I. Introduction

The deployment of Single Window platforms has transformed customs administration across developing economies by improving trade facilitation, revenue collection, and administrative efficiency. In Ghana, the Integrated Customs Management System (ICUMS) was introduced to modernise customs operations by replacing fragmented legacy systems with a unified digital platform. More recently, the Ghana Revenue Authority (GRA) has integrated artificial intelligence tools, including the Publican AI vetting system, to automate the detection of customs valuation anomalies. While these innovations have enhanced efficiency, they have also raised important legal and economic questions about the role of algorithmic decision-making in customs administration.

ICUMS increasingly relies on benchmark pricing, depreciation schedules, and other pre-programmed valuation data that may override genuine transaction values, even where importers present valid documentary evidence of legitimate purchases. This practice raises concerns about compliance with the tax laws, the WTO Agreement on Customs Valuation, and the broader principles of due process and the rule of law. This paper examines the legal and some economic implications of ICUMS's valuation practices and compares Ghana's approach with jurisdictions that use automated valuation systems as administrative risk-management tools rather than as determinative valuation mechanisms.

II. Legal Architecture of the Domestic and WTO Hierarchy of Customs Valuation

Ghana is a signatory to the World Trade Organisation (WTO) and has ratified the WTO Trade Facilitation Agreement (TFA), 2017 and the Agreement on Implementation of Article VII of the GATT, 1994 (WTO Valuation Agreement). In particular, Ghana codified the WTO Valuation Agreement domestically under Section 67 of the Customs Act, 2015 (Act 891). This law establishes a strict, mandatory sequence of six valuation methods. Customs authorities at the ports are legally prohibited from skipping any step in this hierarchy. They cannot jump to a subsequent method unless the preceding one has been thoroughly investigated and disproven through transparent, human-led administration. It is also important to add that under Article 1 of the WTO Valuation Agreement, the primary foundation for assessing customs duties is the Transaction Value, defined explicitly as the actual price paid or payable for the goods when sold for export to the country of importation.

The WTO explicitly forbids customs administrations from utilising arbitrary, fictitious, or, by extension, artificially generated values through the use of AI assistance systems. The WTO Valuation Agreement permits customs to discard the transaction value only under strict, sequential statutory exceptions:

a) Method 1: The Transaction Value (the Mandatory Baseline)

Method 1 serves as the mandatory primary baseline for all customs calculations. It is defined as the actual price paid or payable for the goods when sold for export to the country of importation, adjusted for specific costs like packaging or inland transport. Under international law, this real-world price, even if it reflects a massive discount won at a legitimate public auction, trade discounts,s etc., must be accepted unless Customs has verifiable, fact-based reasons to suspect fraud, forgery, or an uncompetitive relationship between buyer and seller.

To illustrate how this operates under strict statutory conditions, consider an importer who buys a clean-title 2025 Mercedes-Benz GLC 43 AMG at a genuine U.S. auction for an exceptional price of $12,000. Under international law, this real-world price of $12,000 must be accepted as the base for import duty calculations in Ghana, provided the transaction was finalised at arm's length. This method cannot be circumvented unless Customs provides the importer with written notice of their doubt and allows him to present evidence such as banking logs, wire transfers, and auction contracts, etc., confirming that the $12,000 was the actual consideration exchanged in the transaction that was legitimate and finalised at arm’s length.

b) Method 2: The Transaction Value of Identical Goods

Method 2 is triggered only when Method 1 has been legally discarded or, following a careful evaluation of all the circumstances, is determined to be inappropriate. Under Method 2, Customs evaluates the transaction value of physically identical goods, produced in the same country, by the same manufacturer, and exported at or about the same time as the item being cleared. Fundamentally, Method 2 cannot be utilised if the uniquely priced asset lacks an identical counterpart in the port's recent import history. Furthermore, it cannot be bypassed for Method 3 unless Customs proves that no identical goods have been imported under standard commercial conditions, preventing the state from ignoring localised pricing matches in favour of higher benchmarks.

