This chapter discusses opportunities and challenges presented to customs administrations by information and communication technologies (ICT) and other modern technologies. It examines why some customs administrations struggle with low performance despite having implemented modern ICT for operations. It then discusses some potential causes of low performance, including persistent as well as newly created manual procedures in declaration processing, incomplete exploitation of ICT declaration processing systems’ functions, inadequate ICT support to enterprise- level management or back- end operations, and the inability to process and analyze data. Considering the root causes of low performance while technologies are available, the chapter also reviews the potential customs application of various disruptive technologies, such as data analytics, artificial intelligence, and scanned image analytics.
Opportunities and Challenges Provided by ICT in Customs Administrations
The COVID- 19 outbreak in 2020 reconfirmed for the world the benefits of ICT. Thanks to ICT use in customs procedures,1 customs administrations can be resilient against a raging pandemic like COVID- 19 and secure supply chains without compromising social distancing, trade facilitation, or levels of compliance. ICT are ever- evolving technologies, and customs administrations should continue to exploit the opportunities presented for reform and modernization. This should not be limited to customs procedures but also include the critical supports to the organization’s decision- making, including resource mobilization planning, institutional risk management, and enterprise performance assessment.
“Some customs administrations struggle with low performance despite having modern ICT.”
In some countries, customs administrations have implemented ICT for several years but did so in a perfunctory manner that did not derive much real benefit. There is a need for mitigation measures to identify the cost incurred by continuing the obsolete business process, conduct business process reengineering, and ensure effective change management among customs managers and staff. These mitigation measures may add more time and delay in ICT implementation but are worthy to reduce conflict cost and to maximize benefits of ICT use in the future.
Evolution of Digitization to Digitalization2
In the 20th century, the early ICT adopters, particularly the “communication” part, were mainly developed countries. As trade volume rose and with limited increases in human resources, these customs administrations turned to ICT automation as a means to clear goods efficiently with effective control.3 Progress in several other areas continued— for example, harmonizing the implementation of legislation, reducing face- to- face contact, keeping track of operations to fight corruption, transforming paper- based processes toward increasingly paperless customs, removing discretionary human intervention, and increasing accountability for decisions. Entering the 21st century, developing countries also started adopting ICT in their customs operations.
“Customs administrations are in the journey from digitization to digitalization.”
Early ICT initiatives by customs administrations were “digitization” efforts, rendering paper to digital artifacts first, while the transformative “digitalization” undertakings took longer. Early “digitization” efforts culminated in standalone systems with no interface to each other (see “Silo Mentality” in this chapter). Similarly, as the public became accustomed to the use of ICT, many trade- related other government agencies (OGAs) followed suit. Members of the trading community found themselves having to use a myriad of such systems, for example, import certificate application systems. These OGAs’ systems deal with the same goods consignment as customs while there is often no interface among these systems, creating redundant manual data entries and repetitious data residing in these systems.
As manual data input is not resource- efficient and has a high risk of input errors and data manipulation with corrupt motives, the silo systems are gradually being connected to each other and with customs clearance systems for electronic data exchange. Electronic data exchanges save resources in inputting, checking, and correcting data entry; they prevent data manipulation and secure data integrity and consistency by comparing and reconciling the systems’ data, which significantly enhances customs operations, for example, cargo traceability. In the 21st century, the internet became more ubiquitous, and customs clearance systems in developing countries were gradually upgraded to be web- based with increased interoperability with other ICT systems.4 Customs administrations seek better and wider connectivity in three ways: within the customs administration (for example, customs clearance systems and other silo systems and electronic devices, such as cargo tracking devices); between the customs ICT system and other national parties’ ICT systems (for example, tax- customs data exchange, trade single windows, port community systems); and between the customs administration and foreign partners (for example, foreign customs administrations, chamber of commerce, quarantine). Standardization of data models, such as the WCO data model,5 facilitates interface and interoperability across different ICT systems through a standard vocabulary, definition, format, and quality standards for exchanged data.
Several models capture such evolving stages of ICT use in organizations. This section introduces a model specifically designed for the customs area, the Digital Customs Maturity Model (DCMM),6 the concepts of which are summarized in Figure 7.1.
As indicated in the DCMM, the progress in a customs administration’s ICT adoption is guided by a three- staged “vision”: smart clearance, efficient risk management, and effective controls. It is also supported by two- staged policy instruments: data security and protection and business continuity plan. It is important to understand that ICT implementation is a continual journey based on the national priorities, policy considerations, and resource availability of each customs administration. The DCMM identified six stages of ICT maturity in customs administration, and many low- income countries struggling within the first two stages, “initiate” and “implement,” may consider DCMM as a benchmark for their everlasting journey to better use of ICT.
DCMM may appear to have ICT maturity centered on a workstream of customs clearance processes. Many customs administrations have worked on another workstream, customs back- end operations, including risk management profiling, databases supporting valuation, cargo inspection, post- clearance audits, tariff management, guarantees and bonds, warehouses, authorized economic operators’ schemes, and so on. Some countries call such systems “customs management systems” (CMS). Many customs administrations in developed and some developing countries have worked on another workstream, enterprise resource planning (ERP).7 ERP is for organizational management support, notably, assessing the current situation and supporting corporate- level decision-making, for example, investment planning and associated human resource reallocation. (Details are discussed in “Weak Enterprise- Level Management Support” in this chapter.) The interlinked ICT systems for the three workstreams help customs reform and modernize the “administration,” not only “procedures.”
Challenges, Pitfalls, and Mitigation Strategies
ICT has been adopted in customs administrations since the 1980s. There has been notable progress, but there are still many aspects of customs management that have not fully harnessed ICT technologies. For example, ICT systems have often been applied in limited scope, primarily to automate the customs clearance processing. Outside customs clearance processing, ICT can and should be applied to support management decision- making— the raison d’être of ICT investment for institutional modernization— but few developing countries’ customs administrations have used ICT in this area. The following paragraphs discuss major challenges and pitfalls in the use of ICT and also mitigation strategies against these.
Persistent Manual Procedures
Unlike decades ago, fewer customs administrations today face reluctance and deliberate sabotage of the introduction of ICT customs clearance systems. Nevertheless, several customs administrations are still hesitant to fully exploit the functionality of ICT systems; worse still, some introduce new manual processes that undermine the merits of ICT. The situation persistently creates vulnerability in control and corruption.
“Low-performing customs administrations continue manual procedures while ICT solutions are available.”
Face -Vetting
Several customs administrations persist in requiring traders to submit the hard copy of the declaration with a handwritten signature, along with hard copies of supporting documents to the customs office for processing the declaration. Until this is done and formally accepted by customs officers, customs does not start processing the declaration even if all the declared data are stored in the clearance processing ICT system. This practice, dubbed face-vetting, is obsolete and was vividly exposed through social distancing requirements in the COVID- 19 pandemic. Accepting electronic copies while allowing for deferred physical submis-sion8 can mitigate such practice. Where proper risk assessment is done, a substantive portion of consignments can be accorded the green channel (no control) treatment; thus, customs does not need the supporting documentation. Face-vetting at the outset of the declaration process is unnecessary.
Cargo Management
Through the use of customs clearance systems, declaration processing in many countries is automated, albeit to varying degrees. However, some developing countries do not activate the ICT modules for cargo management areas, notably those relating to manifest, transit, bonded warehousing, and temporary admission. For example, if the manifest is not well managed, customs controls supported by ICT are only on declared cargo while many cargoes may have arrived in the country but may not be declared (possibly smuggling). Cargo management is not related to goods’ value but the lack of it exposes the customs administration to other issues, such as the inability to trace and reconcile data, and vulnerability to revenue and other control leakage (for example, illicit drugs, explosives). It is also closely linked with physical release of the cargo.
Customs Valuation Pre-Declarations
Some customs administrations have introduced new manual procedures that undermine customs clearance automation. One example is “customs valuation pre- declaration,” in which traders are obliged to submit the value of their goods for customs verification prior to the goods’ arrival. Customs valuation pre- declarations may be a good trade facilitation practice as pre- arrival data processing on the condition that it is similarly automated and supported by the ICT system’s selectivity and that there is no duplication with selectivity on the same shipment before the import clearance. In some countries, this condition is not met where overall time required to release the imports may be reduced but traders’ time and cost to deal with customs would increase (because of 100 percent documentary verification at the customs valuation pre- declaration stage). Figure 7.2 compares ordinary import declaration, customs valuation pre-declaration, and pre-arrival declaration.
