Want more Insights from SAS? Subscribe to our Insights newsletter. Or check back often to get more insights on the topics you care about, including analytics, big data, data management, marketing, and risk & fraud.
Fighting customs fraud
What does big data mean for customs agencies?
By Jérôme Bryssinck, Director of Government Fraud Solutions, EMEA & AP at SAS
Fraudsters know that there are large profits to be made by avoiding customs duties and importing illegal goods. The chances of detection are small, so it’s no surprise this is a target area for organized crime.
Security risks have become a top priority for customs departments. To spot criminals, customs authorities have moved from relying on manual searches and officers’ detective skills to rules-based systems and database searching. However, these systems are now also under pressure as smugglers become more sophisticated and better at outmaneuvering existing systems.
For many years, banks have been continuously looking for suspicious or unusual behavior in real time. Big data is making detecting unusual transactions easier.
Scoping the challenge
Officials today require detailed information about each shipment so as to scrutinize its veracity. A skilled customs expert will look beyond the immediate shipment and search for identifying links, such as directors and known fraud cases, as well as being aware of what’s been declared and whether it poses a specific risk.
Doing this manually is time-consuming. To conduct risk management thoroughly, agencies must gather data from multiple sources, including the business itself, other customs and tax agencies, and the police. Staff shortages compound the difficulties, but agencies must act fast to respond to information that requires immediate action.
All this means that customs officials can often only check a small percentage of goods and must therefore carefully target their approach.
Today most agencies, at best, are only armed with systems that automate the separate customs management processes they run. First-generation rules-based solutions often lack the advanced analytics necessary to optimize their rules. They typically generate large volumes of false positives, resulting in wasted time, reduced operational efficiency, increased costs, mistrust and ultimately system atrophy. Coupled with this, there is a lack of information sharing between government agencies due to political and cultural issues, concerns about privacy and security, poor policies, and inadequate technologies.
Fortunately, a new breed of risk management solutions is emerging, offering a range of operational advantages over its predecessors. These systems can make real-time decisions on very low-latency data. For example, agencies can use risk management systems on information provided on shipments before loading at the port of origin and deny loading for suspicious shipments. These advanced risk management systems enable agencies to rank import transactions and manage associated information to create risk scores. Field offices can plan the deployment of staff to best match threat priorities.
For many years, banks have been continuously looking for suspicious or unusual behavior in real time. Big data is making detecting unusual transactions easier. Similar techniques can now be used by customs departments. For example, the 2010 Yemen parcel bomb plot could easily have been avoided had a risk detection system been in place. During this incident, two packages, each containing a bomb of plastic explosives and a detonating mechanism, were found on separate cargo planes. The shipments originated from Yemen, with a final destination in the US. Yemen is not a common route for office supplies to the US. Also, the companies involved in the trade were highly suspicious. Therefore, a risk detection system would have picked up such an anomaly and would have prevented the loading of this particular cargo.
Another key benefit to the new systems is that they deploy self-learning advanced hybrid analytics to reduce false positives. With false positive rates minimized, customs officers can focus their time on fraudulent shipments rather than on unwarranted alerts, increasing efficiency.
Critically, too, the latest solutions can score imports and exports in real time. This allows 24/7 monitoring and accelerates the decision and detection process.
The days of expensive, in-house fraud management projects built from scratch are ending. Increasingly, governments are looking to pre-packaged solutions that offer low risk and costs, scalability and a high level of future-proofing. The automated approach these technologies offer to risk management is here to stay. They are already operational and profitable within several customs departments globally, where they are cutting costs, increasing revenue and bringing more fraudsters to justice.