On 4 July 2026 a Chinese coast guard task group led by the cutter Xiushan relieved one led by the Daishan in the water east of Taiwan, and continued what the service called law-enforcement patrols. Nothing was fired. No vessel was stopped. In most respects it was the least dramatic thing that happened in the strait that month.
It was also the most informative. A relief in place is how a service sustains a presence it means to keep, and it is a different act from a patrol. The handover said that the presence east of Taiwan was not an episode.
What it did not involve was the People’s Liberation Army Navy — the largest navy in the world by number of hulls, which was not required and would have been worse than useless. That choice is the subject of this note.
The key judgment
Beijing applies maritime pressure around Taiwan through a coast guard rather than a navy because the coast guard makes a legal claim that a warship cannot, and because it is structurally difficult to answer. A white hull conducting an inspection asserts that the waters are subject to Chinese jurisdiction. A grey hull doing the identical thing asserts only that China has ships there. The first is a claim about law; the second is a fact about power, and everyone already knows it.
The difficulty this creates for Taiwan is not tactical. It is categorical. A military response to an act framed as law enforcement escalates a situation that the other side has carefully kept below the military threshold. A law-enforcement response — sending Taiwan’s own coast guard to shadow and record — concedes the premise that the waters are subject to competing jurisdiction and that two agencies are policing the same sea. Neither answer is good, and there is no third.
That is the whole mechanism, and its value is entirely dependent on the ambiguity holding. This note’s secondary argument is that the ambiguity is consumable: every time the framing is stretched — by integrating with military exercises, by an incident that produces casualties, by being applied somewhere its plausibility is thinner — it costs something that cannot be easily replenished.
The instrument
The China Coast Guard was consolidated from several civilian maritime agencies from 2013 and placed under the People’s Armed Police in 2018, which put it under Central Military Commission authority. That is the structural fact underneath everything else: it is a law-enforcement service inside a military chain of command, and it presents as the former while reporting through the latter. CSIS notes that the 2021 legislation authorises it to carry out military operations in peacetime, without restricting that to wartime.
It is the largest coast guard in the world by tonnage, with cutters larger than many navies’ frigates. But the size is not the point, and reading this as a capability story gets it backwards. Against a warship these are lightly armed vessels. Their advantage is categorical: they generate pressure while remaining below the threshold that would justify a military answer, and their endurance matters because it converts presence into routine.
What PRC law asserts
Three layers, and the third is the one that does the work.
The 2021 Coast Guard Law authorises enforcement in “maritime areas under the jurisdiction of the People’s Republic of China”, and permits the use of weapons and the removal of foreign vessels.
The Provisions on Administrative Enforcement Procedures for Coast Guard Agencies — Order No. 3, in force from 15 June 2024 — set out the procedure: boarding, inspection and detention of foreign vessels and persons suspected of breaching Chinese law in those waters. Detention is reported at up to 30 days, extending to 60 where a case is treated as complicated.
Neither instrument defines the waters. That is the third layer. The framework avoids the Law of the Sea vocabulary — territorial sea, exclusive economic zone — that would carry defined rights and corresponding obligations, and substitutes a phrase that appears in no convention. The Lowy Institute’s reading, which Vigil finds persuasive, is that the vagueness is deliberate rather than a drafting failure: an undefined claim cannot be formally contested, cannot be mapped, and can be invoked or not invoked according to circumstance, while each individual act is presented as routine administration rather than as a new position.
Taiwan’s foreign ministry objected on 14 June 2024 that the provisions disregard the UN Convention on the Law of the Sea, were announced unilaterally, and would affect navigation safety for international commercial and fishing vessels.
Vigil describes what these instruments assert. It does not treat the assertion as established law, and it does not adjudicate the sovereignty question the assertion rests on. Both governments’ positions are recorded as positions.
Kinmen: a precedent, and its limits
The mechanism has been demonstrated once, and the sequence is worth setting out as a sequence rather than as an episode.
On 14 February 2024 an unregistered Chinese speedboat entered restricted waters around Kinmen, fled an inspection by Taiwan’s coast guard, and capsized. Two of the four people aboard died. Four days later Beijing announced increased patrols in the surrounding waters. Those patrols have recurred since, framed throughout as law enforcement; Taiwan’s coast guard had recorded scores of incursions within the following year.
