By Kati Suominen, Founder and CEO, Nextrade Group and Techical Director, eTrade Alliance

 

With millions of parcels and packages flooding countries’ borders, ecommerce is creating new complexities for customs and border agencies seeking to meet their triple objectives of securing and facilitating trade and collecting revenue. On the one hand, customs struggle to secure trade and collect revenue without opening every package to see what is inside; on the other, customs agencies look to facilitate trade so that businesses and consumers buying products online get them in a timely manner.

 

As a result of the massive inbound ecommerce flows, customs and border agencies around the world have set out to pilot blockchain and artificial intelligence as a means to automate risk management and fraud detection. The eTrade Alliance’s policy mapping shows that over one-half of mapped countries have piloted blockchain in border clearance, or joined broader blockchain platforms such as Tradelens and IDB’s Cadena. What can we learn by now from these pilots?

 

Blockchain Pilots

 

There are a number of blockchain pilots with various aims, such as to enable sharing of documents among the trade community and customs; verifying origin or products entering under free trade agreements preferences; and streamlining interactions among national customs agencies. For example:  

 

• The Korean Customs Service (KCS) has been a pioneer in piloting blockchain. In 2018, KCS conducted various pilot projects, such as Cross-Border Project with Vietnam to improve the reliability of origin certificate data via blockchain; and Export Logistics Blockchain Project with Samsung, Hyundae Globis, Busan Port Terminal, Shinhan Bank and over 60 Korean companies to test blockchain’s usefulness in the accuracy and transparency of shipment data generated by the logistics community. Ten companies based in Vietnam and Singapore on the import side have participated in pilots.

 

To pave the way for the pilots, KCS created a dedicated division for blockchain, and selected as project managers members of staff that had been trained by leading blockchain companies. To develop the pilots, KCS engaged extensively with the stakeholders, holding many meetings and workshops where the stakeholders could define data that could be shared, and share information related to export logistics and their respective business processes. KCS’s ICT Development Division led the technology’s adoption; the blockchain platform was developed by Samsung and KCnet. It was geared to using blockchain to generate and share information such as trade documents as export declarations, bills of lading, and letters of credit, among others (figure 1).

 

 

 

Figure 1 – Setting Before and after a Korean Customs Service (KCS) Blockchain Pilot with Private Sector

 

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

Blockchain significantly improved KCS’s operations – reduced manual work and increased particularly the reliability of data on shipments and supply chains as the data are collected from multiple sources and tamper-evident. KCS sees great potential in leveraging the Big Data on blockchain, and has trained dozens of staff to better leverage big data.

 

• The U.S. Customs and Border Protection (CPB) has worked with the private sector to  develop several blockchain use cases, such as tracking licenses, permits, certificate of origin reporting, free trade agreement compliance, and bonded movement tracking. In 2018, CBP ran a proof of concept to test blockchain’s feasibility for verifying certificates of origin for entries under North American Free Trade Agreement and Central America Free Trade Agreement. The test moved the data requested in two CBP forms onto blockchain, enabling nearly instantaneous communications between CBP and the business community and receipt of files.

 

The POC expedited processing, enabled 10 systems integrations, and overall proved that blockchain technology can improve the processing and tracking of trade-related documents –particularly to improve the auditability and evaluation of entry documents; transparency; and identification of suppliers and manufacturers. Blockchain also enabled digitized submission of certificates of origin, standardized the process for filers, and overall expedited processes for filers. The POC was particularly helpful in showing that the CBP’s blockchain could interoperate with a third party blockchain – in this case, UPS’s systems. This potential for customs to connect to pre-existing databases and systems would be enhanced further by global blockchain standards.

 

CBP has also set out to facilitate ecommerce through a 24-month pilot announced in 2019 with nine companies to voluntarily submit data on one-off ecommerce shipments below the $800 de minimis threshold. The pilot is to provide CBP more data on these goods that are fast-tracked thorough customs, to identify and target high-risk shipments such as narcotics, weapons, and products presenting health and safety risks.  The pilot covers goods entering the U.S. by air, truck, or rail.