Under Method 2 computation, if Customs proves that the $12,000 price was fraudulent or non-arm's length, they must evaluate the transaction value of physically identical goods, meaning another 2025 Mercedes-Benz GLC 43 AMG produced by the same manufacturer in the same country, and exported to Ghana at or about the same time. This method cannot be utilised if the uniquely priced asset lacks an identical counterpart in the port's recent import history. Furthermore, it cannot be bypassed under Method 3 unless Customs proves that no identical goods have been imported recently.

c) Method 3: The Transaction Value of Similar Goods

Method 3 is reserved for cases where identical items are not found. Thus, customs officials should proceed to Method 3 only after determining that Method 2 is inappropriate. This method focuses on vehicles or goods produced in the same country that, while not identical in all respects, share similar characteristics, components, and materials that enable them to perform the same functions and be commercially interchangeable. For example, a sports luxury vehicle from a competing brand with equivalent engine capabilities and manufacturing origins might be referenced. This method must be fully exhausted and proven inapplicable before moving to Method 4, ensuring that Customs actively seeks real-world market equivalents before reverting to theoretical calculations.

In our illustration, if no identical 2025 GLC 43 AMG data exists (i.e. if Method 2 is inapplicable), Customs would look at the real-world transaction values of imported vehicles with equivalent engine capabilities, luxury, and manufacturing origins. In this case, potential alternatives to consider may be the BMW X3 M50, BMW X4 M40i, Porsche Macan S, or Audi SQ5 Sportback.

d) Method 4: The Deductive Value Method

Method 4 shifts the valuation framework from export data, unlike methods 2 and 3, to local/domestic resale data. Here, the customs value is calculated based on the unit price at which the imported goods, or identical/similar imported goods, are sold in the greatest aggregate quantity within the importing country to unrelated buyers, minus standard local deductions like post-importation profits, transport costs, and domestic taxes. This method cannot be skipped for Method 5 unless the goods undergo significant transformation locally or are never resold in the domestic market, rendering an onward sales price impossible to verify.

The term "sold in the greatest aggregate quantity" refers to the specific unit price at which the highest total volume of physical goods is sold to unrelated buyers on the local market, rather than the highest price or the most frequent number of individual sales transactions. Under Section 67 of Act 891, customs identifies this baseline by aggregating all local sales records for identical or similar goods. For instance, if 500 units are sold at GHS1,000 per unit, 600 units at GHS800 per unit, and 400 units at GHS600 per unit, the unit price of GHS800 represents the price at which the greatest aggregate quantity is sold because the largest number of units, i.e. 600, is sold at that price. By targeting this dominant commercial rate, the law ensures that the valuation baseline reflects the true, bulk market reality of the business, protecting importers from having their customs duties calculated using an unrepresentative, inflated retail price or a single-item transaction value.

If applied to our vehicle illustration, Customs would look at the domestic retail price of that specific model in Ghanaian showrooms or auto dealers, subtracting standard local deductions like post-importation profits, transport costs, and domestic taxes to arrive at a backwards or reverse-calculated customs value.

e) Method 5: The Computed Value Method

Method 5 looks entirely backward at the cost of production. It determines value by summing the cost of manufacturing materials, fabrication processes, standard profits, and general expenses incurred by the producer in the country of export, along with international freight and insurance. Given that this method requires deep, voluntary access to a foreign manufacturer's private accounting books and cost structures, it is rarely feasible for used or auctioned goods. However, Customs is legally barred from jumping to the final step until they have formally established that this manufacturer data is unavailable.

To our vehicle illustration, if Method 5 is applicable, Customs will determine the value of the 2025 Mercedes-Benz GLC 43 AMG by summing the cost of manufacturing materials, fabrication processes, standard profits, and general expenses incurred by Mercedes-Benz in the country of export (the US), along with international freight and insurance. Only then may Customs resort to Method 6 after Method 5 has been determined to be inapplicable or infeasible.

f) Method 6: The Fall-back Method/Default Method

Method 6 represents the absolute last resort in the legal hierarchy. It permits the flexible application of the previous five methods using data available in the importing country, provided the calculation remains reasonable and transparent. It is under this final method that the internal databases, showroom Manufacturer’s Suggested Resale Prices (MSRPs), and automated benchmarks of the ICUMS platform legally sit.

Returning to our illustration, when the Integrated Customs Management System (ICUMS) at the Port of Tema flags the $12,000 invoice, bypasses the intermediate human checks (Methods 2 to 5), and automatically forces an assessment based on the car's original retail price of approximately $70,000, it is executing a Method 6 calculation. Because Method 6 is explicitly restricted to a last resort status, this automated practice, which immediately ignores the importer's genuine transaction record and applies a hardcoded showroom benchmark, represents a severe violation of both Act 891 and international treaty obligations.

