Risque associé à l'utilisation de la loi de Benford pour détecter les fraudes dans le secteur de la mode

Objectives. This paper aims to show that it's not always possible to detect fraud in sales volume with Benford's law. Data. I use video games hardwares sales volume, in Japan (from april 1989), in United-States, France, Germany and United-Kigngdom (from november 2000). Study Design. After a short review of litterature and an introduction of my method, I test the adequation with Benford's law of my 56 weekly sales volume time series with khi-square statistics. Then, I present the bias and their significance. Results. These tests show the inadequacy of our fashion sales time series with the Benford's Law despite the quality of our data base. Thus, for video games hardwares sales volume, Benford's law may be inefficient to detect frauds.