Silo Mentality
The divergent goals of customs, in terms of revenue collection, trade facilitation, border security, and so on, can induce a “silo mentality,” also known as “departmentalization,” in which there is a reluctance to share information outside one’s division and across the organization and there can be a tendency to increase one division’s output at the cost of other divisions’ results or the administration’s results. This has a negative impact on the corporate culture of customs and the efficiency and effectiveness of the administration. Divisions within customs administrations often do not want to lose their authority or influence within the administration, resulting in a fear of integration efforts or changes to existing applications and silo systems9 that do not interface with each other. These standalone systems often hold the same data as elsewhere, causing duplication and possible data inconsistencies.
The silo systems are often developed within divisions without considering customs- wide IT policies or the IT division’s support. These divisional systems may not comply with customs administration- wide IT policies, such as procurement, hardware/software licensing, after-service contracts, anti- virus software, standard data modeling and coding, data accessibility controls, and
data protections. Under such circumstances, the interoperability between the silo systems or with the customs clearance system is difficult. Creation of an inclusive ICT strategy covering the entire customs administration’s departments/divisions, operations, and services is a mitigation measure and will open a path to further reform and modernization.10
“Low-performing customs do not have an administration- wide ICT strategy, and divisions create their own silo- systems without interfacing with others.”
Weak Enterprise-Level Management Support
Customs administrations worldwide face the daunting challenge of modernization. For many administrations, technology has emerged as the platform for modernization and is a catalyst for various services to converge. To keep abreast with modernization, customs administrations are expected to go beyond automating customs procedures to leveraging existing and new technologies to transform into high- performance organizations, harnessing digitalization and improving service quality for stakeholders as well as contributing to a positive business climate that is conducive to national economic progress. There are indeed many corporate issues beyond procedures, for example, enterprise resource planning (ERP), institutional risk management, human resource management, organizational performance and productivity management, and training and development.
A commonly observed phenomenon in low- performing customs administrations is that ICT investment in corporate- level issues is marginalized. Some customs administrations have only focused on digitizing customs clearance procedures and not digitalizing the institution per se (WCO 2016). This approach may be a result of the nature of customs ICT projects, which is predominantly for thematic issues, such as trade facilitation, but it is myopic: failure in revenue collection, trade facilitation and enforcement, or any other problems of poor performing customs administrations are rooted in a lack of accurate management information. The result is that decisions on investment and resource allocation do not address root problems and poor performance can persist or be exacerbated.
Therefore, customs administrations should prioritize ICT investment for organizational reform and modernization in order to ensure management efficiency and to appropriately assess the performance of its operations. In addition, a focus on human resources development, the allocation of appropriate resources, and the implementation of appropriate training for customs officers is necessary for effective customs operations. Since customs officers require highly technical knowledge and practical expertise, these training programs must be aligned with actual policies and operations.
Many ICT initiatives and projects have shown that public sector organizations generally have different organizational structures and management cultures than their private sector counterparts. Nevertheless, private-sector management approaches present some good practices worth examining. Examples include comprehensive
ERP, strategic planning, human resource management, and breaking down silo-oriented structures to integrate key planning and management functions and structures to one enterprise- level process.
“Low-performing customs do not adopt ICT to improve management decision-making.”
In recognizing the tremendous benefits gained from ERP, several public sector organizations have, in recent decades, taken a significant step by implementing ERP systems in an effort to integrate structures and to provide better governance and transparency. An ERP system covers, among others, the following common functional areas:
Financial accounting: For example, general ledger, fixed assets, payables, receivables and collections, reconciliation, cash management, financial consolidation
Management accounting: For example, budgeting, costing, cost management, activity-based costing
Human resources: For example, recruiting, training, rostering, career record, payroll, benefits, retirement and pension plans, diversity management, sanction record
Project management: For example, project planning, resource planning, project costing, work breakdown structure, billing, time and expense, performance units, activity management
Data services: For example, various “ self- service” interfaces for staff and external stakeholders
The ERP system often incorporates best practices. An ERP adapted for the public sector is often dubbed as government resource planning (GRP) where the software structure, modularization, core algorithms, and main interfaces are specifically adapted for government agencies.
All the preceding functional areas are lacking in the majority of ICTs in low-performing customs administrations, and these areas are often the under-performing functions within the customs administration. One of the key benefits for customs administrations to adopt an ERP approach is that it helps in coordinating the disconnected and uncoordinated data, information, resources, and assets within the administration while helping to integrate the various computing systems to provide a seamless overview to enable strategic planning and decision-making. With ERP, customs management can have timely and accurate data collection to assess both the surrounding operating environment and customs performance, all of which can be used to make timely and reliable decisions at the corporate level.
Disconnect between the Customs Clearance Processes and Back-End Operations
Although most administrations have customs clearance systems, they often find it difficult to implement other complementary components, such as customs back- end operations, as a coherent use of ICT within the customs administration. Digitalization progress has been patchy, as the lion’s share of efforts has been on automating the customs declaration procedure. Quite often, information silos are built within the customs administration, where the declaration processing system is detached from other customs back- end operations, such as risk management, ERP, revenue accounting, or human resource planning. There has been less effort and investment in applying ICT to automate the back- end processing, improve front- and back- end coordination, and enhance organizational performance.
Selectivity
As described in detail in Chapter 5, customs clearance systems have selectivity modules of various effectiveness that filter declarations by predetermined selectivity criteria. If the selectivity criteria are poorly determined, the results will be poor targeting and excessive control with little results. “Silo Mentality” in this chapter explains that without proper management at the customs administration level, each division/unit may add its selectivity criteria11 and the selectivity may end up with more controls, which is contrary to the original objective of more targeting. Therefore, a cross- departmental risk management support ICT system, which is different from the customs clearance system but mutually interfaced, is necessary to improve the selectivity criteria management, including weighting by prioritization. Such risk management support systems can only function with quality information, notably control result reports, intelligence information, and data analysis, which many low- performing customs administrations lack.
Data Reconciliation
Another disjointed area is the poor data reconciliation during customs clearance processing. In reality, most customs clearance systems have this functionality, but some customs administrations do not activate it. Manifest and declaration data must be reconciled to ensure that all discharged cargo is covered by a customs regime. If there is a discrepancy in the reconciliation, it indicates a high risk of diversion (smuggling). Similarly, discrepancies between transit departure data and arrival data, warehousing in/out and inventory data, and temporary admission entry and exit data are all illustrative of the necessity for data reconciliation. Very interestingly, (1) these areas often belong to one of either the law enforcement division, cargo control division, or a division responsible for monitoring the physical state and movement of cargo and not to the import procedure division; (2) these areas are the most vulnerable to smuggling (entering without declaration) and diversion, and there is little information about the situational reality or control results; and (3) the lower the performance of customs, the greater the likelihood that these functions have not been activated.
“Low-performing customs do not use ICT in cargo management and data reconciliation.”
Tax–Customs Data Exchange
Many developing countries have launched trade single- window projects through which trade and declarations data are exchanged between customs ICT systems and OGAs’ systems, but several of them have not yet realized data exchange between tax administration systems and customs systems. These two administrations are very often under the same ministry, and tax– customs data exchange is a promising practice. Chapters 3, 5, and 6 deal with the advantages of customs– tax cooperation in revenue collection in more detail.
There are several options for realizing tax– customs data exchange. The keys to success are (1) using transactional data12 and not aggregate data and (2) data protection and privacy policies and practices in place to protect proprietary business data. Some countries (for example, Benin) create a common electronic platform in the Ministry of Finance through which one administration’s staff, duly authorized, can share and access data from the other administration’s system. Other countries like Cambodia have developed direct interfaces between the two systems. If budget is a concern, a quicker fix is to allow system access rights to members of the other administration’s staff (for example, customs risk management staff have access rights to the tax administration database).
Compliance with Data Privacy and Protection Legislation
Customs administrations handling confidential business information need to take appropriate measures to conserve the privacy and protection of the data, as required by legislation. Import declaration data contain businesses’ confidential information, which competitors may want to obtain— for example, the name, address, telephone number, and so on of the exporter and the manufacturer, the imported goods’ description, prices, and the quantity.13
In recent years, there has been a spate of new legislation in many countries regarding the protection of data and privacy rights— for example, the EU’s General Data Protection Regulation (GDPR) and the Personal Data Protection Act (PDPA). These laws have reshaped how public administrations approach data privacy and the protection of information. Under GDPR/PDPA, customs administrations are defined as “controllers and processors of personal data” and must have in place technical and organizational measures to ensure an appropriate level of security to ensure that there is no misuse, loss, unauthorized access, undesirable disclosure, and unauthorized alteration of data to prevent any risk of litigation. When assessing the appropriate level of security, customs administrations must consider the risks that data processing presents, particularly from accidental or unlawful destruction of, loss of, access to, or disclosure of personal data.
