Hong Kong told payment firms to fix monitoring before the next inspection round

The HKMA’s thematic review on stored-value and payment licensees flags outsourced monitoring, dormant clients, and slow sanctions rescreening. The warning landed before the inspection wave, not after a fine.

Harbour and towers at blue hour
Hong Kong · Payments Hong Kong harbour at blue hour. The review is about payment licensees, not the skyline.

A review that arrives before the penalty

The Hong Kong Monetary Authority published a thematic review of anti-money-laundering controls at stored-value and payment licensees. The public point of that kind of review is to say what supervisors have already seen, in time for firms to fix it before inspections turn the same points into findings. The gaps described in the reporting of this review are practical. Monitoring that has been outsourced and then not checked. Dormant clients left on the book without a fresh look. Sanctions screening that does not rerun when lists change, or reruns so late that payments have already gone. None of that is a new idea. All of it is the difference between a licence that looks tidy and a firm that can show last month’s work. Faircroft treats a thematic review as a document, not as atmosphere. If you bank a Hong Kong payment company, or you clear for one, the review is now something your counterparty has read or should have read. “We are licensed by the HKMA” is not a response to a review that says licensees are missing these controls. The licence is why they are in scope.

Outsourcing is not a handoff

Payment firms buy monitoring from a vendor and then behave as if the vendor is the compliance officer. The review’s complaint, as it has been summarised in the supervisory material, is that the firm still owns the scenarios, the backlog, and the decision to close an alert. A contract does not close an alert. Ask the respondent who tunes the scenarios, who samples the output, and what happened the last time the vendor missed a name. If the answer is a slide with the vendor’s logo, you do not have an answer. Dormant clients are the same shape of problem. A wallet or an account that has not moved in a year is not low risk because it is quiet. Quiet is where a stale identity and a stale sanctions check go to live. The review wants a review of those names, on a clock, not a rule that says dormant means ignored. Sanctions rescreening fails on latency. Lists update. Customers do not sit still. A batch that runs weekly will clear a name on Monday that was designated on Tuesday and pay out on Wednesday. The control is the gap between the list and the payment, measured in hours where the product moves in hours. Stored-value products move in hours. A bank-shaped monthly rescreen pasted onto a wallet is the mismatch this review is pointing at.

What a correspondent should take from it

You do not need to be in Hong Kong to hold the risk. A European or American bank that provides the dollar leg, the card settlement, or the reserve account is downstream of these licensees. Ask whether the firm was in the thematic cohort. Ask for the date it mapped the review to its own gaps. Ask for one dormant-client review and one sanctions rescreen log, redacted. If the firm says the review does not apply to it, ask why, in a sentence a supervisor would accept. Payment companies are good at that sentence. The log is the test. Inspection waves follow thematic reviews. The point of publishing the gaps first is to make “we had not heard” unavailable later. A 2026 onboarding pack that cites a Hong Kong payment licence and does not cite this review is a pack from the year before the review. Update it.

What it is not

It is not a fine on a named wallet. It is not a finding that stored-value products are prohibited. It is not a reason to exit Hong Kong. It is a supervisor telling a licensed population, in advance, where the files are thin. Firms that fix the three points — outsourced monitoring with an owner, dormant clients actually reviewed, sanctions rerun at the speed of the product — will have a boring inspection. Firms that wait for the notice letter will have a longer one. Correspondents should be able to tell which of those two they are banking. The way to tell is to ask for the page, not the logo.

Three logs, and what “good” looks like

Ask for the outsourcing log first. It should name the vendor, the scenarios the firm can change without raising a ticket into a void, the last sample the firm pulled from the vendor’s output, and the person who accepted that sample. A sample the vendor marked as clean, which the firm did not read, is the gap. Second, the dormant book: how many customers had no movement for the period the firm itself defines as dormant, how many of those were reviewed, and how many were exited or restricted. A policy that defines dormant and a count of zero reviews is the finding waiting to be written up. Third, sanctions latency: the timestamp of the list update and the timestamp of the next screen against payments already in flight. If those timestamps are a week apart and the product settles in minutes, the control is decorative. Keep the three logs, or the refusal, next to the licence extract. When the inspection wave follows the thematic review, your file should already say what you asked in 2025 and what changed in the limit. Hong Kong payment companies will keep using the licence as the first sentence of every questionnaire. Your second sentence is the review. Firms that can answer it are the ones worth the operational work of a dollar account. Firms that cannot are telling you the inspection will be the first time anyone outside the vendor looks at the alerts. That is a fact about your risk, and it belongs in the approval, not in a footnote under the logo. Date the thematic review June 2025 in the pack so a 2026 renewal cannot pretend the document is still forthcoming. It is not. The inspection that follows it will assume you had a year to pull those three logs and to decide what the dollar account was for. Write that decision in one sentence: keep, cap, or exit.