Date of Award

Spring 2001

Document Type


Degree Name

Doctor of Business Administration (DBA)


School of Accountancy

First Advisor

Michael S. Luehlfing


This study investigates the premise that certain types of negative information are associated with auditor switches. A data set of 305 auditor switches from 1976 to 1994, extracted from the Compustat data base (limited subscription), was analyzed using tests of proportions and nonparametric sign tests. The data set consists of negative information extracted from the switching companies' income statements (i.e., net losses or extraordinary items) or calculated from the items extracted (i.e., net income adjusted to reverse the effect of extraordinary items).

The initial results, based on tests which assumed random movement of net income, did not support the notion that net losses or decreases in net income (with or without adjustment to reverse the effect of extraordinary items) occurred disproportionately in the year after the auditor switch compared to the year before the auditor switch. However, supplemental analyses which used a different expectation for net income (i.e., a 1972 finding by Ball and Watts that accounting net income each year is greater than or equal to the preceding year's net income) do support the notion that more negative information is reported in income statements following an auditor switch than preceding the auditor switch.

This study uses a data source, the Compustat data base, which has not been used much in previously published journal articles on auditor switching. (In fact, when the study was begun [in 1996], the author found no previously published study of auditor switching which had used this source.) Investigation of the underlying data indicates that (1) switches among non-Big audit firms are not available before 1989, (2) auditor code changes before 1990 which appear to indicate a change from non-Big audit firms to Big audit firms might actually be the result of a business combination of audit firms, and therefore should be subjected to further investigation, and (3) auditor codes should be visually examined over multiple years (for reasonableness) before using the codes as a basis for auditor switching studies. Additionally, the data set employed indicates that auditor switches by publicly-owned companies may frequently be a predictor of bad financial news which has not yet occurred.