Norway’s Wild Aquatic Products Enter China: The Registration and Health Certificate Chain

  • Customs Notice 2026 No. 159 opened the Chinese market to Norwegian Wild Aquatic Products on 17 November 2026, effective from the date of issue, with no transition period and no annex of named establishments.
    海关总署 2026 年第 159 号公告自 2026 年 11 月 17 日发布之日起实施,挪威野生水产品自该日起获准输华,没有过渡期,也没有随附的企业名单。
  • The scope covers wild, human-consumption aquatic animal products and their preparations plus marine plant products such as algae, and expressly excludes amphibians, reptiles, aquatic mammals, live aquatic animals and breeding material.
    准入范围是野生的、供人类食用的水生动物产品及其制品,以及藻类等海洋植物产品及其制品;明确排除两栖类、爬行类、水生哺乳动物类、活水生动物及水生动植物繁殖材料。
  • Overseas Plant Registration is the first link in the chain and it runs on a Recommended Registration basis: the establishment must be approved and effectively supervised by the Norwegian authority, and recommended by it to China.
    链条的第一环是境外生产企业注册,走的是推荐注册的路子:企业须获挪威官方批准并受其有效监督,并由挪方向中方推荐。
  • An establishment that is not registered may not export to China at all, so the registration number is a gate rather than a formality.
    未经注册的企业完全不得向中国出口,因此注册编号是一道闸门,不是一道手续。
  • The Health Certificate is the second link: it is issued for every consignment, printed at least in Chinese and English with English mandatory, and completed with the producing establishment’s information.
    第二环是卫生证书:每批货物都要出具,至少以中文、英文印制且英文为必选语言,并完整填写生产企业信息。
  • A change to the certificate format must be filed with China at least one month before the new version takes effect, which turns certificate version control into a live operational task.
    证书格式如有变更,挪方须至少在新版本生效前一个月向中方备案;这使证书版本管理成为一项日常操作。
  • At declaration the China registration number goes into the product qualification field under licence code 519 and the use field must read for human consumption; a filing that omits them is not accepted.
    申报时须在报关单「产品资质」项下许可证类别代码 519 的证书栏填写在华注册编号,「用途」栏填报「食用」;未规范填报的,海关不接受申报。
  • Registration runs for five years and renews automatically on expiry, unless the product sits on the no-auto-renewal list, the establishment is under rectification, or China has suspended imports of the relevant food from the country.
    注册有效期为五年,届满自动延续;但产品列入不予自动延续注册清单、企业处于整改期间、或者中方已暂停该国相关食品进口的除外。

Norway’s Wild Aquatic Products Enter China: The Registration and Health Certificate Chain | 挪威野生水产品获准输华:读懂「注册 + 卫生证书」这条链

What the notice does, and what it leaves alone

On 17 November 2026 the General Administration of Customs issued Notice 2026 No. 159 on the inspection and quarantine requirements for wild aquatic products imported from Norway. It takes effect on the date of issue. There is no transition period, no phase-in by product line, and no annex listing approved establishments. The text is short, and almost all of it is addressed to two parties other than the importer: the Norwegian competent authority and the Norwegian establishment.

The legal base is a stack rather than a single instrument, and the way the stack is assembled tells you where to look when something goes wrong. Three statutes and their implementing regulations supply the authority: food safety, entry and exit animal and plant quarantine, and import and export commodity inspection. Two departmental measures supply the procedure: the measures on supervision and administration of import and export food safety, and the regulation on registration of overseas food production establishments. A bilateral protocol between the General Administration of Customs and Norway’s Ministry of Trade, Industry and Fisheries supplies the technical standard. The notice is the switch that turns the protocol on inside China.

Scope is defined by one inclusion and one exclusion. The inclusion is wild, human-consumption aquatic animal products and their preparations, together with marine plant products such as algae and their preparations. The exclusion is drawn tightly: amphibians, reptiles, aquatic mammals, live aquatic animals, and aquatic animal and plant breeding material all sit outside.

That exclusion list is where a surprising number of commercial conversations end. A supplier of live king crab is not covered by this notice even though king crab is wild and is eaten. Live aquatic animals move under a separate arrangement, and this notice does not disturb it. The same is true of breeding material such as salmon ova. What the notice covers is dead product — frozen, chilled, dried, salted, smoked, canned, and the algae lines that sit alongside them. An importer who reads only the headline, “Norwegian wild aquatic products are now permitted”, will overestimate what has opened.