 

• Chile, Colombia, Costa Rica, Mexico, and Peru have made significant progress enhancing the interoperability among their respective customs agencies. In 2018, they piloted a project supported by the Inter-American Development Bank called CADENA, which uses blockchain to share mutual recognition agreements (MRAs) and Authorized Economic Operator programs. The existing process of sharing AEO data is arcane, as it is done by bilateral emails among the different customs agencies. By using blockchain, the various customs will have one-stop access to the same information, securely and in one place – and AEO beneficiaries will secure AEO benefits immediately when they receive their AEO certification (figure 2).

 

CADENA’s development was a nimble and collaborative process. In early 2018, IDB staged a workshop to identify customs pain points with the governments. All customs pointed to the frictions in sharing data on AEO certified companies. As a result, the IDB created an interdisciplinary team of trade and technology experts across the Bank and experts from the various customs agencies; together, the team drafted a vision and business case for data sharing on AEOs, and chose blockchain as a tool to accomplish their vision. The team proceeded to pilot the solution in various phases; the IDB funded the pilot project with a modest budget.

 

The first phase resulted in definition of the functionalities, technical requirements, and data management requirements for the envisioned solution. An ad hoc governance structure was defined, consisting of the participating customs administrations, IDB, and the technology vendor. The parties also defined data privacy and user identity management policies, to ensure the platforms’ security, and agreed to create a Power App to enable customs officials and AEO-certified companies to easily access the platform. The participating customs agencies opted to integrate the solution into their legacy systems, to keep the pilot focus on the exchange of data, and worked iteratively in frequent communication through the pilot.

 

The gains from the pilot solution included reduced time for granting benefits to new AEO-certified firms in the countries of destination for their cargo operations; increased transparency and traceability of cross border data; access to real-time data for new AEO certified companies and AEO suspensions and cancellations in real-time across countries’ customs. The pilot also increased the global trade community’s awareness about the potential of blockchain to solve pain points in border clearance and new ideas to leverage blockchain, for example to enhance coordination among trade agencies in the various countries’ single windows.

 

Figure 2 – Setting Before and after Inter-American Development Bank’s CADENA Blockchain Pilot

 

 

​

​

​

​

​

​

​

​

​

​

​

​

​

​

 

 

Many other governments are driving blockchain initiatives:

 

• The European Commission’s Directorate-General for Taxation and Customs Union (DG-TAXUD) tested blockchain in 2017 in temporary admission and excise movement control, a computerized system for monitoring the movement of alcoholic, tobacco, and energy goods in the EU. The purpose of this system is to combat fiscal fraud with real-time data and checks on goods being moved under duty-suspension, and ensure secure movement of excise goods for which duty has still to be paid, and simplify procedures for traders of these goods. The POC was used to track the goods across three EU countries. results were “very encouraging,” and found that there was still more to do with data confidentiality.

 

• In Japan, New Energy and Industrial Technology Development Organization (NEDO), a Japanese government-controlled organization, intends to improve the sharing of trade-related information across Japanese trade ecosystem. The company has partnered with the Japanese IT giant NTT Data to develop a blockchain trade platform of the Japanese Ministry of Economy, Trade, and Industry (METI).

 

• In Mexico, customs, customs brokers, Hutchison Ports, and the Port of Veracruz have together piloted a blockchain solution that provides them with common, real-time data on the location and documents associated with a given import.

 

• Numerous countries such as Australia, Bahrain, Canada, Ghana, Indonesia, Jordan, Peru, Saudi Arabia, Singapore, Thailand, Uruguay, and Vietnam have piloted and joined Tradelens, a blockchain platform kickstarted by IBM and Maersk to connect the trade and supply chain ecosystem. At the end of 2019, Tradelens reported it was supported by “two-thirds of container shipping lines globally, over 80 terminals and ports, 17 customs authorities, dozens of inland providers, many corporates and banks, and several leading global and regional freight forwarders” and helped transmit two million events and over 15,000 documents daily.

 

• A new Mercosur-wide blockchain initiative called bCONNECT is used by Argentina, Brazil, Paraguay, and Uruguay to test blockchain technology in customs security and identity verification of persons who feed information to the blockchain platform.

