본 연구는 비상장기업을 대상으로 Burgstahler and Dichev(1997, 이하 B&D)의 연구에서 제시된 이익조정이 의심되는 기업(적자회피 및 이익감소회피)과 그렇지 않은 기업 간의 타인자본비용(부채조달비용)에 차이가 있는지를 분석하였다. 즉 이익조정이 의심되는 기업에 대한 채권투자자의 반응을 살펴보았다. 이와 더불어 적자회피와 이익감소회피 기업 간의 타인 자본비용에 미치는 효과 면에서 상대적 중요성에 차이가 있는지를 살펴보았다. 이를 위하여 본 연구는 타인자본비용의 대용변수로 부채차입이자율과 차입이자율 스프레드를 이용하였고(Jiang 2008), 이익조정이 의심되는 기업은 B&D(1997)의 이익분포의 횡단면적 불연속성의 특성을 통해 파악되는 적자회피와 이익감소회피의 구간을 설정한 후 분석했다. 분석기간은 2004년부터 2009년까지이고 최종표본은 분석에 이용가능 했던 40,362개 기업/연 자료였다. 연구결과는 다음과 같다. 첫째, 타인자본비용에 영향을 미치는 일정 변수를 통제한 후에도 이익 조정이 의심되는 적자회피 또는 이익감소회피기업은 각각 그렇지 않은 경우와 비교할 때 타인 자본비용이 더 높은 것으로 나타났다. 이런 결과는 OLS 외에도 횡단면-시계열적 종속성 문제가 조정된 Newey and West(1987)와 군집성을 통제한 Clustering 검증결과 모두 강건성을 가지고 있었다. 그러나 Fixed effect를 고려한 분석에서는 극단치가 조정된 소수의 순위등급변수로 측정한 경우 적자회피기업만 유의한 결과를 보였다. 따라서 분석결과에서 고정효과까지 통제하면 채권투자자들은 적자회피기업만을 이익조정이 의심되는 기업으로 간주하여 이들 비상 장기업은 채권시장에서 부채조달시 패널티를 받는 것으로 나타났다. 둘째, 추가분석 결과에 따르면 적자회피기업과 타인자본비용 간의 양(+)의 관계는 이익감소회피기업의 경우와 비교할 때 통계적으로 유의한 차이를 보이는 것으로 나타났다. 즉 채권투자자들은 이익조정이 의심되는 기업으로 이익감소회피보다 적자회피기업에 대해 더 가중치를 두고 대출이자율 결정에 반영한다는 결과였다. 이는 채권시장에서 이익감소회피보다 적자회피기업의 정보위험을 더 높게 평가하고 있음을 나타낸다. 본 연구는 그동안 연구가 부족했던 비상장기업을 중심으로 목표이익을 달성할 유인이 있는 기업들에 대해 타인자본비용 관점에서 어떤 체계적인 관련성이 있는지를 보여주었다는데 의미가 있다. 또한 본 연구의 발견은 이익조정 관련연구에 추가적 공헌을 제공할 뿐만 아니라, 이익조정에 관심 있는 실무계 및 규제기관에게도 비상장기업들에서 부채차입을 할 때 목표이익에 근접된 이익을 보고하는 경영자의 보고전략이 채권시장에서 어떻게 평가되고 있는지를 이해하는 데 있어서도 도움을 준다는 점에서 유익한 시사점을 제공한다.

This study examines how the earnings management behavior of nonlisted firms affects the level of firms` cost of debt. More specifically we investigate whether the creditors price when the borrowers report the small positive earnings or the small earnings increases as identified in Burgstahler and Dichev (B&D, 1997). B&D document that firms manage their reported earnings to avoid the earnings decreases or the losses, thus increasing the incidence of producing the small earnings increases and the small positive earnings. The purpose of this study is to reexamine the earnings management perspective of B&D targeting the non-listed firms in the debt market in Korea. We explore whether the creditors penalize the non-listed firms reporting the small earnings increases and the small earnings, thus increasing the costs of debt. In addition, we compare the two groups: the firms having small positive earnings and the firms having small earnings increases, and test whether there are any significant differences in the effect on the levels of cost of debt between the two different earnings management area. Following the prior studies (e.g., Jiang 2008, etc.), we employ the average interest rates and the average spread of interest rates as proxies for the cost of debt. As B&D show the cross-sectional discontinuity of earnings distribution around zero, we identify the regions with small earnings increases and small positive earnings. This study uses the total 40,362 firm-year observations for the period from 2004 to 2009. Our empirical results present the evidence that when the firms reporting small positive earnings and small earnings increases show significantly higher cost of debt respectively, than firms not falling into the regions, after controlling for other correlated variables. It suggests that creditors perceive that firms report the small positive earnings in order to avoid losses and the small earnings increases to avoid earnings decreases. In other words, when firms report the B&D`s earnings regions the creditors regard it as a result of managers` opportunistic behavior, and thus penalize by increasing the costs of debt. This OLS results are qualitatively similar to the results under the Newey and West (1984)`s test controlling for the heteroskedasticity and auto-correlation problems as well as the clustering methodology to control for time-series dependency. Despite the robustness, however, when fixed effects are considered, the results of non-listed firms with small earnings increase show not significant but those of firms with small positive earnings are still significant. Furthermore, we find that the firms reporting small positive earnings to avoid losses show significantly higher cost of debt effect than the firms reporting small earnings increases to avoid earnings decreases. It implies that creditors are more concerned about the firms reporting losses than the firms reporting negative earnings changes, and thus more severely penalize the firms reporting the small positive earnings to avoid losses than the firms reporting small earnings increases to avoid earnings decrease. In other words, creditors value more highly information risk of firms reporting small positive earnings than firms reporting small earnings increase. In sum, we conclude that the earnings management regions identified in the crosssectional discontinuity of earnings distribution around zero presented by B&D have negative impact on the non-listed firms` costs of debt compared to the firms not falling into the regions. Creditors believe that earnings in the two regions are manipulated due to the managers` opportunism rather than earnings are highly realizable in the future. Thus, creditors impose a penalty rather than give a reward in terms of the costs of borrowings. In this regard, our findings suggest that although the firms opportunistically manage earnings in order to report positive earnings and earnings changes and thus anticipate any other substantial benefits from the earnings management behavior, it is highly probable that they rather suffer from the increases in the cost of debts. On the other hand, the debt market perceives the firms reporting small positive earnings or earnings changes as those which manipulate earnings to avoid losses and earnings decreases. This study contributes to the extant literature in several ways. First, it is the first study to examine the earnings management behaviors identified by B&D (1997) of the non-listed firms, more specifically the effect of earnings management regions of B&D on the level of cost of debt. There exist not many prior studies regarding the non-listed private firms in Korea despite the relative importance of non-listed firms in the capital market in Korea. Second, this study tests the efficiency hypothesis in the debt market as a joint hypothesis. It is notable that the earnings management behavior of non-listed firms documented in this study is likely to adversely affect the efficient resource allocation in the capital market. Thus, our findings provide a crucial insight to various stakeholders of non-listed firms in Korea, such as managers, banks, financial analysts, auditors, government etc., in the sense that the earnings management behaviors of non-listed firms identified by B&D are priced in the private debt market.