Title:
|
IDENTIFICATION OF BANKRUPTCY FRAUD IN DUTCH ORGANIZATIONS |
Author(s):
|
Bernard P. Veldkamp , Theo De Vries |
ISBN:
|
978-972-8924-63-8 |
Editors:
|
Hans Weghorn and Ajith P. Abraham |
Year:
|
2008 |
Edition:
|
Single |
Keywords:
|
Bankruptcy fraude, Fraude detection, Application, Classification Trees, Artificial Neural Nets |
Type:
|
Short Paper |
First Page:
|
63 |
Last Page:
|
66 |
Language:
|
English |
Cover:
|
|
Full Contents:
|
click to dowload
|
Paper Abstract:
|
The damage of bankruptcy fraud is substantial. In response, the Dutch Ministry of Justice started a project to reduce the
number of bankruptcy fraud cases by increasing the probability of prosecution. But what is the best method to develop a
model for indicating fraud based on data streams coming from the Chambers of Commerce, the Tax Administration, and
the Directorate of Public Prosecution (criminal records)? In this paper, classification trees and artificial neural nets are
applied. Data were collected during the period 2005 2007 in the region North-East of the Netherlands. Within this
period of time, 941 bankruptcies occurred in this region. All of these cases were thoroughly investigated by officers of
the Directorate of public prosecution - one by one. For 152 cases bankruptcy fraud was detected. Only some initial
analyses have been completed yet. At company level, we succeeded in detecting 15-20 % of the frauds
(likely followed by a verdict). At person level 30 % of the board members that were indicated as fraudulent
were detected. Around 65% of the fraudulent board members were correctly indicated. The main indicators
in the model turned out to be whether board members had a criminal record, and the number of verdicts.
In comparison with current practice, where only 2 percent of the frauds are detected, this automated
procedure for fraud detection seems to be a promising help for the Directorate of public prosecution in
fighting this kind of white collar crime. |
|
|
|
|