회계투명성의 국가 간 비교 및 자본비용과의 상관관계에 대한 실증연구

Are Korean Firms as Opaque as Perceived? An International Comparison of Accounting Transparency

우리나라는 1997년 외환위기 이후 회계투명성 제고를 위해 다양한 노력을 경주해왔음에도 국제경영개발원이나 세계경제포럼 등 국제평가기구에 의한 회계투명성 순위는 여전히 (최)하위권에 머물고 있다. 그러나 이러한 결과는 응답자의 주관적 인식(perception)이 반영된 설문답변에 전적으로 기초한 것으로 객관적 평가라 할 수 없다. 이에 본 연구는 학술적으로 통용되는 다양한 계량분석모형을 통해 회계투명성에 대한 실증분석을 시도했다. 실증분석 결과 우리나라의 회계투명성은 「① 재량적 발생액 규모 기준으로 중위권(56개국중 22위) , ② 재량적 발생액 변동성 기준으로 상위권(55개국중 8위) , ③ 이익반응계수 기준으로 중상위권(51개국중 19위) 을 기록했으며, ④ 전염효과도 부재」하여 국제평가기구가 발표한 국가 회계투명성 순위처럼 (최)하위권은 아닌 것으로 나타났다. 이는 그간 도입·시행된 각종 선진 회계시스템으로 인해 평균적인 회계투명성이 개선된 결과로 판단된다. 한편, 미국과 우리나라를 대상으로 회계품질(투명성)과 자본비용 간의 상관관계를 분석한 결과 미국은 높은 회계품질이 자본조달 비용을 경제적으로 유의미한 수준으로 낮추는 데 기여하는 반면, 우리나라는 그러한 관계가 관측되지 않아 기업이 자발적으로 고품질의 회계정보를 생산하기 위한 사회·경제적 효익이 사실상 부재한 것으로 나타났다. 이러한 연구결과는 향후 회계정책의 패러다임이 투명한 회계정보의 생산(기업 역할)만을 강조해온 기존의 공급자적 관점 에서 기업의 능동적인 고품질 회계정보 생산을 적극적으로 유도(시장 역할)하는 수요자적 관점 으로 전환될 필요가 있음을 시사한다.

Despite all the efforts from practitioners, regulators, and accounting scholars in Korea, accounting transparency in Korea has been ranked at the lowest according to the previous management surveys by International Institute for Management Development (IMD) and World Economic Forum (WEF). In this paper, we question about the validity of those surveys because they might reflect merely biased perception of respondents on accounting practices in Korea. To address this question, we employ empirical proxies for accounting transparency and analyze a large set of archival data on financial statements of firms from 56 countries surrounding IMD survey respondents. We find that Korea does not appear to be in the most opaque group based on several empirical proxies for earnings quality. Specifically, Korea is ranked 1) at the 22nd in terms of the level discretionary accruals among 56 data-available countries, 2) at the 8th in terms of discretionary accruals volatility (standard deviation) among 55 data-available countries, and 3) at the 19th in terms of earnings-returns relations among 51 data-available countries. Furthermore, we do not find evidence that investors rely on information of industry peers in earnings valuation, indicating that the lack of firm-specific private information is not severe in the Korean stock market. In sum, we do not find evidence supporting that accounting transparency in Korea deserves the lowest ranking around the world. Rather, the empirical proxies for earnings quality consistently prove that the de facto quality of accounting transparency in Korea is above the median of the sample countries, consistent with what academicians and practitioners have argued for the last few years. However, further analyses reveal that costs of raising debt (or equity) are not related to the transparency constructs we used above for Korean firms, implying the lack of benefits of improving accounting transparency in Korea. By contrast, the transparency constructs show a strongly negative association with costs of debt (or equity) for U.S. firms, indicating that more transparent firms are well compensated by lower financing costs in the U.S. financial market. We thus suspect that the lack of benefits for accounting transparency in Korea (i.e., statistically no association between earnings quality and external financing costs) would partially provoke the unfavorable responses to survey questions about accounting practices in Korea. If a financial market does not appropriately appreciate the quality of earnings across firms and thus fails to reflect it in capital resource allocation, managers would consider all the regulatory inputs to improve accounting transparency as unnecessary cost sources that are not properly accompanied by corresponding benefits. A related concern is that such negative perception would drive out transparent firms from the market in the long run (a.k.a., adverse selection problem). Caveats are in order. To empirically address our research question, we cannot help making some assumptions which unavoidably sacrifice academic rigor. First, we compare earnings quality constructs from a single estimation model (i.e., a pooled regression) across countries. By doing so, we assume that the single model fits well to every economy, which is unlikely. Therefore, our earlier inferences would not hold valid to the extent that the fitness of the models significantly varies across countries. Second, we do not intend to explain why accounting transparency differs across countries. Prior studies document a set of country-level characteristics that explain cross-sectional variations of earnings quality across countries. However, including all of those country-level differences would not serve this paper because we are not interested in why earnings quality differs across countries but whether earnings quality differs (and whether Korea actually belongs to the lowest group).