Statistical Methods for Fighting Financial Crimes Agus SUDJIANTO Bank of America Charlotte, NC 28255 (Agus.Sudjianto@bankofamerica.com) Sheela NAIR Bank of America Charlotte, NC 28255 (Sheela.nair@bankofamerica.com) Ming YUAN Georgia Institute of Technology Atlanta, GA 30332 (myuan@isye.gatech.edu) Aijun ZHANG Bank of America Charlotte, NC 28255 (Aijun.zhang@bankofamerica.com) Daniel KERN Bank of America Charlotte, NC 28255 (Daniel.c.kern@bankofamerica.com) Fernando CELA-DÍAZ Bank of America Charlotte, NC 28255 (Fernando.cela-diaz@bankofamerica.com) 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 de- liberately 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 un- known 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. KEY WORDS: Anomaly detection; Classification; Fraud detection; Machine learning; Money launder- ing. 1. INTRODUCTION Financial crimes refer to a broad category of crimes against property, committed by individuals and organizations to obtain a personal or business advantage. As described by the Federal Bureau of Investigation (2005), they are characterized by de- ceit, concealment, or violation of trust, and are not dependent upon the application or threat of physical force or violence. Ex- amples include money laundering, credit/debit card fraud, em- bezzlement, counterfeiting, mortgage fraud, and insider trad- ing, to name a few. These crimes cost several billions of dollars a year and affect the lives of millions of people. By law, U.S. financial institutions are required to carry a substantial amount of the responsibility for combating financial crimes. Of course, these institutions must also protect their customers and share- holders from financial loss. This has led to extensive research and the development of detection systems. This article focuses on two specific types of financial crimes that pose large threats: money laundering and retail banking fraud. In money laundering, the criminals hide the true origin of funds by sending them through a series of seemingly legit- imate transactions. The main purpose of laundering money is to conceal the fact that funds were acquired as a result of some form of criminal activity. These laundered funds may, in turn, be used to foster further illegal activities such as the financing of terrorist activity, trafficking of illegal drugs, support of pros- titution rings, or smuggling of weapons. Even the laundering of legitimate funds to avoid reporting them to the government (e.g., tax evasion) leads to substantial costs for society. The U.S. Internal Revenue Service estimates over $300 billion in taxes went unpaid in 2001 alone. In retail banking fraud, the criminal attempts to achieve fi- nancial gain at the expense of legitimate customers or financial institutions through any retail banking transaction channel, such as credit cards, debit cards/ATM’s, online banking, or checks. Most of the fraud detection research focused on credit card fraud, which was estimated at close to $1 billion in the U.S. and $10 billion worldwide (Ghosh and Reilly 1994; Aleskerov, Freisleben, and Rao 1997). This poses a serious problem for financial institutions as they increasingly assume responsibil- ity for all unauthorized transactions. Debit card fraud is also a growing concern as these cards gain volume share; debit/ATM fraud losses in the U.S. were estimated at $2.75 billion (Gart- ner 2005). A common pattern, skimming, involves collecting personal identification numbers (PIN) from compromised point of sale readers or ATM’s and cloning the card’s magnetic strip. This enables criminals to obtain cash without any human con- tact in a way that is particularly difficult to trace. The financial © 2010 American Statistical Association and the American Society for Quality TECHNOMETRICS, FEBRUARY 2010, VOL. 52, NO. 1 DOI 10.1198/TECH.2010.07032 5