University of Twente Student Theses

Login

Indicators of Bankruptcy Fraud: towards a predictive model

Braunsdorf, M. (2014) Indicators of Bankruptcy Fraud: towards a predictive model.

[img]
Preview
PDF
329kB
Abstract:This study examines which indicators are important in signaling bankruptcy fraud and how important professionals rate those indicators. From the literature examination, there seem to be 31 indicators that are believed to signal bankruptcy fraud. The research question is which indicators can predict fraud most successfully. This is answered through investigating the hypotheses: not all indicators appear with the same magnitude in fraudulent and non-fraudulent cases of bankruptcy and that the indicators receive different ratings of importance by the curators. 15 fraudulent bankruptcy cases were searched in the media and the dossiers were retrieved and analyzed for the presence of indicators, alongside with 45 non-fraudulent cases of bankruptcy. Furthermore a survey, also containing the 31 indicators was conducted. On this questionnaire 28 curators rated the indicators on a 5-point-Likkert scale. The results showed that both hypotheses are supported. There are several indicators that were significantly more often present in the fraudulent than in the non-fraudulent cases. Also, it was shown, that the curators do not rate the different indicators of fraud with the same importance, but that there are indicators that are rated significantly higher than others. Subsequently, with the help of that information a model for fraud prediction is derived and tested for its predictive value with 3 additional cases of fraudulent bankruptcies and 9 non-fraudulent bankruptcies. A model was created that could predict fraud through the presence of different indicators, found in the bankruptcy dossiers. From the two studies it became obvious, that the curators already apply a good focus, but that there are also shortcomings in the treatment of certain indicators. The studies yielded overlapping, but not identical conclusions. For future research it is advised to examine additional sources of data and to merge those findings into the model if possible. Through those findings a greater depth of information is achieved which eventually can lead to the formation of a more elaborate model for fraud detection.
Item Type:Essay (Bachelor)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:77 psychology
Programme:Psychology BSc (56604)
Link to this item:http://purl.utwente.nl/essays/65627
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page