What the notice does not do is worth stating with equal precision, because the gaps are where consultants invent requirements that do not exist. It creates no quota and touches no tariff line. It names no ports of entry. It provides no quarantine approval step of the kind that attaches to live animals; the obligations it does create are registration, product conditions and certification. It sets no advance notification period before shipment. And it imposes nothing on the Chinese importer by name — the duties in the text are addressed to the Norwegian side, which is precisely why the importer’s work is verification rather than application.

That asymmetry is the single most useful thing to understand about this notice. China has told Norway what Norway must guarantee. The importer’s job is to confirm, consignment by consignment, that the guarantee was actually given. Everything that follows is a description of how to do that.

It is also worth being clear about the instrument. A customs notice is an administrative document issued under the authority of the statutes and measures listed above; it does not create a new obligation that those instruments do not already support. That matters in two directions. It means the notice cannot be the whole answer to a compliance question, because the labelling standard it refers to, the registration rules it assumes and the residue limits it implies all live in other documents. And it means the notice is comparatively easy to amend: a change of scope or of certificate requirements can be made by a further notice rather than by a legislative process. Importers who treat the text as a fixed reference will be right for a while and then wrong without warning.

The practical translation is that this notice should be filed alongside the two measures it invokes, not on its own. An establishment’s registration file, the current version of the registration regulation and its implementing announcement, and the applicable Chinese food safety standards form a set. Reading the notice without that set produces a false sense of completeness, because the notice is deliberately silent on the parts it has delegated.

The first link: registration, and who moves it

The notice requires that establishments producing wild aquatic products for China — and the definition explicitly includes producing, processing and storing establishments — be approved by the Norwegian authority and remain under its effective supervision. Their food safety, hygiene management and protection systems must conform to Chinese and Norwegian food safety law. And they must be recommended by Norway to China for registration. The sentence that follows is the operative one: without registration, export to China is not permitted.

Three consequences follow, and each of them is routinely missed.

The first is that registration is a Norwegian act before it is a Chinese one. The establishment does not apply to Beijing. It is approved and recommended by its own competent authority, and China registers on that recommendation. An importer who asks a supplier “are you registered with China” and receives a confident yes has learned very little, because the follow-up question is whether the recommendation is still live. Recommendations can be withdrawn, and a registration can be suspended.

The second is that storage is inside the perimeter. The notice’s definition of establishment covers storage, and the implementing rules for the registration regulation separately bring cold stores used for storing aquatic products within the registration requirement. A Norwegian fishing vessel, a processing plant and a cold store may be three different registered entities, and a consignment may pass through all three. The registration number that appears in the documentation must correspond to the entity that actually performed the operation being certified.

The third is that the registration regime itself changed recently, and the change is favourable in one respect and unforgiving in another. The current regulation on registration of overseas food production establishments was published on 14 October 2025 and took effect on 1 June 2026, replacing the 2021 version. Under it, aquatic products sit on the catalogue of foods for which official recommendation is required — so the route described above is not optional for this trade. Registration runs for five years. At expiry it renews automatically for a further five, subject to three exceptions: the food is on the published no-auto-renewal list, the establishment is under rectification for non-compliance, or China has suspended imports of the relevant food from that country.

The no-auto-renewal list currently contains meat and meat products, and edible bird’s nest and its products. Aquatic products are not on it, which means a Norwegian establishment in good standing does not have to file a renewal application and does not have to watch a three-to-twelve-month window. That is a genuine convenience, and it is also a trap for the inattentive: an establishment that assumes it must file, and files unnecessarily, wastes effort; an establishment that assumes renewal is automatic while it is under rectification discovers at the worst moment that it is not.

Change of particulars is the sharper edge. Where an establishment moves premises, changes its legal representative, or has its Norwegian registration number changed — anything that materially affects its food safety and hygiene management and protection system — the registration cannot be amended. The establishment must apply afresh, and the existing China registration number lapses from the date of that notification. For an importer, this is a supplier-risk item with a date attached: a plant relocation announced in the trade press is not just a logistics event, it is an interruption in the registration number that appears on the certificate.

