An Empirical Investigation On Correlation Between Auditor Change And Earnings Quality

It has become one of the hotspots in foreign and domestic academy research that is about the behavior that the public corporations change the auditors in recent years. This paper positively investigates incremental effect that auditor change affects earnings quality by using change auditor corporations for research samples in Chinese public corporations in 2004 to 2006, at the same time, taking the unchanged auditor corporations for control samples. The author takes a perspective for decision-making usefulness of financial information and measure earnings quality in two dimensions from earnings management and earnings conservatism. According to the test, the author finds, comparing to unchanged auditor corporations, changing auditor will cause the degree of earnings management increasing significantly in public corporations. In additional investigation, the author finds auditor change will lead to the negative earnings manipulation in the public corporations, but the test using Basu(1997)'s earnings conservatism model finds that the changing auditor will bring incremental effect for earnings conservatism. This incremental effect will disappear after the author modifies Basu(1997) model and the author finds it is due to "big bath" of changing auditor corporations. At the same time, in order to control endogeneity by auditor change, the author adopts Heckman (1979)'s two step regression approach and finds the self-selection exists in auditor change, but the conclusion doesn't have substantial variety after this problem is controlled. The author also tests by the group in the direction of auditor change and find the above influence that auditor change brings to earning quality is due to the change from non-big4 to non-big4. In additional, the author also finds earnings in Chinese public corporations emerges conservatism as a whole. The author also designs various sensitivity tests, but the conclusion doesn't have substantial variety, which indicates the conclusion in this paper is robust.