An incident supplied an occasion. A jurisdictional claim was asserted on the back of it. Patrols and inspections were conducted under that claim. Repetition made them unremarkable. In May 2024 the pattern acquired a name — the “Kinmen model” — from a CCTV-affiliated media account, which reported an unnamed official saying off the record that it could be extended across the strait; Global Times wrote that the coast guard could refer to the model to establish normalised patrols where it judged Chinese fishermen’s rights to have been infringed.
Kinmen is not Taiwan. The differences are not incidental and they are the reason this note treats it as a precedent for a mechanism rather than a template for a campaign. Kinmen lies a few kilometres off the mainland coast opposite Xiamen — closer to China than to the island that governs it. The boundary at issue there was already contested before February 2024, so an enforcement claim had somewhere to attach. The traffic is small craft and fishing vessels, not container ships on published schedules with international owners, charterers and insurers. And the waters are ones where Chinese vessels were already present for ordinary reasons.
None of that transfers automatically to shipping bound for Kaohsiung. Extension to Matsu or Penghu is an inference drawn by outside analysts, not a PRC statement. What Kinmen establishes is narrower and more important than a template: it establishes that the mechanism works — that jurisdiction can be asserted, enforced and normalised without a warship and without any threshold being crossed.
Justice Mission-2025: three things, kept apart
The exercises of 29–31 December 2025 are the largest recent data point, and they are also where analysis most often goes wrong, because three different kinds of fact get merged. Vigil separates them.
What Beijing declared. The Eastern Theater Command announced the exercise on the morning of 29 December, naming among its objectives combat-readiness patrols, seizing comprehensive superiority, three-dimensional external-line deterrence, and “blockading key ports and territory”. That last phrase is exercise language published by the PLA. It is not evidence that a blockade occurred, was attempted, or is planned, and no interference with commercial shipping was reported.
What was observed. Taiwan’s defence ministry reported 89 aircraft and 28 naval vessels on the first day, of which 13 were coast guard rather than navy, and 71 aircraft with a further 15 coast guard vessels on the second. Announced areas were published partly through Eastern Theater Command navigation warnings and partly through the maritime safety administration, with one announced by radio only; accounts differ on the total, with Taiwan’s ministry identifying seven zones and one compilation counting eight. Several announced areas overlapped waters Taiwan claims as territorial sea — and no Chinese vessel entered that territorial sea. Roughly ten rockets are reported to have impacted within the contiguous zone, twelve to twenty-four nautical miles out, the first such firings since 2022.
Those three facts — a declared zone’s geometry, a vessel’s actual entry, a munition’s impact — are routinely collapsed into “China entered Taiwan’s territorial waters”. They are different claims with different evidence and different legal weight.
What Vigil assesses. The significant feature for this note’s argument is the coast guard participation. Thirteen to fifteen white hulls operated inside a joint military exercise alongside navy, air force and rocket forces. CSIS identifies coordinated military–police exercises as an indicator of a quarantine capability being developed. It is also, and this cuts the other way, a cost: the law-enforcement framing depends on the coast guard appearing to act independently, and a pattern of joint operation makes that harder to maintain.
Why not the navy
Set the two instruments side by side at a strategic level.
The PLA Navy is unmistakably a military instrument. Anything it does to shipping is an act of one state’s armed forces against another’s commerce, which is the definition of a blockade in customary law and is generally regarded as an act of war. It has greater coercive capability by every physical measure. That capability is exactly what makes it unusable for pressure short of war: its involvement resolves the ambiguity that the pressure depends on, converts a policing dispute into a casus belli, and hands every third party a clear question with a clear answer.
The coast guard has less capability and more utility. It asserts law rather than force. It gives third-party governments an excuse not to act — the incident is a policing matter between two authorities, and states with no position on sovereignty are not obliged to take one. It gives commercial operators a reason to comply rather than to escalate: a shipping line that submits to an inspection has had a bad day, not a war.
None of which makes it harmless. The coast guard is not independent of the military system, it operates under Central Military Commission authority, and its legislation authorises military operations in peacetime. The distinction Vigil draws is about what the instrument does politically, not about whether it is benign.
Where the pressure actually lands
If enforcement were ever extended toward main-island traffic, the mechanism would not run primarily through governments. It would run through commercial decisions, and Vigil’s treatment of that changed during this sprint.
The clearest evidence is from August 2022, when exercises following the Pelosi visit included announced closure areas. Commercial traffic routed around them within days — on the announcement itself, before any assessment of whether the risk was real. That response is fast, documented, and driven by risk management rather than by verified danger.
