본 연구는 배당변화가 미래신용위험에 대한 예측정보를 포함하고 있는지를 살펴보았다. Miller and Modigliani(1961)는 완전자본시장을 가정할 때 배당이 기업가치와 무관함을 보였으나, 많은 실증연구에서는 이와 다른 결과를 보고하고 있다. 이러한 차이를 설명하는 중요한 이론 중 하나는 배당이 정보를 전달효과가 있으며, 더 나아가 역선택 위험에 따른 모니터링 비용을 낮출 수 있다는 것이다. 본 연구는 배당변화가 미래신용위험에 관한 정보를 포함하는지에 초점을 두었다. 한편으로는 경영자는 안정적으로 배당을 증가시킴으로써 장기적으로 정보위험에 따른 신용스프레드를 감소시킬 수 있다. 다른 한편으로는 경영자는 파산위험과 같은 나쁜 뉴스에 대한 전달을 지연하려고 하는데, 파산위험이 높다고 예상하는 경우 경영자는 현금보유정책의 일환으로 배당을 감소시킬 수 있다. 이러한 메커니즘에 의해 배당정보에는 미래신용위험에 관한정보를 포함한다는 예상이 가능하다. 실증분석결과, 배당변화를 통하여 향후 2년까지의 미래신용위험을 예측할 수 있었다. 특히, 이러한 예측가능성은 회계이익의 불투명성이 높은 경우와, 배당을 감소시킨 경우에 더 두드러지게 나타났다. 본 연구는 배당변화가 미래신용위험에 대한 정보를 포함하고 있는지 여부로 확장하였다는 의의가 있다. 본 연구결과는 배당의 정보전달의 효과가 정보위험을 낮추려는 경영자의 노력에 의해 나타날 수 있을 뿐만 아니라 파산위험을 낮추려는 현금보유정책의 일환으로도 나타날 수 있다는 것이다. 이 같은 연구결과는 채권투자자와 신용평가기관에게도 유용한 시사점을 제공할 것으로 기대된다.

This study examines whether changes in dividend effectively predict future credit risk. Miller and Modigliani (1961) show that a firm’s dividend policy is irrelevant to its value under a perfect capital market where there are no information asymmetry and transaction costs. But many empirical studies have reported inconsistent results with Miller and Modigliani (1961). One of the most appealing arguments for this inconsistency is that dividends convey managers’ superior information to investors. Further, it is argued that dividends reduce adverse selection costs that are caused by information asymmetry. Most analytical studies premise that managers deliberately use dividends policy as a signal to convey private information. But Brav et al. (2005) show that while most managers agree with that association exists between dividend changes and information, only some managers deliberately use dividends as a signaling device. This raises an issue that the mechanism of signaling and information contents of dividend changes need to be re-examined from a different perspective. On one hand, a stable increase in dividends could lower information risk due to information asymmetry between managers and bondholders. But managers`` efforts to lower information risk by increasing dividend may reflect future credit risk (Odders-White and Ready 2006, Cheng and Neamtiu 2009). On the other hand, a decrease in dividends could reflect the managers`` efforts to lower future bankruptcy risk. This is because managers should increase cash holdings by reducing dividends when they expect accessibility to capital market to be low. Given that managers tend to delay releasing bad news till a certain threshold period (Kothari et al. 2009), investors could infer bankruptcy risk from dividend changes. Based on above discussion, we hypothesize (i) that dividend changes predict changes in future credit risk; (ii) that the predictability of dividend changes for future credit risk is higher for firms with more opaque earnings than those with less opaque earnings; and (iii) dividend decreases predict future credit risk better than no-changes or increases in dividends. The final sample consists of 1,475 (1,147) firm-years that are listed in Korea Exchange for the period 1999~2015 for which dividend changes and one-year-ahead (two-year-ahead) bond ratings are available. To be included in the final sample, a firm-year should be in non-banking industries, with December fiscal year end, and pay dividends for two consecutive years. Also, consistent with prior studies, we delete a firm-year for which dividend growth exceeds 500% to mitigate unusual effect of extremely low dividends for previous year. The results of this study generally support our hypotheses. First, dividend changes significantly predict future credit risk that is proxied by one-year-ahead or two-year-ahead bond ratings. Second, the predictability of dividend changes for future credit risk is more pronounced for firms with more opaque earnings than for firms with less opaque earnings. This suggests that the signaling effect of dividends is generally stronger as earnings become more opaque and that opaque-earnings-firms are less likely to disclose voluntarily bankruptcy risk to bondholders. Third, the predictability of dividend changes for future credit risk is more pronounced for dividend-decreasing firms than for other firms. This shows that the role of dividend changes in predicting future credit risk depends on their signs. In other word, although non-negative dividend changes could lead to only adverse selection risk, negative dividend changes could inform bankruptcy risk as well as adverse selection risk. We believe that the result in this study deepens our understanding about the role of dividend changes in conveying relevant information about future credit risk. Further, such a role of dividend changes is not uniform across direction of dividend changes. Overall our finding will provide useful implication to bondholder as well as credit agency.