One structural feature of the current regulation is worth knowing even though it changes nothing about the obligation. Where a country’s food safety management system has been recognised by the General Administration of Customs, and one of a set of conditions is met — an existing food safety cooperation agreement, a broader agreement or memorandum containing food safety cooperation, or a positive risk assessment — the two authorities may agree in writing to use list registration. Under that route the competent authority submits a list of establishments in batches, together with the application information, a declaration that the listed establishments meet the registration conditions, and an undertaking to keep performing the obligations in the bilateral document. China reviews the list and registers those that qualify.

For a Norwegian exporter the relevance is procedural rather than substantive. List registration shortens the path for a group of establishments at once, and it makes the Norwegian authority’s own quality control the effective filter. For an importer it means that a supplier’s registration status may have been established in a batch rather than individually, which is neither better nor worse but does change what a verification exercise should look at: not whether the establishment applied correctly, but whether the authority’s list is still current.

Two administrative facts round out the picture. Registration is free — the implementing announcement states expressly that no fee is charged for the registration of overseas food and primary edible agricultural product establishments, which is worth knowing because intermediaries sometimes charge for it. And the registration system, at the address published by customs, is the channel for applications, changes, renewals, suspensions and restorations, and also the place where an establishment’s China registration number and validity can be checked against the published list. That published list, not a supplier’s assurance, is the reference a compliance file should cite.

The second link: the certificate, consignment by consignment

The certificate is where the two sides meet on paper. Norway is to inspect and quarantine each consignment and issue a health certificate in a format confirmed by both sides, attesting that the consignment meets Chinese and Norwegian food safety law and national standards. The certificate must be printed at least in Chinese and English, and English is mandatory when the form is completed. It must carry the producing establishment’s information in full.

Three details in that paragraph carry more weight than their length suggests.

“At least Chinese and English” sets a floor, not a ceiling. A certificate in Norwegian, Chinese and English is compliant. A certificate in Chinese only is not, because English is expressly mandatory. A certificate in Norwegian and English without Chinese is also non-compliant, because the floor is both languages. This sounds trivial until a shipment arrives with a certificate produced by a system that was configured for a different market.

“Complete establishment information” is what makes the certificate checkable against the registration database. A certificate that names the exporter and the consignee but describes the producer generically gives a border inspector nothing to match. In practice the registration number is the field that does the matching, and a certificate whose number differs from the one in the declaration is the most common single defect in this trade.

“Format confirmed by both sides” means the certificate is a controlled document, not a template that a Norwegian laboratory can adjust. And the notice adds the control mechanism explicitly: if the format changes, Norway must file with China at least one month before the new version takes effect. That one-month filing window is the reason a certificate version is a thing an importer should record, alongside its date of issue. A consignment certified on an old format after a change, or on a new format before China has been notified, sits in an uncomfortable position even if every substantive fact in it is true.

The certificate is also batch-specific, and this is the difference between a certificate and a qualification. A supplier’s registration qualifies the establishment; the certificate qualifies one consignment. A shipment split across two production lots needs the certification to match the split, and a consolidated container drawn from two establishments needs two certificates or a certificate that properly covers both. Consolidation is where small importers most often improvise, and it is where the improvisation is most visible.

Six product conditions that sit behind the certificate

The notice lists what Norway must ensure about the product itself. Read as a compliance document, the list is a set of things the certificate is asserting, and each one has an operational counterpart on the Norwegian side that an importer can ask about.

The first is provenance. The product must be caught legally in Norwegian or international waters, or come from a registered aquatic product establishment in a country or region that China has already opened. The second clause matters for re-exported material: a Norwegian processor that buys raw material from a third country can only put it into China-bound product if that third country is itself on China’s access list and the supplying establishment is registered. This is the point at which a Norwegian supplier’s own supply chain becomes part of the Chinese importer’s diligence.

The second is the absence of three defined problems: that products already shipped or intended for shipment do not seriously violate Chinese food safety law, national standards or the protocol; that no disease listed in China’s catalogue of animal quarantine diseases, or notifiable to the World Organisation for Animal Health and relevant to the protocol’s products, is occurring in Norway in a way that seriously affects safety; and that no major food safety incident, major public health event at an establishment, or contamination of fishing waters is doing the same. Each of the three is qualified by the words “seriously affects”, which is a judgement rather than a threshold. An importer should read them as triggers for a conversation when something happens in the Norwegian sector, not as boxes to be ticked once.