Due to the multiple challenges and opportunities that the GDPR/PDPA brings, customs must be proactive to prepare to meet these challenges, as compliance requires considerable effort in reforming how customs store, use, share, maintain, and record personal and other sensitive data, which requires significant changes to current processes and systems.
For this reason, customs administrations must ensure that all their ICT systems are well protected against both unauthorized internal usage, external attacks, and data leaks. To be GDPR/ PDPA compliant, the mechanism for treating personal data should integrate appropriate data protection principles and safeguards (for example, using pseudonymization or full anonymization where appropriate). Customs administrations must deploy their ICT systems with privacy in mind, for instance, using the highest- possible privacy settings by default, so that data sets are not automatically publicly available and cannot be used to identify an entity, natural, or legal person. No personal data may be processed unless this processing is done under one of six lawful bases: consent, contract, public task, vital interest, legitimate interest, or legal requirement. When the processing is based on consent, customs must have a provision so that the data owner has the right to revoke the consent at any time. Data exchanges are to be restricted to legitimate data receivers with the equivalent level of compliance with GDPR/PDPA.
“Strengthened privacy and data protection becomes an inevitable agenda for customs’ use of ICT.”
In addition, the structure of employees’ and contractors’ legal agreements under which they access or use data on the customs systems must protect the privacy of taxpayers and traders. Customs administrations must put in necessary measures to prevent data leaks and to quickly mitigate the negative impacts of such leaks in the event of an occurrence. All staff should understand the sensitivity of customs and trade data. Data protection and loss cannot be the responsibility of an individual unit or individual staff; it must be institutionally owned by senior management leading a cross- functional effort covering all stakeholders.
Inability to Process and Analyze Big Data
Customs administrations are accorded wide legislative powers to require economic operators to submit data which are mostly in structured form (manifest, declaration). In addition, semi- structured or unstructured data in form of X- ray and photo images, scans of supporting documents, video recordings, data from devices such as weighbridges, cargo tracking Global Positioning System (GPS), data from e- commerce parcels, and more add to the tsunami of data collected by customs administrations. The reality is customs administrations are ill equipped to leverage all these data. Furthermore, most data are not adequately shared but remain sequestered within the customs clearance system or other silo- systems until they are deleted to make way for new data. The lack of storage capacity, data mining, and analytical expertise creates a situation where customs is unable to make productive use of these data for improvements or refinements to customs processes.
The importance of leveraging customs data should be recognized not only by customs or its parent ministry but by the government as well. The range of trade-related data that customs holds places it in a unique position to leverage these data through data science in multiple varied ways beyond just the confines of customs. Through data anonymization, other parties including the public and private sectors can utilize such data to forecast emerging patterns and do better planning.
Customs administrations should seek more investments from their governments to derive the full benefit and even explore new revenue channels by monetizing the massive data they collect.
Poor Design and Management of ICT Projects
There have been many examples of failures in customs ICT projects, which were exacerbated by inadequate planning, inexperienced project teams, and poor design and implementation. There have been customs clearance ICT installation projects suspended due to a shortage of budget. Such overspending was often attributed to delays in project progress— for example, delays due to failure of equipment tendering or a government procurement agency’s scrutiny while external experts’ cost augments as they are idling for the next phase. Also, when there was no shortage of funding for the project, the lack of the prerequisite business process reengineering (BPR), the appropriate organizational restructuring, legal reform, and change management process resulted in a customs clearance ICT system in place but with little change in customs performance.
Often, ICT projects were left to the technical staff to direct, which led to unsuccessful results as there was an overemphasis on the technical aspects while neglecting the business and managerial needs. Instead of business improvement, these endeavors automated obsolete procedures and risk became irrelevant— digitization but not digitaliza-tion.14 Anecdotal experience indicates that digitization of obsolete procedures can block further customs reform and modernization. For example, often heard is “This business process is obsolete, but we invested a substantive amount in its digitization, and we do not want to be criticized for wasting this investment; thus, we need to continue using this business process” and “Changing the digitized business process is costly; thus, although we know the benefits of the new business process, we will keep using the current business process.”
“Experiences show that digitization of obsolete procedures can block further customs reform and modernization.”
Governance and Financing of Customs Clearance ICT Systems
Governance of customs clearance ICT systems is often left aside while it is key to success or failure and sets the characteristics of the system through spending and investment decisions. In many countries, the customs clearance ICT system is not only for customs revenue collection but also for cargo management and other purposes. There are many diverse system users.15 Each of them may request to improve the system for their own interest, but there should be a longer- term agenda, such as enhancing security (access control, cyber security), improving resilience against system interruption, business continuity planning (BCP), and disaster recovery planning. The question is how to coordinate and prioritize these requests. The governance models observed in the world are Ministry of Finance (MOF); special board (options: board members are composed of only MOF directors, mixture of ministries, mixture of representatives of both public- sector and private- sector users); special purposed company set up by law (options: fully owned by MOF, owned by shareholders including private sector); and outsourced private sector.
Financing decisions are often linked with governance (for example, the case of special purpose vehicles [SPVs])16 and are also important for sustainability of the system. For system installation, two funding sources are popular: government self- funding and donor funding. Except those in fragile countries that lack project management capacity, donor- funded projects are usually “ recipient- executing projects” where tendering and procurement responsibility belongs to the country subject to prior consent of the donor.17 Therefore, these two financing models are de facto similar.
Recently, a third option of financing has emerged: the build- operate- transfer type of public- private partnership ( BTO- PPP), where the government does not pay for system installation. The company will build and operate the system at its own cost and recuperate its investment (and return) through collecting a user fee from the private- sector users until the contract expires, when the system, infrastructure, and equipment are transferred to the beneficiary country.18 Since this is a bilateral contract between the government and the service provider, the contract’s contents are often not disclosed. A cautious approach is necessary before signing the contract, and attention should be paid to whether the user fee rate is reasonable, the agreed service levels (including upgrade) and how to assess if the service levels are met, how the knowledge and assets necessary to continue operating the system after the contract expiration are transferred to the customs administration, and how to define the detailed status of the expiration of the contract. The contracts often contain capacity development articles which however lack details, and general training, such as the customs valuation method, is provided but not much on knowledge transfer of management and maintenance of the ICT system in question. Customs clearance ICT is a country’s critical soft infrastructure and monopoly; tying the custom clearance system contract with other services (for example, transit management, X- ray scanning services, and so on) requires careful examination. There can be a risk that a country relies too much on the use of this company’s services. In other words, if these terms are clear and the money value meets the services, this can be an option.
Secured financing for the running and maintenance costs is critically important for sustainabili-ty of the system, for example, running the operation, maintenance of tariff tables, modifications based on new legal provisions, system debugs, improvements in useability, and system upgrades. Most low- income countries and some other countries (for example, Japan) collect a user fee that may be kept in a public– private trust fund or directly finance the system operator. Different countries have different fee types, rates, and fee mixes (for example, registration fee, annual subscription fee, data volume usage fee, and so on). Donors such as the World Bank, seeing the sustainability of the system as imperative, usually request the creation of a user fee schedule with collection methods, for which technical assistance is also provided. A “customs clearance fee” has a long history and has been accepted by GATT/WTO for decades.19 From the perspective of sustainability of the sys-tem,20 attention is needed to certain trade agreements that waive the user fees on goods originating in certain countries. In addition, although it is debatable whether the government can or should collect a user fee for compulsory use for taxation, this concept is very common in tax administration ICT system projects. Here, again, the characterization of the customs clearance ICT system and its governance model becomes important.
“The governance structure will characterize the customs ICT system through spending and investment decisions.”
Going beyond Digitization, Opportunities for Digitalization in Organizational Reform and Modernization
As explained earlier, some customs administrations seem stuck in digitization and unable to achieve digitalization, which is the leveraged use of ICT technologies and digitized data to create the real digital transformation. Digitization has played a key role in customs reform and modernization. As seen in Chapter 1, the customs administration is a multifaceted government agency, and the application of digitalization, done in a coherent and well- planned manner, is now a priority.
