How Graph Analytics Revolutionizes the War Against Human Trafficking

The United Nations Office of Drug Crime estimates that $800 billion to $2 trillion, or 2-5% of global GDP, is laundered globally every year. 

What is driving this massive amount of money laundering? After drug trafficking, human trafficking is the second top motive for money laundering. Forced labor generates over $150 billion per year by robbing a stunning 25 million people of their freedom and basic human dignity (that’s about the size of Australia’s total population). Almost two-thirds ($99 billion) of this revenue comes from sexual exploitation. Sexual exploitation is thus the most profitable form of human trafficking, as victims account for less than 20% of people trafficked.

How Traffickers Exploit Inefficiencies

Traffickers leverage the financial system for many reasons: to be paid by their clients, to book their travel and accommodations, and to finance their sordid operations. An organized criminal network is often involved in various forms of human trafficking, such as sex trafficking, forced labor, and debt bondage.

The unidentified relationships among the people and entities involved can make anti-money laundering extremely difficult for financial institutions to detect. There are some well-documented red flags, however, for which financial institutions should monitor:

• Expenditures on websites that advertise trafficked persons
• Transactions made in locations known to law enforcement as sex worker locations
• Frequent or excessive travel expenses (foreign, domestic, and local)
• Cash deposits made in multiple cities, often avoiding the use of bank tellers
• Repeated or excessive spending on food, lodging, or personal needs for trafficked persons by the trafficker. Similarly, the victim’s financial activities may reflect none of these expected daily expenditures.

Some of the most crucial red flags are not financial activities, however. Therefore, anti-money laundering investigation teams in banks should have access to not only internal transaction data but also external data.

Some non-financial red flags include:

• The suspected victim’s mobile number is linked to escort service advertisements
• An unusual number of joint account holders or authorized users on bank accounts
• Multiple victims with identical bank account details, such as phone numbers and addresses

Traditional monitoring systems are ineffective because they are designed to only account for internal transaction flows and are unable to analyze human trafficking behavior thoroughly.

Furthermore, even when traditional investigation systems do detect behavior related to human trafficking, they reveal neither hidden relationships among involved parties nor anything about these parties’ activities on risky advertisement websites, on social networks, or elsewhere in the media. This leads to inaccurate investigator decisions and increased financial crime risk exposure for financial institutions.

Increase Agility to Fight Human Trafficking

The nature of human tracking varies drastically by countries. In Brazil, for example, human traffickers exploit Brazilian victims in forced labor, including on farms and in factories and restaurants, after the victims join particular churches or religious cults. In Cambodia, a lack of jobs has led women and girls to leave their homes in rural areas to seek work in tourist destination cities. These women often falling victim to sex trafficking when they arrive in such cities.

In Ethiopia, traffickers often deceive parents of children living in rural areas into sending their children to major cities to work as domestic workers. In India, the government officially abolished bonded labor in 1976, but the system of forced labor still exists. In the United Kingdom, gangs force British children to carry drugs. In the United States, traffickers prey upon children in the foster care system.

An agile platform that allows global institutions to tailor policies to meet the ever-changing environment as per their country of operations is critical to the success of the program.

Graph Analytics: A Game Changer in the War Against Human Trafficking

Innovative technologies such as graph analytics, machine learning, and AI make it possible to fight highly complex organized financial crimes successfully. The technology does so by assessing complete customer behavior and by increasing financial crime compliance program agility.

Here is an example of how innovation can be applied to transaction monitoring, investigation, and optimization to fight human trafficking.

Graphic 1: Revolutionizing the War Against Human Trafficking

Network correlation-based monitoring: Traditional rules-based detection systems are designed to monitor individual customer activity. Graph analytics can uncover the complete network behavior of multiple victims (customers) operating under the same trafficker. Activities that may appear legitimate when investigated in silos may be a small part of an extensive organized criminal network. Correlating the activities of various related parties and their red flags using powerful graph analytics is a more efficient way to detect complicated financial crimes.

In Graphic 1 above, Customer 4 and Customer 5 may appear unrelated in the traditional system, leading to two separate cases. With graph analytics, the relationship between both customers based on shared information (i.e., their tax identification number) is uncovered. This can then link a more extensive criminal network comprising other related parties such as Customers 1, 2, and 3.

Combining internal and external data for investigations: In the traditional investigation process, users access various systems for media scans, screening, and keyword searches. Since there is no single platform to associate all the information, investigators find it hard to see the holistic view of various involved parties. A unified investigation platform allows investigators to blend both internal and external data, leveraging graph analytics. This provides a full picture of risk indicators, which requires external data.

Negative news, suspicious keywords, and high-risk websites can be analyzed holistically. Such a platform provides not only quicker and more efficient investigations but also high-quality case narratives for suspicious activity reporting and communicating evidence to law enforcement agencies.

Deep learning using graph: Graph analytics opens new avenues of deep learning using graph algorithms such as Graph Similarity, which determines the degree of similarity between graphs by making inferences from similarities between their respective entities (“nodes”) and relationships (“edges”). Intuitively, the nodes in two graphs are similar if their neighbors (and their connectivity, in terms of edge, to their neighbors) are similar. Again, their neighbors are identical if their neighborhoods are identical, and so on. This intuition guides the possibility of using belief propagation (BP) to measure Graph Similarity, due to BP's dependence on neighborhood structure. Deep learning in human trafficking can be valuable since similar past behaviors can be clusters that can predict new cases. 

A Safer World

Human trafficking is a heinous crime and a human rights violation that is often referred to as “modern slavery.” Unfortunately, despite all the efforts of governments and civil society to combat this financial crime, it still ensnares many victims around the world. In 2009, only 26 countries had an institution that systematically collected and disseminated data on trafficking cases. While the number rose to 65 in 2018, many countries in Africa and Asia continue to have low conviction rates, and at the same time, detect fewer victims.

The good news is that affordable, innovative, anti-money laundering solutions can help. Financial institutions can tackle highly complex crime networks with them, contributing to a safer world for everyone.