The third is pharmacologically active substances and additives. Neither party’s prohibited drugs or additives may be used, directly or indirectly, and restricted or permitted ones must be used as prescribed. The notice does not reproduce either party’s list, so the practical reference is the Chinese residue monitoring framework and whatever Norway publishes. For importers of farmed-then-released or enhanced stocks, this is the condition with the longest tail.

The fourth is the inspection result: no pathogenic micro-organisms, toxic or harmful substances or foreign matter listed in Chinese and Norwegian law, no disease from China’s catalogue, and no disease notifiable to the World Organisation for Animal Health. This is what the certificate attests to, and it is why the certificate cannot be issued before the testing is done.

The fifth is process control across the whole chain — fishing, processing, packing, storage, transport, transit and export — against Chinese and Norwegian safety and hygiene requirements and against traceability requirements. Traceability is the operative word. A consignment that cannot be traced from the Chinese port back to the catching area and the registered establishment has failed this condition even if every test result is clean.

The sixth is packaging and labelling: the packaging and label markings must comply with China’s national food safety standards and the rules on imported food safety administration. Note what this does and does not say. It incorporates the Chinese labelling standard by reference rather than restating it, which means the labelling question is answered by the standard and not by the notice. And because the product is a food, a separate provision of the registration regulation applies: a registered establishment exporting food to China must mark on the packaging either its China registration number or the number issued by its own competent authority. The notice does not repeat that requirement, but the notice and the regulation are read together at the border, which is why the number on the carton and the number in the declaration are expected to be the same number.

Where the chain usually breaks

Six failures account for most of the friction in this trade, and five of them are documentary.

The first is a mismatch between the registration number on the certificate and the registration number in the declaration. This happens when a supplier substitutes a sister establishment, when a plant has relocated and been re-registered, or when a trader copies a number from an earlier shipment. The customs system validates the number against the registration database; a number that does not correspond to the declared product and origin stops the filing.

The second is a product that falls inside the exclusion rather than the inclusion. Live aquatic animals, breeding material and aquatic mammals are the obvious cases. Less obvious is a preparation whose character has changed — a product so transformed that it is no longer described by the aquatic product category under which the establishment registered. Classification and registration are two different questions, and passing one does not answer the other.

The third is algae treated as an afterthought. Algae and other marine plant products are expressly inside the scope, and their preparations too. A trader who assumes that “plant” means “not an animal product, therefore outside the animal registration regime” will discover that the notice does not distinguish. The registration obligation attaches to the establishment that produces the product for China, whatever kingdom the product belongs to.

The fourth is language. Certificates printed in one language, or in two languages that do not include both Chinese and English, are refused on form rather than on substance. This is the cheapest failure to prevent and one of the most frequent, because certificate templates are configured centrally and changed rarely.

The fifth is a storage link that was never registered. A cold store in a transit hub, a consolidator’s facility, a second plant that repacks — any of these can be an establishment within the meaning of the rules, and an unregistered one breaks the chain even when the fishing and processing links are impeccable.

The sixth is not documentary at all: it is the assumption that access is permanent. The notice provides that for products with serious problems, or establishments with repeated nonconformities, China may enhance inspection and quarantine or suspend imports. A suspension is imposed on an establishment or a category, and it takes effect faster than a supply contract can be rewritten.

Declaration, and the five-year clock

At the border the notice’s conditions become data fields. Imported food must declare, under the product qualification heading of the customs declaration, the China registration number of the overseas establishment corresponding to the country of origin, in the certificate field for imported food overseas establishment registration, licence category code 519. The use field must state that the goods are for human consumption. A declaration that omits either is not accepted.

Two points about that requirement deserve emphasis. The first is that it is a validation, not a description: the number is checked against a database, so a plausible-looking number fails in the same way as a missing one. The second is that a false filing is treated as more than a clerical error — the implementing announcement provides that information falsely filed to obtain customs documents will be dealt with according to law. For a customs broker, this raises the cost of accepting a registration number on trust from a supplier.