The first step is for customs administrations to undertake a comprehensive review of their ICT applications to manage the shift to where the leveraged use of innovative technologies is applied in a holistic and integrated manner. In this regard, publicly available literature provides examples of successful ICT practice in the customs context.21 Recognizing the transformational importance of ICT, it is useful for customs administrations to align their ICT investments with organization and national goals as well as the priority in a structured manner. This practice, commonly known as enterprise architecture (EA), supports digital transformation, ICT growth, and the modernization of ICT. EA provides a template for defining the objectives, standardizing business operations, and incorporating systems in different layers, and applying proper governance rules. The EA approach helps customs administrations design and build an integrated ICT environment to achieve desired benefits.
In this way, the linkage between organizational goals and priorities and ICT efforts becomes very clear and provides the necessary context in which management can exercise effective decision- making to leverage ICT. It enables a balanced and clear decision- making process, where the different levels— strategic, tactical, and operational— can be aligned. This avoids the pitfalls where the implementation of ICT is skewed toward operational and tactical aspects yet is underused in strategic planning, decision- making, performance management, and resource utilization.
Digitalization or digital technologies can be used to enhance customs administrations in the following manner:
Increased automated processing: Many customs administrations deploy a significant number of staff to process manifests and declarations lodged online, for example, to reconcile the information, verify the data with supporting documents, validate the goods’ classifications and valuation, and so on. There can be increased use of automated processing to reduce the manual processing. The lodged data can be digitally analyzed with increased accuracy using disruptive technologies described in the next section.
Changing the nature of declarations: With technological advancements, the amount of data required to be lodged by the declarant can be lessened by customs collecting the supporting document information from the sources instead of from the declarant.22 For example, the issuance of permits and licenses from other government agencies can be easily verified by customs, which only needs a reference number and not the license itself. This can eliminate the need for the declarant to enclose the permit or license with the declaration, thereby saving time and effort. The same can apply to certificates of origin (C/O). Through such source data verification (SDV), verification of authenticity that consumes customs resources can be reduced along with the volume of submissions. Going beyond this, the importer can simply send a message to customs that all the information is ready for customs clearance processing, and it invites customs to remotely visit the trader’s ICT server to audit the necessary documents. In this case, the nature of the process changes from “submission” to “declaration of the start of customs audit and control.”
Changing the location of processing: Technologies enable customs officers to operate remotely (for example, valuation and back- office functions). With modern ICT, such functions can be centralized in centers of excellence, which would solve persistent problems of inconsistent rulings and shortage of skilled staff for all the border posts.
Growing use of behavioral insights as a compliance tool: To continue leveraging the vast volume of data amassed by customs, growing numbers of customs administrations have reported the increased use of data mining and analytics to gain improved behavioral insights into trade and supply chain processes. Through “open data initiatives,”23 customs can share data and insights with other departments and ministries (for example, planning to design more practical economic policies and interventions).
Smarter compliance and risk management: Customs administrations have to take an increasingly proactive approach to compliance management and risk management and where possible seek to intervene proactively and at earlier stages in the import process rather than at the point where a declaration has been filed.
Introduction of governance by design: The increased availability and sharing of data now allow governance by design approaches to cover a variety of data sources, including use of blockchain technologies to secure trust in chains of information.
Disruptive Technologies: Features, Opportunities for Customs, and Challenges
Common Features of Disruptive Technologies and Implications for Customs Administrations
The term disruptive technology can be defined as one that displaces an established technology and “disrupts” the industry or a groundbreaking product that creates a completely new industry (Christensen et al. 2016). The principles of disruptive innovation create an opportunity for customs administrations and stakeholders to take a step back, analyze their current situations, and identify what areas can be improved and where opportunities exist that can benefit from innovative solutions and more.
The pace of technology in the private sector is always faster than in the public sector to advance its thirst for higher profits. Customs should recognize where disruptive technology can be used to keep pace with the private sector. Adapting and leveraging these technologies to tackle the evolving risks and threats is critical to customs’ future success. Table 7.1 indicates where and how disruptive technologies can contribute to mitigating the major risk and threats in customs operations.
Major Risks and Threats That Customs Faces and the Potential of Disruptive Technologies
Major Risks and Threats That Customs Faces and the Potential of Disruptive Technologies
External Driver Leading to Risks and Threats | Opportunities for Use of Disruptive Technologies | Examples |
---|---|---|
Increased volumes and complexities of international trade: For example, proliferation of free trade agreements (FTA); complex preferential rules of origin | Large | Web-enabled trusted exchange of electronic C/O between FTA partners |
New business models and requirements: E- commerce and small parcels; innovative methods of moving goods across borders and trade financing; crypto currency; tax base-erosion and profit-shifting | Large | Use of tokenization,24 for example, unpaid invoices as tokens, to open more financing options for small and medium-sized enterprises (SMEs), besides the traditional banks |
Increased security threats and organized crime: Terrorism; pandemics; illicit activities; financing terrorists and organized crimes through evasion and avoidance of duties and taxes; cross- border fiscal fraud; smuggling of drugs, prohibited goods; money laundering; and counterfeit goods | Large | Applying machine learning (ML) / artificial intelligence (AI), for example, enabling digital ID, for improved profiling and targeting, interception of content and traffic data, forensic analysis, detection, tracing and disrupting crimeware |
A new approach to the “border”: New measures for border control; authorized economic operation (AEO) initiatives; biosecurity | Large | Use of nonintrusive inspection (NII) technologies, Internet-of-Things (IOT) devices (drones, sensors, GPS) and biometrics for enhanced coordinated border management |
Diversified demands for control from society: Anti-corruption, equality, public health, biosecurity, fauna and flora, environmental concerns | Moderate to large | Use of paperless trade platforms (single window, port community systems) and social media, chatbots, to meet demands and expectations |
New trading patterns: Increased number of connected parties; trust; 3D printing | Large | Cloud computing, Federated Architecture (FA), 5G networks enhance connectivity |
Increase in revenue fraud: Threats related to duty and tax evasion and avoidance | Moderate to large | Leverage data mining, big data, AI for accurate classification and valuation, and fraud detection |
Major Risks and Threats That Customs Faces and the Potential of Disruptive Technologies
External Driver Leading to Risks and Threats | Opportunities for Use of Disruptive Technologies | Examples |
---|---|---|
Increased volumes and complexities of international trade: For example, proliferation of free trade agreements (FTA); complex preferential rules of origin | Large | Web-enabled trusted exchange of electronic C/O between FTA partners |
New business models and requirements: E- commerce and small parcels; innovative methods of moving goods across borders and trade financing; crypto currency; tax base-erosion and profit-shifting | Large | Use of tokenization,24 for example, unpaid invoices as tokens, to open more financing options for small and medium-sized enterprises (SMEs), besides the traditional banks |
Increased security threats and organized crime: Terrorism; pandemics; illicit activities; financing terrorists and organized crimes through evasion and avoidance of duties and taxes; cross- border fiscal fraud; smuggling of drugs, prohibited goods; money laundering; and counterfeit goods | Large | Applying machine learning (ML) / artificial intelligence (AI), for example, enabling digital ID, for improved profiling and targeting, interception of content and traffic data, forensic analysis, detection, tracing and disrupting crimeware |
A new approach to the “border”: New measures for border control; authorized economic operation (AEO) initiatives; biosecurity | Large | Use of nonintrusive inspection (NII) technologies, Internet-of-Things (IOT) devices (drones, sensors, GPS) and biometrics for enhanced coordinated border management |
Diversified demands for control from society: Anti-corruption, equality, public health, biosecurity, fauna and flora, environmental concerns | Moderate to large | Use of paperless trade platforms (single window, port community systems) and social media, chatbots, to meet demands and expectations |
New trading patterns: Increased number of connected parties; trust; 3D printing | Large | Cloud computing, Federated Architecture (FA), 5G networks enhance connectivity |
Increase in revenue fraud: Threats related to duty and tax evasion and avoidance | Moderate to large | Leverage data mining, big data, AI for accurate classification and valuation, and fraud detection |
Considerations in Applying Disruptive Technologies in Customs Administration
Disruptive technologies are very attractive, but there are four common considerations that apply to customs administration:
Customs have to discern between hype and the actual usefulness of these technologies. It is important that customs leaders be fully abreast of the promise and usefulness of such technologies as well as any pitfalls.
Cost- benefit or return on investment analysis should be carefully conducted and alternative options examined. Newer technologies may cost more than traditional ones, with the possibility of hidden costs. Hence, it is important for customs administrations to be clear- minded about the desired outcomes and how these technologies can deliver these in the most cost- beneficial manner. Customs also should be aware that the cost may be incurred not by it but by traders. Also, not taking newer technologies may be costly as nonproductive legacy technology continues and new opportunity misses.
