Accounting Fraud: an estimation of detection probability

• Autor(a) principal: Wuerges, Artur Filipe Ewald
• Data de Publicação: 2014
• Outros Autores: Borba, José Alonso
• Tipo de documento: Artigo
• Idioma: eng; por
• Título da fonte: Revista Brasileira de Gestão de Negócios (Online)
• Texto Completo: https://rbgn.fecap.br/RBGN/article/view/1555
• topic: Account ing fraud. AAER. Misclassification. Logit. Factor analysis.; Fraude contable. AAER. Errores de clasificación. Logit. Análisis factorial.; Fraude contábil. AAER. Erros de classificação. Logit. Análise fatorial.

Financial statement fraud (FSF) is costly forinvestors and can damage the credibility of theaudit profession. To prevent and detect fraud, itis helpful to know its causes. The binary choicemodels (e.g. logit and probit) commonly usedin the extant literature, however, fail to accountfor undetected cases of fraud and thus presentunreliable hypotheses tests. Using a sample of118 companies accused of fraud by the Securitiesand Exchange Commission (SEC), we estimateda logit model that corrects the problems arisingfrom undetected frauds in U.S. companies. Toavoid multicollinearity problems, we extractedseven factors from 28 variables using the principalfactors method. Our results indicate that only 1.43percent of the instances of FSF were publicized bythe SEC. Of the six significant variables includedin the traditional, uncorrected logit model, threewere found to be actually non-significant inthe corrected model. The likelihood of FSF is5.12 times higher when the firm’s auditor issuesan adverse or qualified report.