 

AI in Customs

 

A growing number of countries is applying AI – such as machine learning, machine vision, and natural language processing – to detect customs fraud and illicit trade, match products to the right HS codes to collect the right revenue, and to enable ecommerce buyers and sellers to estimate the total cost of delivery. Brazil and Singapore have been among the pioneers:

 

• Brazil has since 2014 applied artificial intelligence tool called SISAM to assess the risks associated with imports. The system is based on both supervised and unsupervised learning. With supervised learning, the algorithm is taught with import declarations that have been inspected by customs officers, to identify correlations between such fields as the importer’s identifier; the importer’s national economic activity classification (the code that defines the production activity of a company); nomenclature codes of the goods being imported; countries involved in the production; commercialization and transportation of the goods; customs broker that registered the declaration; and manufacturers and suppliers of the goods. This enables the computers to learn patterns and alert officials when it works on its own with new declarations and encounters outliers, declaration that do not meet the pattern of licit shipments. For example there can be incompatibilities between how goods are described and what their declared HS codes are.

 

SISAM has 8.5 billion different patterns, and estimates the probability of 30 types of errors, such as false descriptions of goods, errors in the nomenclature codes, errors in the declared countries of origin, missing import licenses, non-applicable tax regimes, and wrong preferential tariff claims. While customs officers resisted the application at first, over time they came to see it as useful and today they check a shipment based on SISAM’s suggestions 30 percent of the time. Brazil also uses a tool called ANIITA to extract data from multiple systems and show the most important information for the customs clearance and risk assessment process on a single screen, saving users time they would spend on manually browsing many systems.  

 

In addition, in 2017, Brazil adopted PATROA, a real-time risk identification tool that reads transactions as soon as they are registered. PATROA can email or instant message an officer. The officer can respond and stop the transaction right away if convenient. PATROA can also decide not to call an officer immediately, but to start saving information – for example, data on anomalous behaviors of an authorized economic operator (AEO).

 

• Also Singapore is using AI and machine learning and machine vision in its customs clearance. Singaporean customs identifies anomalies in shipments on the basis of predefined criteria and historical datasets with such data as shipment clearance times and declaration-related information. Singapore is using machine learning to analyze hypotheses and associations in data faster than could be done by human staff using traditional econometric techniques. Machine learning has enabled Singapore to discover customs fraud; for example, Singapore identified underpayment of duty for cigarettes through analytics that showed that the weight of goods declared in a declaration for cigarettes was lower than the historical norm. Singapore Customs is applying AI in machine vision to read the country’s library of X-ray images of containers, in order to automate the detection of anomalous patterns. Japanese customs has used AI similarly since 2017 to analyze X-ray images of postal shipments.

 

• The Dutch Customs Agency (DCA) has worked since 2018 to pilot web-crawling to compare data on customs declarations with the same items from ecommerce platforms. DCA had found that about one in three customs declaration has errors or is fraudulent, and goods descriptions for the same item vary widely, such as “smartphone” or “cell phone”. By using web-crawling to match prices and weight of items declared with items and their weight that are sold on marketplaces, DCA was able to uncover the likelihood of under- and over-valuation and also identify need for the item to pay VAT and qualify for de minimis.

 

• The Jamaica Customs Agency (JCA) is as of January 2020 using AI to enable online buyers of goods to estimate duties for items sent by ecommerce – only before they are sent, so that the Agency, courier companies, and residents of Jamaica can have a sense of tariffs of inbound ecommerce shipments before they are shipped. The estimator is straight-forward: the user goes on JCA website, inserts the url of the website the item is selling on, and obtains the item’s 10-digital HS code and import duty and all other taxes and fees associated with it (environmental tax, general consumption tax, and standards compliance fee and stamp duty). This enables buyers and couriers to anticipate the total cost of delivery and customs to manage risk and revenue. The service is offered via a partnership with Semantics3 and can be accessed via the Jamaica Customs website, ASYCUDA World portal or directly using the website.

 

Many other governments are leveraging AI in customs clearance, in tandem with blockchain. For example:

 

• The Indian government announced in 2019 plans to introduce “machine clearance” in its customs, using blockchain, machine learning, and artificial intelligence. A pilot was launched in December 2019 for select importers; 3,800 of them were authorized economic operators (AEO) and made up 40 percent import volume. The effort is expected to lower clearance times from 12 hours to 12 minutes. Indian customs also plans to use image scanners for cargo that does not pass initial machine clearance and keep the data as intelligence for future reference on blockchain.  

