Processus de détection et évaluation de la fraude sociale
This article presents an econometric model of social fraud contributions which takes into account the process of monitoring and detection. The introduction of these two processes in the model allows to correct two important biases and so propose an unbiased fraud estimation. The first bias is inherent to data from ?selective? controls of contributors assumed to be more likely fraudulent. The second is related to the possible failure in detection of the entire fraud during inspections. The estimates are based on individual data of small and medium enterprises of the Lyon metropolitan area, drawn from the confidential database of the social security administration. Our results confirm that the selection bias leads to over-estimate fraud, while the detection bias tends to underestimate fraud. According to our results, in the absence of correction of these two biases, fraud would be over-estimated of approximately 13%. Classification JEL : C34, D81, H26, J22.
When requesting a correction, please mention this item's handle: RePEc:cai:recosp:reco_605_1235. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (Jean-Baptiste de Vathaire)
If references are entirely missing, you can add them using this form.