以非線性及多期盈餘不對稱反應重新檢驗會計穩健性與公司特質之關聯

Abstract

穩健性影響會計理論及實務良多，Basu(1997)提出穩健性存在於盈餘對壞消息的認列門檻較好消息低，基於價格領先盈餘的假設，將當期正(負)的股價報酬率作為好(壞)消息的代理變數，利用盈餘對資訊的不對稱反應來測試盈餘是否存在穩健性，後續許多文獻亦利用盈餘對資訊的不對稱反應比較不同公司或年代的穩健程度。Lin和Liu (2011)根據會計穩健的性質，認為盈餘的不對稱性在好或壞消息程度較大的時候會更加明顯(非線性盈餘不對稱反應)，而消息公開後盈餘的不對稱性會隨時間經過減弱(多期盈餘不對稱反應)；除外，在利用不對稱盈餘反應比較穩健程度時，應將盈餘對好壞消息反應分別比較。 本研究利用非線性及多期盈餘不對稱反應來比較不同公司與會計穩健性之關聯性，所檢視的公司特質為債權人及股東對股利政策之衝突、公司特有風險以及成長機會，本研究利用舉債程度、股利發放程度及營業風險作為債權人及股東對股利政策之衝突的代理變數；利用公司規模及權益市價對帳面價值之比率作為成長機會的代理變數。 實證結果顯示在盈餘不對稱反應、盈餘對好或壞消息的各別反應、非線性和多期盈餘不對稱反應下，公司特質與穩健的關聯性有不一致的情形，顯示過去文獻利用Basu (1997)比較穩健性之結論，需要以非線性及多期盈餘不對稱反應重新檢視。

Basu (1997) perceives conservatism as lower verification threshold for bad news than good news. He devises a reverse regression equation of earnings and stock returns to capture asymmetric timeliness of earnings. Prior research uses asymmetric timeliness of Basu (1997) model to compare the extents of conservatism. However, Lin and Liu (2011) argue that the components of asymmetric timeliness should be considered respectively and that the effect of nonlinearity and multi-period lag should be considered due to the essence of conservatism. This paper examines the relations between conservatism and firm characteristics by incorporating nonlinearity and multi-period lag effect into Basu (1997) model. Firm characteristics examined in this paper are bondholder-shareholder conflicts over dividend policy, firm-specific uncertainty, and growth opportunity. I use leverage, dividend payment and operating uncertainty as proxies for bondholder-shareholder conflicts over dividend policy and use size and market-to-book ratio as proxies for growth opportunity. The empirical results indicate that firms with different characteristics exhibit different extents of conservatism to different magnitudes of news and to different type of news. Further, multi-period asymmetric timeliness captures the extent of conservatism while not captured by concurrent asymmetric timeliness. The empirical results imply the importance of re-examination of results of prior literature.