Many of these disruptive technologies inherently bring forth data streams from new sources. This generates even more data. Rich data are always useful provided that customs builds the associated capabilities to harness the increased data. Failing that, customs would be swarmed in seas of data without gaining any benefits.
Lastly, technologies are tools and not the objectives. They are ever- evolving, and today’s disruptive technologies will become obsolete in the future. People are innovative: fraudsters can be beneficiaries of disruptive technologies; it is easily imagined that fraudsters will run artificial intelligence (AI) to find out how they can smuggle goods without being detected by customs.
Possible Implication of Disruptive Technologies to Fragile States25
While the customs clearance ICT system is not a disruptive technology, it may be the most critical element for fragile states (FS) customs administrations to properly collect revenue, do fiscal reporting, control commercial fraud, produce trade statistics, and fight against smuggling of firearms, illicit drugs, and other antisocial items. Nevertheless, FS have tremendous difficulty installing and running such ICT systems for several reasons— for example, budget constraints to build data center buildings and procure processing and network servers and other necessary equipment and recruit or outsource IT operators for hardware, data entry screen design, coding, and table and data maintenance; local customs staff do not have experience in working with ICT; and so on.
Disruptive technologies may provide possible support to FS customs administrations in several ways. Cloud computing would reduce the needs for a local data center, a full set of equipment, and IT operators by using a foreign server and IT operators— mitigating the constraints of infrastructure and budget. Table and data management, selectivity criteria management, and document verification can be jointly conducted with an outsourced company (as on- the- job- training [OJT]), where tariff classification would be supported by artificial intelligence with natural language processing. Physical inspection is supported by contracted foreign experts situated outside the country through real- time compressed video communication and augmented reality technology (as OJT). An X- scanner is provided by donors for security control and its operation, including assessment of the scanned image, and is supported by real- time compressed video communication and scanned image analytics with automated threat detection. Certainly, there are prerequisites to realize this, particularly the development of high- speed and secure telecommunication networks enabling cloud computing and other data exchange with foreign servers. If fiber optic communication is not available or not reliable, satellite data communication can be considered. Although cloud computing and satellite data communication incur costs, they may be cheaper and more reliable than building a local data center and procuring a processing server and equipment. In using outsourced services, a transparent and accountable service contract with clear service level agreements is desired; otherwise such a contract should be a term contract with clear exiting clauses including transfer of knowledge and facilities.
Data Analytics
As mentioned previously, customs administrations are voracious collectors of data. Three vectors— volume, variety, and velocity— are useful to understand how “big data” is very different from old school data management. “Volume” is commonly associated with big data because the volume of data handled by the customs administration becomes unprecedentedly large. With the advent of cross- border e- commerce (see Chapter 2) and the shift from “containerization to parcelization,” the volume of data submitted to customs will soar exponentially.26 “Velocity” is the measure of how fast the data are coming in. For example, a cross-border e- commerce operator has geared up its operations to process 16,000 packages per hour in China. The “variety” of data that customs can obtain reaches almost staggering and incomprehensible proportions, such as unstructured and semistructured scanned documents, X- ray images, video feeds, and GPS readings.
“Without data analytics capacity, customs would just be swarmed in seas of data without gaining any benefits.”
The ever- improving ability to mine big data by utilizing data analytics tools represents the big driver in trade today. Customs administrations have to rethink how they will use data to gain new insights or experiment by looking proac-tively into new questions. Data analytics is the process of examining raw data in order to identify patterns and draw conclusions. It is to obtain, cleanse, review, analyze, and pull insights from raw data for more effective operations and to provide support for better strategic decision-making.
Although there are different approaches and outcomes of data analytics based on the objectives, data availability, and resource availability, data analytics methods can be broadly grouped into the following four stages (Gartner analytics ascendancy model [Laney and Kart 2012]) (indicating an order of maturity assessed by difficulty and value gained):
Descriptive analysis: What happened and/or what is happening now based on historical and incoming data
Diagnostic analysis: Reviewing past performance to determine causes
Predictive analysis: An analysis of likely scenarios. The deliverables are usually a predictive forecast.
Prescriptive analysis: Reveals what should be done. This is the most valuable kind of analysis and usually results in recommendations for next steps.
In exploiting data analytics, there are a few major constraints that customs needs to consider and address the following:
Data storage and quality: Firstly the ability to store and archive the tsunami of data is a prerequisite. Not all data had been stored in the past, and storage of data incurs a cost, although storage has become increasingly affordable. With a better appreciation of its great intrinsic value, customs needs to invest in more data storage. Data quality is the next challenge where great attention is needed. Obtaining quality data can be achieved by data cleansing of the original source and correcting data issues during the extraction, transforming, and loading (ETL) phase.
Knowledgeable staff: As referenced in Chapter 3, this is an area where competency gaps particularly exist. Customs administrations need to create a conducive environment to advance data analytics by establishing multidis-ciplinary teams of trained data scientists. One approach currently being pursued by some customs administrations is the establishment of data analytics centers of excellence, staffed with data scientists skilled in data mining, algorithms, predictive analysis, probability models, and other techniques.27 They may report directly to the customs senior management in handling sensitive information. The recruitment of such qualified staff who are in high demand can be daunting; therefore, special employment schemes might be needed to attract them into customs work.
Adequacy of ICT systems: Additional investments in ICT equipment and software are needed for data analytics. These include large data storage and warehouses and software programming languages (such as R, which is open source). Customs should explore the possibility of using cloud- computing models to service its data analytics needs (see “Cloud Computing” later in this chapter).
Artificial Intelligence (AI)
Artificial intelligence (AI) traditionally refers to an artificial creation of humanlike intelligence that can learn, reason, plan, perceive, or process natural language. Machine learning (ML) is a particular approach to AI that makes use of learning algorithms to make inferences from data to learn new tasks, identify patterns, and make decisions with minimal human intervention. ML can be regarded as a method of data analytics that automates predictive analysis. The more data, the better the ML and logical output. Figure 7.3 depicts the relationship between big data analytics, machine learning, and AI.
Relationship among the Fields of Big Data Analytics, Machine Learning, and Artificial Intelligence
Source: Giordani 2018.Relationship among the Fields of Big Data Analytics, Machine Learning, and Artificial Intelligence
Source: Giordani 2018.Relationship among the Fields of Big Data Analytics, Machine Learning, and Artificial Intelligence
Source: Giordani 2018.ML can be conducted as supervised or unsupervised by humans. In supervised ML, “labeled data” are used— meaning these data are already tagged with the correct answer. Supervised ML learns from labeled training data and predicts outcomes for unforeseen data. Supervised ML from historical data will be very helpful for effective risk assessments and accurate targeting decisions. Unsupervised learning, on the other hand, does not have labeled outputs, so its goal is to allow the model to infer within a set of data points. An unsupervised ML approach has been tested in work on fraud detection, misclassification and underreporting declarations (de Roux et al. 2018).28 The obtained results apparently demonstrate that the model doesn’t miss on marking declarations as suspicious and labels previously undetected tax declarations as suspicious, increasing the operational efficiency in the tax supervision process without needing historical labeled data.
Use of AI presents a tremendous opportunity given the big data collection in customs. AI provides the ability to make sense of the vast, ever- increasing data to detect and predict patterns accurately and at greater speed than humans. AI’s potential contribution to customs is vast, including the following:
Modeling duties and taxes collection patterns to ensure the appropriate duties and taxes are collected at the border
Realizing auto- classification of HS code of commodities based on natural language processing, ensuring improved classification, and applying the right tariff rates29
Identifying anomalies more quickly and thereby enabling customs staff to focus on areas of noncompliance
Improving cargo selectivity and targeting using predictive analysis
Enhancing scanned image analysis to improve detection efficiency and augment operator effectiveness
Deploying chatbots with an intelligent knowledge base for better customer service
As mentioned earlier, data accuracy is critical for data analytics, and that applies for AI as well; as the IT adage goes, “garbage in, garbage out.” This critical weakness can sometimes be exploited by miscreants who wish to influence AI’s logical reasoning for their own nefarious reasons.
One common example is the spoofing of AI- enabled facial recognition. The introduction of fake or inaccurate data to customs could create faulty logical reasoning leading to wrong conclusions and a distrust of the accuracy of AI. Therefore, robust security policies and transparent redress mechanisms should be put in place to ensure the data integrity and ongoing improvement of AI processes.
