPAQUE2) as suggested by Jeon and Park (2016), which is the standard deviation of discretionary accruals, which is calculated over years t-2 through t. We use a three-year moving sum of absolute value of discretionary accruals or standard deviation of discretionary accruals (instead of a one-year value) to capture the multi-year effects of earnings management. Therefore, these measure intends to capture both the abnormally high accruals in the year of overstatement and the subsequent reversal of prior accrual overstatements. For example, Dechow et al. (1996) show that firms subjected to enforcement actions by the SEC generally manipulate reported earnings from one to three years before being detected and the overstated accruals of these firms typically reverse fairly quickly. Our sample covers KOSPI and KOSDAQ listed firms in non-financial industries based on the test variable from 2001 to 2015 (based on the dependent variable from 2002 to 2016), thus we use 9,677 firm-year observations in Korean Stock Exchange Market. Our main findings are as follows. First, after controlling for several factors that affect analyst’ forecast errors, we find that a significantly positive association between analyst’ earnings forecast error and financial reporting opacity (both OPAQUE1 and OPAQUE2 proxies), consistent with our predictions. This result implies that firms with higher OPAQUE is associated with less accurate analyst’ earnings forecasts, and more optimistically biased forecasts. Whereas, we find no evidence that a significantly positive association between analyst’ earnings forecast error and the discretionary accruals of a one-year value. Second, we demonstrate that our measure of financial reporting opacity reliably predicts both analyst’ earnings forecasting accuracy and bias. This is, our results are robust to a variety of sensitivity checks such as an alternative specification. For examples, when we classify sample into KOSPI versus KOSDAQ listed firms, when we divided sample into Big 4 auditors versus non-Big 4 auditors, and when we partition the sample into pre- and post-IFRS adoption period, there are a positive and significant relation between analyst’ forecast error and financial reporting opacity, regardless of KOSPI versus KOSDAQ samples, or Big 4 auditors versus non-Big 4 auditors samples, and pre-IFRS period versus post-IFRS period. Therefore, our conclusions are not sensitive to market type, audit quality, and according to before and during IFRS adoption period, we find similar results. Overall, these empirical evidence suggests that the financial analyst perceive financial reporting opacity as a information risk-increasing factor. Using two proxies for financial reporting opacity, we provide novel evidence that analyst’ earnings forecast error increases significantly with financial reporting opacity, which is constructed as the multi-year effects of discretionary accruals. Whereas analyst’ earnings forecast error not increases significantly with the level of discretionary accruals in the short-run. Thus, we confirm that our firm-specific measure of opacity (both OPAQUE1 and OPAQUE2), which prior three-year moving sum or standard deviation of discretionary accruals is a reliable predicator of information risk, compared with the level of discretionary accruals in the one-year, to detecting earnings management. Therefore, the results of this study may provide useful information to academics as well as investors and regulatory bodies that higher financial reporting opacity are associated with less accurate analyst’ earnings forecasts, and more optimistically biased forecasts. Furthermore, considering the importance of analyst role in decision making of investors on the capital market, our results suggests that there is a problem with the accuracy of analyst’s earnings forecast, i.e., firms with opaque financial reports. In addition to providing the first empirical examination of the factors that connect financial reporting opacity to analyst’s earnings forecast errors, the results have implications for earnings management and analyst’s forecast errors studies in prior research.