All in all, the sequential nature of these six valuation methods also preserves an importer’s right to challenge the legal basis for any departure from the primary transaction value. Under section 67 of Act 891, and Article 16 of the WTO Customs Valuation Agreement, an importer may require Customs to explain the basis upon which the declared transaction value was rejected and an alternative method applied. The importer should, therefore, request a written explanation identifying the reasons for rejecting Method 1 and the data and parameters relied upon in applying the subsequent method.

Where applicable, Article 7.3 of the WTO Trade Facilitation Agreement may also permit the release of goods against payment of undisputed duties and the provision of security for the disputed amount, thereby reducing the risk of accumulating port-related costs. Supported by documentary evidence, including bank transfer records, commercial invoices, and manufacturer or supplier price information, the importer may then challenge the valuation through the applicable administrative and judicial mechanisms. The significance of the sequential hierarchy is, therefore, not merely procedural - it imposes a legal obligation on Customs, requiring each departure from the transaction value (Method 1) to be justified within the prescribed valuation framework rather than determined solely by an automated benchmark or database.

III. The Statutory Clearance Clock as an Enforcement Weapon

Importers challenging an algorithmically generated valuation under the Integrated Customs Management System (ICUMS) face a high-stakes race against time. This operational friction severely undermines the practical exercise of due process. The Customs Act, 2015 (Act 891) establishes strict clearance deadlines that run continuously, regardless of whether an importer is actively disputing an automated assessment. Under Sections 52, 53, and 54 of Act 891, general imported goods must be entered and cleared within seven days of discharge, failing which they are deposited into a State Warehouse. If they remain uncleared after a cumulative 21-day period, they are transferred to the Uncleared Cargo List (UCL) and become liable for disposal via public state auction.

Imported motor vehicles are subject to a distinct, yet equally unyielding, statutory regime. Under Section 59(1) of Act 891, a motor vehicle that is not entered and cleared within 60 days after final discharge is automatically forfeited to the State. Once forfeited, Section 59(2) empowers the Commissioner-General to sell or otherwise dispose of the vehicle. While this 60-day buffer appears generous on paper, it functions as a hard ceiling. An importer contesting an inflated ICUMS assessment frequently finds that the administrative machinery required to review the dispute moves far slower than the rapidly ticking statutory clock.

The temporal pressure created by these provisions becomes particularly acute where the importer seeks to challenge an algorithmically generated valuation. A formal objection, engagement with Customs, or escalation through the applicable valuation dispute-resolution mechanisms may require time to resolve, yet the statutory period for clearance does not necessarily cease merely because a valuation dispute has been initiated. Because the law contains no mechanism to automatically pause the clearance clock when a dispute is lodged, importers are forced into an unsustainable trilemma: Capitulate immediately to a disputed, inflated ICUMS benchmark to secure release; or provide an expensive bank guarantee or alternative security, if accessible, to stall enforcement; or continue contesting the assessment while watching the statutory clock run down to zero.

For motor vehicle importers, the stakes of this gamble are total forfeiture of the asset under Section 59 after sixty days of uncleared vehicles. Consequently, the core issue extends beyond the GRA's clear statutory authority to prevent port congestion. Rather, it lies in how these inflexible legal deadlines interact with unyielding automated systems such as the ICUMS. When an algorithmic benchmark dictates the import duties, and the legal clock refuses to pause for human review, the ICUMS ceases to be a mere administrative tool. It transforms into an enforcement weapon that weaponises time to compel compliance, raising profound questions about procedural fairness and the rule of law at the port.

IV. Global Comparative Analysis of AI Systems as Guides, Not Final Arbitrators

Analysing how automated systems function across different international customs environments highlights a distinct policy divide between Ghana and other major trade hubs. In Ghana, the Integrated Customs Management System (ICUMS), coupled with the Publican AI engine, operates effectively as a de facto final arbitrator at frontline points of entry. Although the GRA maintains that the Publican AI does not itself reject customs declarations and that final valuation decisions remain with Customs officers, the practical operation of the system at the port presents a materially different picture. Customs officers are reportedly constrained from finalising a valuation below the amount generated by the AI system, thereby significantly restricting the situational discretion ordinarily available to the frontline officer.