“Data accuracy in machine reading is important; AI without it can become “garbage in, garbage out.”
The fact that AI logical reasoning is superior to humans in some ways will create significant fear of change in some customs staff. Certain staff positions could be replaced by AI. Therefore, the administration needs to guard against risks to implement such analysis and against possible sabotage. Robust change management and optimal resource reallocation plans should be well prepared.
Scanned Image Analytics
The advent of shipping containerization and the birth of “ inter- modalism”30 gave rise to the use of nonintrusive inspection (NII), particularly X- ray container scanners, in customs. The terrorist attacks in New York and the Pentagon on September 11, 2001, brought to the fore the urgent need to protect borders more stringently, and yet facilitate legitimate trade. A subsequent WCO resolution highlighted the critical role of NII in border security (WCO 2005). Moreover, the rise of cross- border e- commerce brings about new challenges: how to detect contraband in the voluminous parcels and curb revenue leakages and other breaches.
A typical X- ray container scanner can scan between 35 to 50 containers per hour,31 while a high- speed parcel scanner can screen 2,500 parcels per hour. The analysis of scanner images is a challenging visual task, even for trained image analysts. The scanned images tend to be cluttered, often with other objects that can closely resemble the targets of interest. In containers in which a variety of goods is present, the imaged objects are in varying and overlapping shades, complicating their interpretation. As such, human error, compounded by eye fatigue, increases the risk for undetected illicit cargo.
Automated image identification is a promising technology that can significantly aid the need for rapid and accurate analysis of scanned images. Current efforts to develop automated image identification and risk assessments with artificial intelligence (AI) are mainly in developmental or pilot stages, and widespread common use has not yet been realized.32 The major scanning technology providers are working on algorithms that will enable machines to recognize the catalogue of objects. Automated image analysis can be separated into image preprocessing and image understanding:
Image preprocessing is a broad category including any treatment made to an image in order to help understand it by either humans or algorithms. Image preprocessing includes image manipulation; image correction, quality improvement, and denoising; material discrimination segmentation; and threat image projection (TIP).
Image understanding concerns decisions that are made based on the image contents. It is split into automated threat detection (ATD)33 and automated contents verification (ACV).
By leveraging advanced data analytics and machine learning, ATD/ACV algorithms can be developed to achieve automated image analysis and identification. There are three steps for automated image analysis: collection of the images, the learning process for recognizing characteristics of images and automatic identification, and auto- detecting and flagging suspicious characteristics as inspection targets. Figure 7.4 shows a model process diagram for the X- ray cargo inspection process and possible uses of automated image analysis: assisted selection and assisted inspection (depending on the timing of scanning).
With ACV, the classification of the goods (HS code) can be automatically assessed from scanned images. Through this, ML- enabled ACV can help flag irregularities between the detected goods and the description in the customs declaration and riskier goods. For the ML, a critical constraint plaguing the wider use of automated image analysis is the lack of historical data sets of images of sizable volume for accurate ACV. Customs has often regarded X- ray scanned images as single use, and they are not stored or archived due to limited storage capacity. Today, customs’ inability to store can be easily overcome, as advanced image compression technology and much cheaper data storage ease cost concerns.
Another constraint with ML occurs in certain contracts where X-ray scanning is outsourced to private service providers and the scanned images are contractually their property. In such cases, customs administrations have no image database unless it is obtained from the provider, which might incur additional costs. Such contracts should be revised to ensure that all data belong to customs administrations.
Recently, customs administrations have been considering a central image analysis center so that all scanned images from border sites are centralized and experienced image analysts are pooled together. For this purpose, a unified file format (UFF) for X- ray images has been developed and adopted by all major scanner equipment manufacturers. The UFF aids the buildup of a nationwide centralized scanned image database where ML tools can run through images in sufficient quantities (usually millions) to build fairly accurate ATD algorithms.
While images of non- threat cargoes are abundant, images of threat cargoes are fewer, thereby forcing the reliance on staged threat images. This is a common problem in the application of supervised ML for image analysis, whereby the lack of reference images to “train” the algorithms affects the accuracy of the ATD. Recently, researchers have begun to test ML where staged threat images are projected into an innocuous stream of images while adding realistic variations to bring balance between the threat and non- threat algorithms. Through international cooperation of customs administrations, the buildup of a sufficiently large data set (containing anonymized data elements like cargo description and HS codes) is accelerated. This would further enhance the wider usage of this technology among customs administrations.
In the near future, it is anticipated that the use of “deep learning” methods, where feature extraction, representation, and classification are learned simultaneously, will show great promise. These types of methods have shown to achieve superior performance in visual tasks, including image categorization. It is perfectly reasonable to believe that these methods can and will outperform humans at visual inspections of X- ray images.
Tracking Devices
Tracking devices using a combination of “ radio- frequency identification” (RFID) and satellite navigation systems (for example, GPS) enable accurate identification of geographical location. Satellite- enabled GPS is now much more accurate, possibly pinpointing to within 30 centimeters. The GPS tracking technology allows customs to monitor transit cargo movement in real time and deter cargo diversion as an integral part of the electronic cargo tracking system (ECTS).34 Use of GPS devices has been incorporated as electronic locks or smart seals, which secure the cargo container, as well as provide tracking and monitoring functionalities. In the event of attempts to break the lock/seal or divert from the route, an alert is automatically triggered to customs.
The use of an RFID device with antenna and associated reader facilitates near-distance communication and data exchange with a more reasonable cost. The identifiable distance can be up to 100 meters. RFID devices or tags attached to cargo or pallets are very useful for inventory monitoring in customs bonded warehouses. Use of such RFID tags helps compile inventory logs automatically and prevent the theft of goods with high excise taxes, such as tobacco products. Also, an RFID device can be attached to the temporary admission signboard or regional transit signboard. By using this technology, verification of carnet and vehicle registration can be done more accurately and quickly without visually reading the signboard or scanning the bar codes.
There are two issues in the use of tracking devices, particularly transit tracking. First is the cost involved both in terms of capital outlay as well as in operating costs. Customs administrations often consider whether cargo tracking costs should be covered by themselves or by the economic operators. In a sense, cargo tracking (vehicle tracking) is a social infrastructure: the beneficiary should not be limited to customs but also include cargo owners (whereabout of their cargoes), truck owners (tracking of their vehicles and drivers), insurance companies (monitoring driving behavior), and police (speed control).35 Good system governance, similar to customs clearance ICT system governance including possible BTO-PPP36 (see “Governance and Financing of Customs Clearance ICT Systems”), would solve the issue of cost burden.
Second, transit movement may be through multiple companies if cargo tracking services are preferably on a regional basis. If it is only national, a cumbersome switching from one company’s service and device to the other company’s ones may be needed at the land border posts of two countries, which will cause delay and a queue at the border.37
Robotic Process Automation
Robotic process automation (RPA) is a newer form of business process automation, which leverages the ability of a machine or software to perform a preprogrammed task repetitively but with much greater efficiency than humans. RPA evolved along with AI/ML, such that a software robot or “bot” can be programmed to mimic most human– computer interactions to carry out error- free tasks at high volume and speed.38 RPA bots can log into applications; handle high- volume, repetitious tasks that include queries, calculations, and maintenance of records and transactions; and then log out. RPA can free up staff from doing menial repetitive work to be deployed to other value- added work such as analysis, operational controls, post-clearance audits, and stakeholders’ engagements.
What distinguishes RPA from traditional business process automation is the ability to be aware of and adapt to changing circumstances, exceptions, and new situations. Once RPA software has been trained to capture and interpret the actions of specific processes, it can then manipulate data, trigger responses, initiate new actions, and communicate with other systems autonomously.
The biggest drawback of RPA is that it cannot make decisions on its own. For example, it cannot decide what is correct; that intelligence needs to come from a human or an AI. Another constraint is its limited ability to deal with dynamic or unexpected changes. RPA works best in steady environments, where the business processes do not change, and interfaces and data formats remain static. Because of this, RPA is best exploited within a narrow set of customs operations, such as the following:
Enabling better customer client service (for example, telephone help desk, complaints, corruption telephone hotline);
Performing rapid and accurate data entry (for example, Optical Code Reader [OCR] bot reading commercial invoices which have multiple diverse templates);
Ensuring business processes comply with regulations and standards (for example, declared value checked against the value on the commercial invoice, detecting major errors, and checking that necessary supporting documents are all submitted);
Allowing processes to be completed more rapidly (for example, producing regular reports);
Providing improved efficiency by digitizing and auditing process data (for example, mining useful data from the accounting books and records)
With the emerging development of cognitive robotic process automation (CRPA) software bots, RPA platforms can automate perceptual and judgment-based tasks through the integration of multiple cognitive capabilities, including natural language processing, ML, and speech recognition. The integration of cognitive technologies is extending RPA to new areas and can help customs administrations to become more efficient and agile in their digital transformation journey. While promising, CRPA is still in its early days and the relationship between RPA and AI is still not fully mature.
