Finding shell company accounts using anomaly detection
Devendra Kumar Luna, Girish Keshav Palshikar, Manoj M. Apte, Arnab Bhattacharya · 2018
Money laundering refers to activities pertaining to hiding the true income, evading taxes, or converting illegally earned money for normal use. These activities are often performed through shell companies that masquerade as real companies but where actual the purpose is to launder money. Shell companies are used in all the three phases of money laundering, namely, placement, layering, and integration, often simultaneously. In this paper, we aim to identify shell companies. We propose to use only bank transactions since that is easily available. In particular, we look at all incoming and outgoing transactions from a particular bank account along with its various attributes, and use anomaly detection techniques to identify the accounts that pertain to shell companies. Our aim is to create an initial list of potential shell company candidates which can be investigated by financial experts later. Due to lack of real data, we propose a banking transactions simulator (BTS) to simulate both honest as well as shell company transactions by studying a host of actual real-world fraud cases. We apply anomaly detection algorithms to detect candidate shell companies. Results indicate that we are able to identify the shell companies with a high degree of precision and recall.1