Statistical Methods for Fighting Financial Crimes

Abstract

Financial crimes affect millions of people every year and financial institutions must employ methods to protect themselves and their customers. The use of statistical methods to address these problems faces many challenges. Financial crimes are rare events that lead to extreme class imbalances. Criminals deliberately attempt to conceal the nature of their actions and quickly change their strategies over time, resulting in class overlap and concept drift. In some cases, legal constraints and investigation delays make it impossible to actually verify suspected crimes in a timely manner, resulting in class mislabeling or unknown labels. In addition, the volume and complexity of financial data require algorithms to be not only effective, but also efficiently trained and executed. This article focuses on two important types of financial crimes: fraud and money laundering. It discusses some of the traditional statistical techniques that have been applied as well as more recent machine learning and data mining algorithms. The goal of the article is to introduce the subject and to provide a survey of broad classes of methodologies accompanied by selected illustrative examples.

Machine Learning in Banking Risk Management: A Literature Review

“…There has been extensive research on detecting financial crimes using traditional statistical methods, and more recently, using machine-learning techniques. Clustering algorithms identify customers with similar behavioural patterns and can help to find groups of people working together to commit money laundering (Sudjianto et al 2010). A major challenge for banks, given the large volume of transactions per day and the non-uniform nature of many, is to be able to sort through all the transactions and identify those that are of suspicious nature.…”

Thoughts on Data Science in business and industry

“…Complex algorithms enable banks to identify certain activities as potentially fraudulent. Statistical methods for anomaly detection are crucial . This has a game theory aspect to it, with the banks trying to stay one step ahead of the criminals, and the latter trying to figure out how to “game” the bank's algorithms.…”

Goodness-of-Fit Testing for the Newcomb-Benford Law With Application to the Detection of Customs Fraud

“…Another disadvantage of testing the global NB null hypothesis is that rejection does not shed light on how many digits deviate from the law, nor on which digits are responsible for rejection. We are interested in anti-fraud analysis of customs data arising from international trade, where the goal is to detect illegal actions such as tax evasion and money laundering (Deng et al, 2009;Sudjianto et al, 2010). In this context, multiple-digit deviation from the NB law may seem more suspicious than single-digit non-conformity, as a signal of data fabrication.…”