Cloud Computing
Cloud computing is the on- demand availability of computer system resources, especially data storage and computing power, without direct active management by the user. Cloud computing has effectively solved the financial and infrastruc-tural problems associated with operating and maintaining software applications, as it eases the total cost of ownership previously required. Advantages of using cloud computing services (such as software as a service [SaaS], platform as a service [PaaS], and infrastructure as a service [IaaS]) go beyond costs. The time to develop specific software applications that often take months can be drastically reduced in a cloud computing environment, and the development tools and environment are centralized in the cloud.
Another benefit is the improved efficiency of IT resources via cloud virtualiza-tion. The services approach/ pay- per- use model of cloud computing services affords clients greater flexibility based on their budget and need. Also, cloud computing services allow access to systems and application via multiple devices.
One of the chief concerns in customs’ use of cloud computing is the legal framework linked to sovereignty, security, and privacy. Customs administrations are custodians of sensitive trade and citizen data under the national legal framework and keeping the public trust in its security, privacy and confidentiality is paramount. If there is any data loss, theft or manipulation in the cloud infrastructure located in a foreign country, there are concerns as to how customs can enforce cross- border control and seek legal redress. The legal framework has not caught up with the technology evolution and its use. Work needs to be done to sufficiently protect the government’s ability to enforce their laws over foreign cloud service providers and transgressors.
There have been various efforts to address these concerns, albeit partially.39 Some governments have stipulated that any organization wanting to utilize cloud computing either must use cloud services that have data centers located within the country or must keep local copies of all records.40 More governments have found that adopting private and hybrid clouds helps alleviate some of these constraints, such as government private cloud facilities. At the same time, the COVID- 19 pandemic has amplified the usefulness of cloud computing technology for the public sector.
Data privacy concerns, especially in the light of more stringent GDPR/PDPA, present another potential constraint. The legal definition of “personal data” can be much broader in some countries and jurisdictions, and therefore hosting such data in a foreign country server is problematic. This situation surfaced in the case of advance passenger information data when one economic bloc refused to share the passenger’s name record with the other foreign authorities because this economic bloc considers the foreign private data protection to be inferior.41
Blockchain
Blockchain (BC), popularized by cryptocurrency, has been hailed as a technology with significant potential for disruptive innovations in international trade. BC enables many parties to collectively work on transactions and share information securely as any log records of modification/processing of information are easily verifiable by comparing the log information of all parties. BC- based technology is best used for a transaction with many uses through distributed ledger technologies (DLT), which allows customs, other government agencies, and the trading community to share data over a distributed ledger secured through cryptography. All parties keep the same ledger of records and activities, and any change to the ledger is automatically updated in the ledger of all parties. By doing so, the authenticity and trust of information is secured. Any attempts of forgery are easily detected. This also eliminates a single point of failure and inherently protects sensitive data.
BC can be applied to any authenticated documentation process and induce a significant impact not only on the regulatory process but also on trade financing (see Chapter 2). The parties involved will benefit from secure and trusted data exchange that is immutable, auditable, and tamper- proof. Several customs administrations have joined in BC pilots or have initiated their own projects:42 some explore a BC- enabled cross- border platform in which customs takes part for the exchange of e- certificates of origin or AEO certificates,43 while others examine how trade data, such as declaration information, can be exchanged securely using DLT.
One likely concern in BC, similar to third-party assurance and e- signature, is that while BC secures the information’s authenticity, it does not guarantee that information is correct.44 This can happen, for example, when both exporter and importer connive together to circumvent customs and tax authorities. It is not rare that authentic certificates of origin were delivered by the exporting country’s chamber of commerce based on forged information— the same can happen in BC. Another constraint of BC is that, theoretically, the information can still be forged, and the fraudulent information is maintained in the block if the majority of participating parties are colluding together.
“Blockchain secures the information’s authenticity, but it does not guarantee that the information is correct.”
BC is set to revolutionize international trade in the years to come. Going forward, many issues need to be discussed to resolve the challenges in bringing BC technology into widespread practice in customs administrations. These issues will need to be reviewed in light of the many ongoing pilot programs which are all at the “proof of concept” stage.
Augmented Reality and Virtual Reality
Augmented reality (AR) is an “interactive experience” of a real- world environment which is enhanced by computer- generated perceptual sensory information— for example visual, auditory, haptic, somatosensory, and/or olfactory. AR comprises data, graphics, audio, and other sensory interaction to form computer-generated images that are superimposed on a user’s view of the real world, thus providing a composite view. Virtual reality (VR), on the other hand, is a “simulated experience” that places the user in a created, virtual world that can be similar to or completely different from the real world.
AR and VR are edging their way into key areas of the public sector with the potential to transform the use of data, increase staff performance, and improve the efficiency of public services. Only recently, the core AR software and, most importantly, the devices that will deliver the AR experiences have begun to mature and find practical usage. These include handhelds and mobile devices, primarily smartphones and tablets, and built- for- purpose mobile workforce devices; head- up displays (HUDs) for windshields, screens, and visors; head-mounted displays (HMDs); glasses, goggles, visors, and helmets; contact lenses; virtual retina displays; and spatial displays.
Some possible applications for AR/VR in the customs environment include the following:
Remote inspection: Physical cargo inspection is usually conducted by customs in the presence of the trader for better transparency and accountability. AR/ VR would enable customs remote inspection via the trader using AR devices. Similarly, joint inspection at border posts can be done using AR, controlled remotely by experienced inspectors from different agencies, or located elsewhere.
Interactive with data: Using AR/VR to access and visualize data (including images and videos) in real time during the inspection or audit will save time for customs as well as the public user.
Simulation training: Training in mock environments (for example, imitations of an airport customs counter and a ship structure for search) has been conducted in several countries.45 This can be replaced with AR/VR, which can create realistic and immersive experiential situations in a controlled simulated environment to support training.46
Summary
Customs administrations can improve their performance by fully exploiting the potential of existing and new ICT systems to support not only declaration processing but also internal operations and management decision- making. In doing so, the leadership of customs administrations should be aware that ICT are enablers, so it is imperative to address ICT plans in a holistic manner that supports the overarching strategic plan, as well as to leverage the vast store of data and information collected by customs. Thus, data analytics capacity needs to be further expanded to better design and parametrize the entire customs operations. This should be in tandem with protecting privacy and preventing data breaches so as to maintain trust in the customs administration.
The same issues apply to disruptive technologies, such as AI and scanned image analytics. These offer significant opportunities to improve customs’ performance if, and only if, the benefits and outputs from these technologies and the use of the technologies themselves are clearly defined, agreed, and monitored. If not, the investments produce limited outcomes and become an exercise of acquiring very expensive gadgets that are not useful.
Hence, senior management of customs administrations play a critical role to drive the digital transformation of their complex organizations. Senior management needs to own and organize the proper structures, put in place the necessary supervisory mechanism to ensure that ICT and digitalization efforts are aligned with strategic directions, provide the requisite budget and legal support, assign the right persons responsibility for this work, and ultimately be accountable for the outcomes attained.
References
Beckett, S. 2015. “Robo-Journalism: How a Computer Describes a Sports Match.” BBC Click. https://www.bbc.com/news/technology-34204052.
Christensen, C. M., et al. 2016. “Disruptive Innovation: Intellectual History and Future Paths.” Working Paper 17–057, Harvard Business School.
de Roux, D., et al. (2018). “Tax Fraud Detection for Under- Reporting Declarations Using an Unsupervised Machine Learning Approach.” KDD, London. August 2018.
GATT. 1987. “United States Customs User Fee – Report by the Panel.” L/6264 (November 25, 1987), pp42–43.
Giordani, A. 2018 “Artificial Intelligence in Customs Risk Management for e- Commerce.” Delft University of Technology.
Hammer, H. 1990. “Reengineering Work: Don’t Automate, Obliterate.” Harvard Business Review (July-August 1990) pp104–112.