Around the declaration sits the ordinary apparatus: the contract, the invoice, the packing list, the bill of lading, the certificate of origin where a preferential rate is claimed, and the health certificate. The health certificate is the document that ties the rest together, because it is the only one that speaks to the product’s safety rather than its value or its movement.

Behind all of this sits a clock that most importers never look at, and it is the registration clock. Five years, automatic renewal, three exceptions. The exceptions are the ones to monitor. Rectification is the most common: where China finds that a registered establishment no longer meets the requirements, it orders rectification within a set period, notifies the Norwegian authority to supervise it, and suspends imports from that establishment during the rectification. Once rectification is complete, the establishment submits a supervision report from its authority and a written declaration of conformity, and China reviews and, if satisfied, restores imports.

For an importer, the practical implication is that the registration status of a supplier is a live variable with three possible states — active, suspended pending rectification, and lapsed following a change of particulars — and that the state can change without any notice to the importer. The remedy is not vigilance for its own sake. It is a supplier file that records, for each establishment in the chain, the registration number, the last verification date and the source of the verification.

There is a second duty sitting on the importer, and it is easy to lose sight of because the notice does not mention it. Under the food safety law and its implementing framework, an importer of food is expected to establish and operate a system for reviewing its producing establishments, and not to import from an establishment that fails that review. Customs has restated the point in its own guidance, adding the practical instruction that importers should follow the published information on food refused entry, because a supplier’s pattern of nonconformity often appears there before it appears in a contract dispute.

Read together, the two duties describe a division of labour that is more demanding than either taken alone. The regulator verifies the establishment against a standard and publishes the outcome. The importer verifies the establishment against its own commercial and compliance requirements and keeps the record. Neither of those is satisfied by holding a certificate, because the certificate speaks to one consignment and the duties speak to a relationship.

It is also worth noting what the notice does after a failure, since it shapes how a first shipment should be structured. Nonconforming product is destroyed, returned or otherwise handled — three outcomes with very different cost profiles, and no right for the importer to choose among them. For a first consignment, that asymmetry argues for a small volume, a single establishment, a single production lot and a certificate checked twice: once by the supplier before dispatch, and once by the importer or its broker against the registration record before the declaration is lodged.

What to fix before the first booking

The work is front-loaded, and it is cheaper done once. Eight items cover it.

Confirm the product is inside the scope, not merely Norwegian and not merely wild. Check the species against the exclusion list, and check the preparation against the category under which the establishment registered. Where a product is borderline, resolve the classification question before the commercial question, because the two have different answers and only one of them can be renegotiated.

Map the chain of establishments, not just the supplier. Fishing, processing, packing, storage, consolidation and export may involve more than one registered entity. For each, record the China registration number and confirm it is active.

Ask for the certificate in the form it will actually arrive in, and check the languages. Chinese and English, at least. If the certificate is generated by a system, ask which version of the format that system produces and when it was last changed.

Confirm that the certificate names the producing establishment in a way that matches the registration record. The registration number is the field that will be matched at the border.

Check the traceability file the supplier can produce: catching area or vessel, landing, processing records, storage records. Traceability is a stated product condition, and it is the one that cannot be reconstructed after the fact.

Check the labelling artwork against the Chinese standard before printing. The notice incorporates the standard by reference; it does not restate it, and it will not warn you.

Set the declaration fields with the broker in advance: code 519, the registration number, the use field. Confirm that the broker validates rather than transcribes.

Record the registration expiry and renewal status for each establishment, and diarise a re-check. Automatic renewal is a default, not a guarantee.

The outlook is a slow widening rather than a single event. The protocol signed in October 2026 was described by both sides as a framework for wild-caught Norwegian seafood, and the notice is its first operational output. Species that were not previously approved — the flatfish, wolf fish and sea cucumber lines that the trade has been discussing — become approvable through the same route rather than through a new instrument. That is the useful thing about a framework: it makes the next product a paperwork exercise instead of a negotiation.

For importers, the strategic reading is that the constraint has moved. Market access is no longer the scarce thing for Norwegian wild aquatic products. What is scarce is a supplier base whose registrations are current, whose certificates are correctly versioned and whose traceability survives a challenge. Those are unglamorous assets, and they are what the notice actually rewards.

