Sarbanes-Oxley Act Early Effect: An Empirical Research Using Auditor Change Prediction Data Mining Approaches

Abstract

This study proposes using multiple criteria quadratic programming (MCQP) and other data mining approaches to predict auditor changes with early adopters of the Sarbanes-Oxley Act (SOX). It compares 2003-2004 U.S. firm data with data from 2005-2006 to measure the SOX effect on firms that voluntarily adopted this new regulation nearly (other than the size of the business). The results of the MCQP and other data mining approaches in this auditor change prediction study show that the MCQP method performs marginally better than other data mining approaches using financial and other data to predict auditor changes. In addition, the early SOX effect is not significant empirically using the auditor change prediction model of comparing the prediction rates of early adopters vs. those of later adopters.