Laney, D., and L. Kart. 2012. “Emerging Role of the Data Scientist and the Art of Data Science.” Report G00227058. Stamford, CT: Gartner, Inc.
Rogers, T. W., et al. 2016. “Automated X- Ray Image Analysis for Cargo Security: Critical Review and Future Promise.” Journal of X-Ray Science and Technology (August).
UN/CEFACT. 2020. “Presentation on ’Machine Learning Approaches for Fraud Analytics in Customs,’ 35th UN/CEFACT Forum Webinar Advancements in AI toward Facilitating Cross Border Paperless Trade” (October 12, 2020).
WCO. 2005. “WCO Resolution on the Framework of Standards to Secure and Facilitate Global Trade” (June).
WCO. 2012. “Customs Cooperation Council Recommendation on the Dematerialization of supporting documents,” http://www.wcoomd.org/-/media/wco/public/fr/pdf/topics/wto-atf/dev/recommendation-2012-on-dematerialization-of-supporting-documents.pdf?la=fr).
WCO. 2016. “Summary Report: [213th/214th Sessions of the Permanent Technical Committee (11–14 October, 2016), Including the Joint Meeting of the Permanent Technical Committee and the WCO Working Group on the WTO Trade Facilitation Agreement (11 October, 2016)].”
WCO. 2018a. “IT Guide for Executives” (June 2018). http://www.wcoomd.org/-/media/wco/public/global/pdf/topics/facilitation/instruments-and-tools/tools/it-guide-for-executives/it-guide-executives.pdf?db=web.
WCO. 2018b. “Building technology-driven organizations: China Customs shares its experience.” WCO News #87 (October 2018).
WCO. 2020. “WCO Annual report 2019–2020” (June 2020). http://www.wcoomd.org/-/media/wco/public/global/pdf/media/annual-reports/annual-report-2019_2020.pdf.
Today, 99 percent of customs administrations have some form of ICT customs clearance system according to the author’s calculation based on the data in “WCO Members’ Profiles—Automated Clearance System” (WCO 2020).
For the purpose of this chapter, digitization is used as merely the conversion of analog to digital, and digitalization is defined as the use of digital technologies and digitized data to impact how work gets done and how the organization embarks on its digital transformation to create impactful results.
Early in the 1980s, United Nations Conference on Trade and Development (UNCTAD) launched the ASYCUDA, which is the customs clearance ICT program with the largest market share in the world (ASYCUDA software is free of charge), while other in-house and third-party products are also well used.
For example, this is the main purpose of migration from ASYCUDA++ to ASYCUDA World. The biggest difference is in the electronic message’s structure: fixed-length data elements with electronic message syntax were used before the internet while variable-length with tags for data elements were used with internet. Since all the data elements have tags, electronic data message syntax that is unique to each electronic message template is no longer used, significantly facilitating electronic message exchange.
WCO Homepage (http://www.wcoomd.org/DataModel), living contents and kept updating.
DCMM was proposed by the World Customs Organization (WCO) in 2017. More general models are, for example, the Software Engineering Institute’s Capability Maturity Model Integration (CMMI) and Google’s Digital Maturity Model.
For example, Jordan’s ERP is similar to or sometimes used as a synonym for business intelligence (BI) or management information (MI).
Submission of documents can be within a certain number of days after the start of the declaration.
Such silo systems may include valuation support databases, ofense databases, anti smuggling support databases, trader compliance record databases, dispute settlement tracking databases, exemption regime management databases, passenger management, licensing management, document management, human resource databases, payroll, and intranet for the dissemination of internal and administrative information.
For example, the time stamp data for control and the ID of the officer who did the control are stored in the customs clearance system and can be shared with the HR management system for performance assessment (for example, Cameroon).
For example, under-value, under-quantity, tariff slippage, origin fraud, eligibility fraud, intellectual property infringement, explosives, frearms, illicit drugs, and so on.
Transactional data are data linked to individual trade transactions, in other words, individual customs declarations. They are not publicly available because of trade secrecy.
Corruption cases were reported in Central Africa region, stating that customs officers sold conf-dential business data to the importer’s competitors.
Hammer (1990) has accused managers of having focused on the wrong issues, that is, technology in general—and, more specifically, information technology—has been used primarily for automating existing processes rather than using it as an enabler for making non-value-adding work obsolete.
For example, cargo owners, warehouse operators, road transport operators, customs brokers, banks, freight forwarders, maritime transport agencies, airline companies, guarantors, and OGAs.
An SPV is an entity created only for the purpose of execution of the project, which is different from the government agency or the private company while it/they may sponsor the SPV.
Tresholds are set by type of procurement whether explicit consent is required, or non-objection is obtained automatically after the certain period of time.
For example, Nigeria customs clearance ICT system and single-window systems in Benin and Côte d’Ivoire.
Sometimes called customs processing fee. General Agreement on Tarifs and Trade (GATT)/WTO accepts this provided that it is published, nondiscriminatory, and service- rendered.
A GATT panel interpreted that the exemption from the user fee granted to imports from certain countries increased the burden to the goods from the other countries (GATT 1987). Also, it was reported as inconsistent with the MFN obligation while this interpretation was not the disputed issue and the consistency with GATT Article XXIV (other regulations of commerce) is not clear.
Open data initiatives is an emerging trend among governments acknowledging that government data have many intrinsic values and that when they are made accessible to individuals, organizations, and even other government agencies, they can be promoted in new ways, innovations, and collaborations to realize their full potential.
Tokenization uses a database, called a token vault, which links the sensitive value with the token (a random set of characters). The sensitive data in the vault are often via encrypted and secured.
See more discussion on customs administration in fragile states in Chapter 1.
For example, in 2017, China Customs handled 1.89 billion parcels, inward and outward, and only on November 11, which is a popular shopping day in China (known as “Bachelors’ Day”), the country’s customs offices processed more than 16 million cross-border e-commerce shipments (WCO 2018b).
For example, WCO launches BACUDA (BAnd of CUstoms Data Analysts), a collaborative research project aiming to develop data analytics algorithms for customs administration.
Brazil customs, for example, applies AI in import declaration processing. It has had the AI learning the different types of irregularities in the declarations, for example, tariff classification, country of origin, eligibility of imports (licensing), eligibility of preferential duty rate, and exemption.
Pilots have started in China (UN/CEFACT 2020).
A concept that one transportation contract encompasses different modes of transport and transshipments—for example, ship, train, and road vehicles.
Tanks to a combination of radiography and computed tomography (CT), three-dimensional images are possible.
For example, the Netherlands and Japan respectively apply AI for image analytics in X-ray inspections while few results are obtained.
It is also known as automatic threat recognition (ATR).
Examples of ECTS implementations include Benin, Kenya, Mozambique, Nepal, Tailand, Togo, and Uganda.
For example, car insurance bargains based on the GPS tracking record is very common in the US.
In West African countries, a single-window operating company monopolizes peripheral logistic ICT infrastructure and services, including transit and X-ray scanning. There is a discussion that such a monopoly may increase efficiency but also increase the country’s dependency on one company and reduce contestability.
ECTS regional approach attempts are observed in the East African Community.
To date, there are three broad categories of bots in place: probots (simple, repeatable rules to process data); knowbots (bots that search the internet to gather and store user-specifed information); and chatbots (virtual agents who can respond to customer queries in real time).
Attempt to reduce the risk by limiting the client(s): “private cloud” is the only client while “community cloud” is designated parties, for example, registered traders and customs.
As in the case of New Zealand.
EU refuses to exchange PNR (passenger’s data of the air ticket purchase) with other countries while such exchange is recommended by the WCO and UN bodies. Several EU members authorities request PNR from the foreign countries while they do not allow parties to submit the data outside the EU. EU’s law covers all the data server located in the EU territory, including those for cross-border cloud computing.
Examples are avocado shipments from Kenya to Netherlands (led by TradeLens [IBM/Maersk]) and trade financing (seven Indian commercial banks). Asia-Pacific Economic Cooperation (APEC) also reports some pilots.
For example, Inter-American Development Bank (IADB) supported international exchange of AEO certifcates through CADENA project.
For example, some customs administrations complain that a certain country’s chamber of commerce’s certifcates of origin are not reliable and contain many errors. Such inaccurate information can be encrypted by BC and treated as the authentic information. This problem was reported when e- certifcate of origin backed up by e- signature was discussed. (Because of this problem, this project did not continue.)
For example, airport counter in Uzbek customs, ship structure in Japan customs.
It has been reported that Dutch customs has incorporated virtual reality as a tool for training its officers.