Insurance behaves differently, and the module previously overstated this. Marine hull war-risk cover reprices at a threshold: when the Joint War Committee — the Lloyd’s and London company market body that maintains the Listed Areas — adds a water, standing war clauses trigger cancellation-and-reinstatement mechanics and premiums are negotiated per voyage as a percentage of hull value. In August 2022 that did not happen. Taiwanese waters were not listed at all, and Vigil has found no record of a listing since. The Red Sea shows what a listing does when one comes — rates moving from roughly 0.05 per cent of hull value to 0.5–1.0 per cent per voyage after December 2023, with transits falling sharply — but that is precedent for the mechanism, not a number transferable to Taiwan, and Vigil publishes no premium forecast.
So the honest statement is that rerouting is a fast and demonstrated response, and insurance repricing is a slow one that has not yet been triggered. They are different signals on different clocks, and treating premiums as a sensitive early indicator — as this module did until recently — is not supported by the one episode that tested it.
What would show the model is changing
These are the observables Vigil watches. They are not a score, and there is no composite index: the instruments differ in kind and summing them would imply a single dimension that does not exist.
- Inspection applied to main-port traffic. The clearest threshold in the file, and it remains uncrossed. Kinmen is small-craft enforcement in already contested water; a vessel bound for Kaohsiung stopped under the same claimed authority would be a different act.
- Boarding or detention procedures actually invoked against a commercial vessel. The authority has been published since June 2024 and, so far as Vigil can establish, never exercised against merchant traffic bound for Taiwan. One documented case would be more informative than a year of patrol reports.
- Duration of announced closure areas. August 2022 set the precedent at a few days; Justice Mission ran three. A closure that outlasts a shipping schedule stops being an exercise in commercial terms whatever it is called.
- Continuity of the eastern patrols. The July 2026 relief in place is the evidence so far. Further rotations would confirm a standing presence; a lapse would suggest a demonstration.
- Sustained PLA–coast guard integration. Joint operation extends reach and erodes the civilian framing simultaneously. Which effect dominates is genuinely open.
- A shift to administrative controls — clearance requirements, customs language, notification regimes — which would mark the move from episodic enforcement to a standing system.
- A Joint War Committee listing of Taiwanese waters: a clean, published binary, currently negative.
- Language convergence. The vocabulary has run well ahead of the action. Watch for the two meeting.
What would weaken this reading
The argument is falsifiable, and these are the observations that would damage it.
The navy becomes the primary instrument. If the PLA Navy took the lead in maritime enforcement around Taiwan, the claim that the coast guard is chosen for its legal and political properties would be wrong, or would have been overtaken.
The framing stops being maintained. If coast guard operations became transparently military in presentation — announced as military operations, conducted under military command in public — then the ambiguity would not be what Beijing is buying, and something else would be.
Commercial actors ignore it without consequence. If enforcement were extended and shipping simply continued, having priced the risk as negligible, then the mechanism would not deliver the compliance this note assumes gives it value.
Third parties treat quarantine and blockade identically. The mechanism’s premise is that a law-enforcement framing complicates other governments’ decisions. If the United States, Japan and others declared in advance that they would treat enforcement against Taiwan-bound shipping as they would a blockade — which CSIS suggests as a response — the ambiguity would stop paying.
The jurisdictional framing is abandoned. If Beijing stopped grounding activity in claimed domestic legal authority and asserted a straightforward political prerogative instead, the legal architecture described here would turn out to be decoration rather than mechanism.
What this does not establish
It does not establish that a quarantine or blockade is planned, prepared, intended or likely. Vigil publishes no timeline, no sequence, no probability and no force requirement, and their absence is deliberate rather than a gap awaiting research.
It does not establish that Kinmen scales. That is the single largest open question, and the differences of geography, traffic and pre-existing dispute are substantial enough that the burden runs the other way.
It does not establish what shipping would do. The 2022 episode lasted days and involved no interdiction; it says the commercial response is fast and says nothing reliable about behaviour under sustained pressure.
It does not establish Beijing’s intentions, which are not observable, and nothing above depends on any claim about them. The argument is about what an instrument is useful for, which can be inferred from its properties and its documented use. That is a weaker claim than a claim about purpose, and it is the strongest one the evidence supports.
Finally, capability is not intent, exercise language is not a policy decision, and a legal assertion is not accepted jurisdiction. This module’s whole classification model exists to keep those apart, and this note is written inside it.