Sources

  • 海关总署·海关总署公告 2026 年第 159 号(关于进口挪威野生水产品检验检疫要求的公告,2026-11-17 发布并实施):http://www.customs.gov.cn/customs/2026-11/18/article_2026111816055771888.html
  • 海关总署·关于《中华人民共和国海关进口食品境外生产企业注册管理规定》实施相关事宜的公告(公告〔2026〕27 号,含推荐注册目录、不予自动延续清单、许可证类别代码 519 填报要求):http://www.customs.gov.cn/customs/2026-03/18/article_2026032510113832914.html
  • 中国政府网·国务院公报《中华人民共和国海关总署令(第 280 号)——中华人民共和国海关进口食品境外生产企业注册管理规定》(2025-10-14 公布,2026-06-01 施行):https://www.gov.cn/gongbao/2025/issue_12426/202511/content_7049741.html
  • 海关总署·在线访谈「聚焦进口食品源头监管,释放注册改革惠企红利」(2026-11-11,注册有效期、整改与暂停口径):http://zxft.customs.gov.cn/online?i=1003807

挪威野生水产品获准输华:读懂「注册 + 卫生证书」这条链

这份公告做了什么,又刻意没做什么

2026 年 11 月 17 日,海关总署发布 2026 年第 159 号公告,就进口挪威野生水产品的检验检疫要求作出规定。公告自发布之日起实施,没有过渡期,没有按品类分批推进,也没有随附一份已批准企业名单。文本很短,而且几乎全部篇幅都是写给进口商以外的两方看的:挪威主管部门和挪威生产企业。

法律依据是一叠文件而不是一份文件,这一叠怎么搭起来,决定了出问题时该往哪里找。三部法律及其实施条例提供授权,分别是食品安全法、进出境动植物检疫法、进出口商品检验法。两部部门规章提供程序,分别是进出口食品安全管理办法、海关进口食品境外生产企业注册管理规定。中挪之间的议定书提供技术标准。公告的作用,是把这份议定书在中国境内接通。

范围由一个「纳入」和一个「排除」界定。纳入的是野生的、供人类食用的水生动物产品及其制品,以及藻类等海洋植物产品及其制品。排除划得很紧:两栖类、爬行类、水生哺乳动物类、活水生动物、水生动植物繁殖材料,全部在范围之外。

不少商业对话就终止在这份排除清单上。一家供应活体帝王蟹的挪威企业不在这份公告覆盖之内,尽管帝王蟹是野生的、也是供人食用的。活水生动物走的是另一套安排,这份公告没有动它。繁殖材料同理,例如三文鱼卵。公告覆盖的是已经死亡的产品:冷冻、冰鲜、干制、盐渍、熏制、罐装,以及与之并列的藻类各条产品线。只读了标题「挪威野生水产品现在可以进了」的进口商,会把已经开放的范围估大。

公告没有做的事,值得用同样的精确度说清楚,因为空档正是有人凭空造出要求的地方。它没有设配额,没有动任何税目。它没有指定口岸。它没有设检疫审批环节——活水生动物那一类才附带这个环节;它实际创设的义务是注册、产品条件和证书三项。它没有规定装运前提前报备的天数。它也没有点名对中国进口商施加义务——文本里的义务都是写给挪方的,这恰恰是为什么进口商要做的是核实,而不是申请。

这种不对称,是理解这份公告最有用的一个入口。中国告诉挪威:挪威必须保证什么。进口商要做的是逐批确认:这个保证确实给出了。后面所有内容,都是这件事该怎么做。

文件的层级也值得说清。公告是依据上面列出的法律和规章发布的行政文件,它不创设那些依据本身并不支持的义务。这一点朝两个方向起作用。一方面,公告不能成为某个合规问题的全部答案,因为它援引的标签标准、它默认的注册规则、它隐含的残留限量,都住在别的文件里。另一方面,公告改起来相对容易:范围的调整、证书要求的变更,可以再发一份公告,而不必走立法程序。把文本当成固定参照的进口商,会在一段时间内是对的,然后毫无预警地变成错的。

实务上的翻译是:这份公告应当与它援引的那两部规章放在一起归档,而不是单独归档。一家企业的注册档案、现行注册规定及其配套实施公告、适用的中国食品安全标准,构成一个文件组。脱离这个文件组去读公告,会产生一种虚假的完备感,因为公告对它已经委出去的部分,是刻意保持沉默的。

