Digital Assets & Virtual Assets
CARF in Hong Kong: What the Crypto-Asset Reporting Framework Means from 2027
How the IRD tests a Hong Kong settlement or re-invoicing entity: source of profits, the offshore claim, substance, transfer pricing and FSIE.
Using Hong Kong as a settlement centre is not simply a matter of incorporating a Hong Kong company and opening a multi-currency bank account. The legal analysis follows the functions actually performed by the Hong Kong entity, the contractual and physical flow of goods, the movement of funds, and the allocation of functions and risks across the group.
The distinction matters commercially, because there are two expensive ways to get it wrong. A structure with too little substance attracts an Inland Revenue Department challenge to its offshore profits claim, and may lose it. A structure with too much profit booked against too few functions attracts a transfer pricing adjustment, which may come in Hong Kong, in the counterparty's jurisdiction, or in both.
This guide sets out how the source and substance analysis works for a Hong Kong settlement or re-invoicing entity, where these structures come apart in practice, and what evidence holds up when the IRD asks for it.
The term is used loosely. In practice it covers several distinct models, and the tax analysis differs between them.
Many real structures are hybrids of two or more of these. In our experience the tax risk usually arises where the label given to the entity in the group's own documentation does not match what the entity's people actually do.
Hong Kong taxes profits arising in or derived from Hong Kong from a trade, profession or business carried on here. Profits sourced outside Hong Kong are generally not chargeable to profits tax, whether or not they are taxed anywhere else. That is the offshore claim. The generality is now qualified, because since 2023 the foreign-sourced income exemption regime deems certain categories of offshore income received in Hong Kong by a multinational group entity to be taxable here unless an exception is met. We return to that below.
The governing test, restated by the courts and adopted in DIPN 21, is deceptively short: one looks to see what the taxpayer has done to earn the profits in question, and where he has done it. The IRD calls this the operations test, and applies two refinements that do most of the work in practice. The first is that the focus falls on the effective causes of the profit rather than on antecedent or incidental matters, so preparatory steps, general oversight and the place where the board decided to enter a market are set aside. The second is that locality of profits is a hard, practical matter of fact. There is no formula, and each case turns on the nature of the business and of the transactions in question.
It is at least as important to be clear about what does not determine source.
The most common misconception in this area is that source is a structuring question. It is an evidence question about where work is done.
Most settlement centres are trading entities, and for a trading entity the IRD's starting point under DIPN 21 is the contract effected test: where were the contracts of purchase and of sale effected? "Effected" is read broadly. It covers negotiation, conclusion and execution rather than the technical point of contract formation, and the IRD accepts that email and messaging have made the formal rules of contract formation a poor proxy for commercial reality. The surrounding operations are examined alongside it, including how the goods were procured and stored, how orders were solicited and processed, how the goods were shipped, how financing was arranged and how payment was effected.
The IRD's working presumptions are worth stating plainly, because they set the burden a taxpayer has to displace.
One structural point surprises many groups: trading profits are not apportioned. DIPN 21 states it flatly, that trading profits will be either wholly taxable or wholly non-taxable. There is no partial credit for the overseas half of a genuinely split operation. The familiar 50:50 apportionment accepted for contract processing arrangements with Mainland manufacturers is not an exception to that rule, because those profits are treated as manufacturing profits rather than trading profits. In an import processing arrangement, where a related Mainland entity manufactures as a separate legal entity, the Hong Kong company's activity is trading and its profits are generally fully taxable with no apportionment at all. That all-or-nothing quality is why the structuring has to be right at the outset rather than argued afterwards.
The case law reinforces the same theme. In Li & Fung Trading the Court of Appeal accepted that the acts of overseas affiliates could be attributed to the Hong Kong taxpayer, because the evidence showed those affiliates genuinely carried out the profit-generating work abroad. The point cuts both ways, and the IRD has been explicit that the decision has no wider application to other source cases. In Datatronic and in CG Lighting the courts held that Mainland subsidiaries were not acting as agent for the Hong Kong taxpayer, and DIPN 21 warns that identifying an agent's acts with those of the principal should not be taken to an inappropriate degree. Attribution follows genuine agency and real activity, and it is a question of fact every time.
The practical analysis for a settlement centre comes down to mapping four things and checking that they tell the same story.
Who, physically, does what, and where are they? The question is not about job titles on an organisation chart. It is about the identifiable people who source suppliers, negotiate price, approve credit terms, decide volumes, handle quality disputes, manage shipping and chase payment. If the answer to most of those is that the work is done by group staff in another jurisdiction, the Hong Kong entity is thinly functional. That has consequences in both directions. It supports an offshore claim, but it also undermines any argument that the Hong Kong entity should book a material margin.
These are two separate maps and they frequently diverge. The contractual flow is who sells to whom, on what Incoterms, with title passing where. The physical flow is where the goods are made, stored and shipped, and whether they ever touch Hong Kong. Where goods never enter Hong Kong that supports an offshore position, but it is not decisive on its own, because the contract effected test asks where the deal was done rather than where the container went. Where goods are warehoused, consolidated or processed in Hong Kong, that is a significant onshore factor.
The related trap is document inconsistency. Bills of lading, customs declarations, certificates of origin, insurance documents and purchase orders are all created by people who are not thinking about tax. When the IRD reads them against an offshore claim and finds a Hong Kong shipping address, a Hong Kong-signed purchase order or a Hong Kong contact on the commercial invoice, the claim is weakened by documents the taxpayer created itself.
