Identifying Shell Company Risk with Artificial Intelligence

While not all shell companies are used for illegal means, many are utilized by corrupt officials, drug, sex, and arms traffickers, terrorist organizations, and fraudsters for tax evasion and avoidance, and money laundering. How can financial institutions better identify these pernicious entities?

The Corporate Transparency Act of 2019 and more recent debates in the U.S. Senate propose certain corporations report beneficial owners to FinCEN. Kenneth Blanco, FinCEN’s Director, has pointed to the use of anonymous shell companies as BSA/AML’s biggest and most concerning gaps. At a recent ABA Financial Crimes Enforcement Conference, Blanco argued that if beneficial ownership information were required at company formation, it would be more costly for bad actors to cover their tracks and would help restrict their access to the global banking system.

Detractors argue such legislation would inflict small businesses with burdensome reporting obligations under threat of fines or jail time while some AML experts argue that, if enacted, criminals are likely to find ways to circumvent the Act’s requirements.

While this is debated, the problem persists as legacy AML technology such as transaction monitoring systems (TMS) have little to no ability to identify and assess risk created by shell companies. And while policies, procedures, and processes, if applied correctly, can protect financial institutions from becoming conduits for some fraction of money laundering, terrorist financing, and other financial crimes, identifying shell company risk continues to be elusive.

According to the FFIEC, suspicious activities commonly associated with shell company activity include:

• Insufficient or no information is available to positively identify originators or beneficiaries of funds transfers (using Internet, commercial database searches, or direct inquiries to a respondent bank)
• Payments have no stated purpose, do not reference goods or services, or identify only a contract or invoice number
• Transacting businesses share the same address, provide only a registered agent’s address, or other address inconsistencies
• Many or all of the transfers are sent in large, round dollar, hundred dollar, or thousand dollar amounts
• One company is transferring funds to an unusually large number and variety of beneficiaries
• Frequent involvement of multiple jurisdictions or beneficiaries located in high-risk financial centers

Using these fingerprints left behind by shell companies, advanced technologies can effectively and efficiently identify shell companies and the financial crime they facilitate.

QuantaVerse develops AI-powered observables to identify financial crime. Some of the observables key to identifying shell company behavior include:

• Jurisdiction: the location of the focal entity and/or counterparties is high-risk
• Common address: multiple entities that are transacting with each other have the same address
• One to Many: transfers are going from one originator to many beneficiaries
• Web presence: there are no websites devoted to the entity nor search engine hits
• Phone number: no legitimate phone number
• Value Anomaly In & Out: large transaction payments are coming in or out
• Economic purpose: transactions between two entities are illogical based on their stated lines of business
• Intermediary Bank: the location of the intermediary bank is in a high-risk jurisdiction and/or there are multiple intermediary banks involved in a transaction

The findings of observables are taken into consideration to derive the overall risk score of an entity. For example, if jurisdiction and common address are the only two observables indicating shell company risk, a “Low Risk” score would be indicated. However, if a significant number of related observables indicated risk, a “High Risk” score would be provided.

QuantaVerse is utilizing Local Interpretable Model-Agnostic Explanations (LIME) to make the findings of complex algorithms more understandable. Investigators can see scores for various types of AML-related risks as well as understand how various types of risks contribute to the overall risk assigned to an alert or entity. This provides investigators deeper insights into risk segmentation and how observables should be considered case by case.

The platform’s findings are disclosed to investigators through QuantaVerse FCIRs. Through these reports, the human effort and workload associated with TMS alert triage, investigations, and SARs is automated by up to 70%, allowing investigators to focus on critical-thinking and ultimate adjudication of the case in question.

While FinCEN has previously worked to address shell company risk through the CDD Rule and now debates the merits of certifying beneficial ownership, there is more work to be accomplished. In the meantime, by reviewing this observable, the AI-powered QuantaVerse Financial Crime Platform is well-suited to assist financial institutions in mitigating shell company risk effectively and immediately.