Thus, where an importer presents a genuine commercial invoice reflecting a transaction value below the AI-generated benchmark, the officer cannot simply accept the invoice and finalise the declaration at that value where the system prevents such override. The consequence is that the distinction between an algorithm that formally recommends a value and one that effectively determines it becomes largely semantic, if the human decision-maker cannot depart from the algorithmic output. Publican AI at the Tema Port, therefore, functions as a de facto final arbiter at the point of entry, while any meaningful human challenge is displaced to a subsequent and potentially protracted administrative appeal process. To avoid escalating daily port demurrage and storage charges while disputing the value, importers in Ghana are frequently backed into a corner where they must "Pay Under Protest" up front and fight for a refund later.

In contrast, the United States employs its Automated Commercial Environment (ACE) primarily as an electronic processing and risk-management infrastructure rather than as an autonomous valuation adjudicator. Where Customs and Border Protection (CBP) identifies an entry whose declared value appears inconsistent with available information, the system does not simply substitute a predetermined retail benchmark for the importer's declared transaction value. Instead, the matter may be referred for human review, including through a formal Request for Information on CBP Form 28, which expressly permits CBP to obtain information necessary for the proper appraisement of the goods. The importer may, therefore, be required to substantiate the declared value through commercial invoices, payment records, purchase agreements, and other relevant documentation. The U.S. model, consequently, illustrates a different relationship between automation and administrative discretion: that technology may identify transactions warranting scrutiny, but the determination of whether the declared value is legally acceptable remains subject to human examination and the applicable customs valuation rules.

Similarly, the European Union also provides a contrasting model in which digital customs systems are integrated into a broader risk-management framework rather than permitted to determine customs value autonomously. Through the Union's Customs Risk Management Framework and Customs Risk Management System (CRMS2), customs authorities use common risk criteria and shared data to identify consignments requiring closer scrutiny, while maintaining the legal hierarchy governing customs valuation. Under the Union Customs Code, the transaction value remains the primary basis of valuation. Where customs authorities have reasonable doubts that the declared transaction value accurately represents the amount paid or payable, they may request additional information and supporting documentation; only where those doubts remain unresolved may the transaction value be rejected and a subsequent valuation method applied.

The EU framework, therefore, illustrates a critical distinction between automated risk detection and legal decision-making: an algorithm may identify an anomalous declaration for examination, but the resulting valuation decision remains governed by the prescribed legal hierarchy and subject to human administrative review. The availability of customs guarantees under the Union Customs Code further provides mechanisms for securing potential customs debts, although the release of goods remains subject to the applicable customs requirements and risk controls.

V. Conclusion

Ghana's ICUMS represents a significant technological advancement in customs administration, improving revenue collection, efficiency, and the speed of international trade. Yet its increasing reliance on automated valuation and risk-assessment tools presents a critical rule-of-law challenge where algorithmic benchmarks may, in practice, constrain the application of the legally prescribed customs valuation hierarchy. The central lesson from the United States and European Union is not that customs administration should reject automation, but that automation must remain subordinate to law. Algorithms may identify anomalies and inform customs officers, but they should not effectively determine the legality of a declared transaction value or displace the sequential valuation methods prescribed by the WTO.

Ghana should, therefore, recalibrate the ICUMS and the Publican AI to preserve meaningful human discretion, transparent reasons for rejecting declared values, and an effective opportunity to challenge automated assessments before coercive enforcement consequences arise. In particular, a Release Under Bank Guarantee mechanism for bona fide valuation disputes would allow Customs to secure the State's revenue interests while permitting qualifying goods to be released pending resolution of the disputed valuation. Such a framework would bring technological efficiency into closer alignment with the principles of customs valuation, trade facilitation, and due process. Ultimately, the objective should not be to choose between technology and the rule of law. It should be to ensure that technology serves the law rather than silently becoming the law.

NOTE: The author is aware that Parliament of Ghana passed the Customs Bill, 2026 on July 30, 2026, which, at the time of writing, is awaiting Presidential assent and has, therefore, not yet come into force. Although the Bill, once assented to, will replace the existing customs legislation, its enactment would not, in principle, displace Ghana's obligations under the WTO Customs Valuation Agreement. Accordingly, unless the enacted legislation materially alters the domestic implementation of the WTO valuation framework, the central analysis in this paper, including the primacy of transaction value, the sequential application of alternative valuation methods, and the prohibition against arbitrary or fictitious values, remains applicable.

By Dr A. M. Mashood

Attorney-at-Law (Ghana & New York, USA)

Phone| 0244844565/+1 646-620-3233

Email | ma.abdulmumuni@gmail.com