 

• Pakistan announced in 2019 plans to implement an AI-driven analytics of customs declarations. The system would issue automatic alarm for cargo/declaration mismatch based on HS codes, and help determine replaced cargo, misdeclaration and evasion without physical and intrusive examination.

 

• In 2020 Japanese finance ministry rolled out an AI-driven system for tackling customs fraud. It typically takes a customs officer about 10 seconds of studying an X-ray image to judge whether a package might contain narcotics or other contraband; the AI-driven system reduces that time to a fraction of a second. The system is being rolled out at six sites capable of handling 100 million international pieces annually – including Haneda, Narita and Kansai international airports.

 

Lessons from Blockchain and AI Pilots

 

There is growing interest in using blockchain and AI in border management. While many pilots are nascent and impacts remain to be assessed, there are also valuable lessons-learned:

 

• AI and blockchain have great promise for border agencies to balance the objectives of securing and facilitating trade and collecting revenue. Governments’ interest and pilots of new technologies are exploding and so far, pilots across the board have delivered positive impacts – and, positively, enabled governments and private sector stakeholders to learn about the process of adopting these technologies and their potential for solving many other problems than the early pilots have focused on.     

 

• Defining the problem and solution, engaging all key stakeholders, and conceptualizing process are more critical than the technology solution itself. The most important element in adopting new technologies is people. The sponsors of reforms need to first and foremost build consensuses on the need for, and benefits of, new technology solutions with agency staff and external stakeholders, and on the business models new technologies may require. This step is typically the most important and laborious. The private sector’s needs and ideas need to be brought to the table early on. For example, the U.S CBP has organized a “Pitch Day” for the private sector to present ideas to government officials on ways to use blockchain in customs. The CPB found that the Pitch Day was a great success in bringing numerous ideas forward.

 

• Pilots and POCs are key to understanding the power of technology and improving its application in the future. There is little value to contesting in the abstract the merits and downsides of blockchain or AI in solving a given problem. The best way for the  trade community and agencies to understand the power and limits of technologies and the processes, capabilities, and business models needed to use them is to run a pilot. For example, CADENA enabled Latin American customs to come up with various new use cases for blockchain, while CBP’s pilot enabled the U.S. trade community to understand how blockchain technology can simplify a complex customs procedures and where more research and testing is needed, to better operationalize blockchain in border processes. Pilots have been useful in identifying and testing common standards for connecting blockchain ledgers so that all firms and software could easily connect to customs in the future.

 

• Interoperability and data ownership issues are critical for the trade community to adopt and use blockchain in border clearance. Blockchain interoperability, specifications, and standards are crucial for implementing blockchain. CBP’s pilots have found that the best solution is to allow all participants to retain their own ledgers and to pursue interoperability among these ledgers. More generally, since companies hesitate to join blockchain platforms their competitors or peers have built, companies adopting blockchain do best by adopting joint venture structures and consortium models, as done with Tradelens.

 

• Attaining interoperability does not need to involve years of standards-setting discussions. The IDB has created a shortcut called LACChain, which is essentially a “motherboard” for parties’ blockchain ledgers in Latin America to connect to and interoperate, without the need for onerous one-by-one bilateral negotiations and standards-setting. Such a multilateral platform can also help overcome the free-riding problem that can occur in consortia and “vendor-locking” where a single vendor gets to operate the platform.

 

There are also some important limitations:

 

• Data on a blockchain is not necessarily accurate. Just because data is placed on the blockchain does not mean that it is accurate. Blockchain technology can add efficiency and transparency and can assist in managing and tracking information, but there should not be an expectation of built-in fraud prevention. Blockchain data tracking and data aggregation can help identify issues and target suspicious shipments more efficiently.

 

• Blockchain and AI are among many other potential solutions for customs and trade facilitation. There are many technologies that can be useful in their own way to facilitate trade in the ecommerce era. Robotic process automation can automate mundane and repetitive processes in border agencies, enabling agency staff to invest their time in serving users and performing other higher-value work. Internet of Things applications can enable detection of tampering with shipments. Smart contracts built on blockchains can eventually automate payment transactions from shippers to freight forwarders and customs brokers to customs and ports. More controversial, biometric technologies can help spot fraudulent traders and drones to keep track of the movement of trade.

​

By Kati Suominen, Founder and CEO, Nextrade Group and Techical Director, eTrade Alliance

​