Banking location does not determine the source of trading profits, but the fund flow matters for three other reasons. First, it evidences the commercial reality of who bears credit risk, who funds working capital and who takes the currency exposure. Second, where the Hong Kong entity performs genuine treasury functions such as netting, hedging or intra-group lending, that is a separate service or financing activity with its own source analysis, and it is considerably more likely to be Hong Kong-sourced because the activity is performed here. Third, interest and other passive income generated on the settlement account has its own treatment and may run into the foreign-sourced income rules.
It is entirely possible to have offshore trading profits and onshore treasury income in the same company. That is not itself a problem. Failing to separate them in the accounts and in the return is.
This is the element most often left out of an offshore claim, and it is where the two regimes collide. Hong Kong's transfer pricing rules require related-party transactions to be priced on arm's length terms, and section 50AAF allows the IRD to adjust profits that are understated, or losses that are overstated, as a result of non-arm's-length dealings which confer a potential Hong Kong tax advantage.
The tension is structural and worth stating directly. The more forcefully a group argues that its Hong Kong entity performs no significant functions in Hong Kong, the harder it becomes to justify a large margin sitting in that entity at all. An entity with no people, no assets and no risks is not entitled, on transfer pricing principles, to a substantial residual profit. The offshore claim and the profit allocation therefore have to be designed together. A structure that wins the source argument and then loses a transfer pricing adjustment, whether in Hong Kong or in the counterparty's jurisdiction, has saved nothing.
On documentation, master file and local file obligations are subject to exemptions. A Hong Kong entity is exempt from preparing both where it meets any two of three conditions: total revenue not exceeding HK$400 million, total assets not exceeding HK$300 million, and an average of not more than 100 employees. Local file obligations are also tested category by category, with thresholds of HK$220 million for transfers of properties, HK$110 million each for transactions in financial assets and transfers of intangibles, and HK$44 million for other transactions, which includes services, royalties and rent. Both files are generally due within nine months of the accounting period end and records should be kept for at least seven years. Country-by-country reporting applies to groups with consolidated revenue of HK$6.8 billion or more.
Falling below the thresholds removes the documentation obligation. It does not remove the arm's length requirement itself, and smaller groups regularly conflate the two.
Hong Kong's foreign-sourced income exemption regime took effect on 1 January 2023 and was extended on 1 January 2024. It has changed the calculus for any structure sitting within a multinational group. In broad terms, specified foreign-sourced income received in Hong Kong by an MNE entity is deemed to be Hong Kong-sourced and chargeable unless an exception applies. The covered categories are interest, dividends, income from intellectual property, and disposal gains, which since 2024 extend to gains on property generally rather than only equity interests.
For a settlement centre the practical points are these.
Note the awkwardness this creates. The foreign-sourced income rules reward Hong Kong substance, while the offshore trading claim rewards the absence of Hong Kong operations. A group running both a trading claim and passive income through a single Hong Kong entity is holding two positions that pull in opposite directions, and needs to be able to explain why each is correct on its own facts.
Offshore claims attract close and routine scrutiny, and the courts continue to test them. In 2024 alone the Court of Appeal refined the source rules for sub-licensing royalties in Patrick Cox Asia, alongside several unsuccessful taxpayer challenges. The IRD publishes no statistics specific to offshore claims, so assertions about a rising rejection rate should be treated with caution. What is clear from practice is where these structures fail, and the failure points are consistent.
An offshore claim is made in the profits tax return and is commonly followed by a detailed enquiry. The questions probe the practical conduct of the business rather than its legal form, and they track the factors DIPN 21 adopts: who identified the customers, who sourced the suppliers, where negotiations took place, who set pricing, how instructions flowed day to day, and who bore the risk when something went wrong. For sample transactions, expect to produce:
What persuades is a consistent account corroborated across independent sets of documents. What fails is a legal argument constructed after the event and supported only by invoices. Contemporaneous records, meaning calendar entries, email headers, meeting notes and approval trails, carry far more weight than a well-drafted submission, because they were created before there was anything to prove.
Where the position is material and the facts are borderline, an advance ruling on locality of profits is available from the IRD under section 88A on payment of a cost-recovery fee. It buys certainty for the arrangement as described, which also means the description must be accurate and the business must continue to match it. Two limits are worth knowing before relying on one. A ruling on a recurring arrangement is generally valid only for the current year of assessment and, in general, for no more than two years of assessment from the year of issue. And it cannot be relied on as a precedent by anyone else.
If you are setting up or reviewing a Hong Kong settlement centre, work through the following before the first transaction rather than after the first enquiry letter.
Alan Wong LLP advises groups on Hong Kong settlement, re-invoicing, sourcing and treasury structures. Our work includes designing the functional and contractual arrangements, assessing the source of profits position and the strength of an offshore claim, aligning intercompany contracts with transfer pricing, testing exposure under the foreign-sourced income regime, and putting the evidence framework in place before an IRD enquiry rather than after one. We also act for businesses already under enquiry, responding to IRD questions and, where appropriate, pursuing advance rulings. Whether you are establishing a Hong Kong entity or stress-testing one you already have, please get in touch.
This article is general information current as at August 2026. It is not legal or tax advice, and the analysis of any particular structure depends on its own facts.
Disclaimer: This article is provided for general information only and does not constitute legal advice. It should not be relied upon as a substitute for specific legal advice on any particular matter. No solicitor-client relationship is created by your access to or use of this article. The law may change, and its application will depend on the specific facts and circumstances of each case. To the fullest extent permitted by law, we accept no responsibility for any loss or damage arising from reliance on this article.
Hong Kong's CARF starts 1 January 2027, first exchange 2028 - who must report, what's reported, penalties, and a readiness plan for crypto firms and holders.
Hong Kong's Protection of Critical Infrastructures (Computer Systems) Ordinance is in force from 1 January 2026 — who's caught, the key duties, and a practical checklist.