【正文】 kely to replace management teams due to the following reasons: to recover reputation loss, to guarantee an unbiased SEC investigation, or to satisfy the anxious investors.Agrawal, Jaffe and Karpoff (1999) study firms suspected or charged with fraud during 19811992. Their study includes the following types of fraud: fraud against shareholders, fraud against government, financial reporting fraud and regulatory violations. They do not find systematic evidence of unusually high turnover among senior managers and directors. Even for financial fraud firms, they do not find higher top management and director turnover rates than control firms. Their findings are rather surprising since accounting fraud signals lower firm value and large fraud could be detrimental to shareholders’ interests. Retention of the old management team is not conducive to a smooth investigation by the SEC and may result in significant reputation losses. In this paper we extend turnover analysis to three levels: total management turnover (. Chairman, CEO, President, CFO, Treasurer and Controller), top management turnover (. Chairman, CEO, President) and financial management turnover (. CFO, Treasurer and Controller) over a longer event window (2, 2).Denis amp。 Denis (1995) document that forced resignations of top management officials are usually preceded by large and significant declines in operating performance and are followed by large improvements in their stock performance. Our hypothesis that firms that mit accounting fraud are more likely to suffer from declining financial performance and subsequently incur higher management turnover is consistent with their findings. Mian (2001) studies the choice and replacement of Chief Financial Officers and finds that, in general, CFO turnovers are disciplinary and that they are usually preceded by a decline in operating return on assets, negative excess returns, and abnormally high CEO turnover. We use financial management turnover to evaluate the impact of accounting fraud on firms. financial group. Financial turnover, which includes the CFO, treasurer and controller, is more representative than CFO turnover, mainly because the CFO might not necessarily be the person who is responsible for the accounting will examine whether financial management turnover is preceded by declining firm and stock performance. We also want to document the determinants of management turnovers in the multivariate logistic regression setting.The accounting and auditing enforcement releases (AAER) firms are used as our sample of accounting fraud. A parison of AAER firms to control firms indicates that they are more likely to be financially distressed and have much higher management turnover. In the multivariate framework, we find that top management turnover is significantly positively related to the AAER disclosure, after controlling for firmspecific characteristics.The remainder of the paper is organized as follows: In Section II, we describe the data selection and present the univariate parison between AAER firms and the control sample. In section III, we present the multivariate logistic regression results on determinants of top management and financial management turnover. II. Data on the AAER and Control Firm Samples All firms included in the SEC39。s accounting and auditing enforcement releases (AAER) over the period 19902000 make up the population studied. The SEC enforcement program investigates and takes subsequent injunctive actions or administrative proceedings against registrants and auditors who have violated the financial reporting requirements of the Securities Exchange Act of 1934. Feroz et al. (1991) point out that the SEC ranks and pursues targets according to the probability of success. This is because the SEC has more targets than it can practically pursue and formal investigations are both costly and highly visible. An example would be a group of cases where costs were understated in the previous financial statements, the SEC would only pursue the cases where it can demonstrate that management knew or should have known through better internal controls. Therefore, it is reasonable to assume that AAER firms knowingly or intentionally engaged in accounting fraud.A. Sample The AAER database is used to extract all sample firms during the period 19902000. Since there are a large number of firms with multiple offenses, we only use their earliest offense as the event date due to the following reasons: previous events may have a residual carryover effect。 frequently, consequent disclosures refer to the same accounting fraud that occurred in the past。 finally multiple events from the same firm may bias the independence of the variables, management turnovers and longterm returns. As shown in panel A of Table I, we have a total of 291 unique firms with their first offense included in the dataset, relatively evenly spread out across the decade. These incidents are not clustered around a few years. To record