第一环:注册,以及由谁推动

公告要求,输华野生水产品的生产企业——定义明确包含生产、加工、贮存三类——应获得挪威官方批准,并受其有效监督;其食品安全卫生管理和防护体系应符合中挪两国食品安全法律法规的要求;并且应由挪方向中方推荐注册。紧接着的那句话才是关键:未经注册,不得向中国出口。

由此有三个后果,每一个都经常被漏掉。

第一,注册在成为中方动作之前,先是挪方动作。企业不是向北京递交申请,而是由本国主管部门批准并推荐,中方在此基础上予以注册。进口商问供应商「你们在中国注册了吗」并得到一个笃定的「是」,其实几乎没获得任何信息,因为接下来要问的是:这项推荐是否仍然有效。推荐可以被撤回,注册可以被暂停。

第二,贮存环节也在围栏之内。公告对生产企业的定义包含贮存,而注册规定的配套实施要求又把用于贮存水产品的冷库单独纳入注册管理范围。一艘挪威渔船、一家加工厂、一座冷库,可能是三个不同的注册主体,而一批货可能三者都经过。单证上出现的注册编号,必须对应真正实施了被证明那道工序的主体。

第三,注册制度本身刚改过,改动的方向在一处是宽松的,在另一处毫不留情。现行《海关进口食品境外生产企业注册管理规定》于 2025 年 10 月 14 日公布、2026 年 6 月 1 日施行,取代 2021 年版本。按现行规定,水产品属于需官方推荐注册的目录范围,也就是说,上面那条路径对这门生意不是可选项。注册有效期为五年,届满自动延续五年,但有三类例外:产品列入已公布的不予自动延续注册清单、企业因不符合注册要求处于整改期间、中方已依法暂停该国相关食品进口。

目前的不予自动延续注册清单里是肉与肉制品、燕窝与燕窝制品。水产品不在其中,也就是说,状态正常的挪威企业不必提交延续申请,也不必盯住届满前三到十二个月那个窗口。这是实实在在的便利,同时也是一处给粗心者准备的陷阱:以为必须申报、于是白忙一场的企业浪费的是精力;而处在整改期间却以为会自动延续的企业,会在最糟的时点发现并非如此。

变更登记是更锋利的那道边。企业迁址、法定代表人变更、所在国家(地区)授予的注册编号改变——凡对食品安全卫生管理和防护体系造成重大影响的情形——都不予变更。企业必须重新申请注册,原有的在华注册编号自通知之日起失效。对进口商而言,这是一项带日期的供应方风险:行业新闻里一句「工厂搬迁」,不只是物流事件,而是证书上那个注册编号的一次中断。

现行规定里有一项结构性安排值得知道,尽管它并不改变义务本身。如果某国(地区)的食品安全管理体系已获海关总署认可,并且满足一组条件之一——已与海关总署签署进出口食品安全合作协议、已与中国签署包含食品安全合作内容的协议或备忘录、或经风险评估认为可以采取清单注册方式——双方主管部门可以书面约定采取清单注册方式。走这条路时,主管当局分批提交推荐在华注册的企业清单,连同申请信息、所推荐企业符合注册条件的声明,以及承诺持续履行双边合作文件约定责任的声明;中方审查后,对符合条件的予以注册。

对挪威出口方来说,这件事的意义是程序性的,不是实质性的。清单注册一次缩短一批企业的路径,并把挪威主管当局自己的质量控制变成实际起作用的过滤器。对进口商来说,它意味着某家企业的注册状态可能是在一次批量操作里形成的,而不是单独形成的;这不更好也不更坏,但它改变了核实工作该看什么:不是看这家企业当初申请得对不对,而是看主管当局那份清单是否仍然是最新的。

还有两项行政事实可以把画面补齐。注册不收费——配套公告明确写道,实施进口食品、初级食用农产品境外企业注册登记工作,海关总署不收取任何费用;这一点值得知道,因为中间环节有时会为此收费。另外,海关总署公布的注册系统是申请、变更、延续、暂停、恢复的办理通道,也是查询某家企业华注册编号与有效期的地方,可以与已公布名单比对。合规档案应当引用的,是那份公布名单,而不是供应商的一句保证。

第二环:证书,一批一批地来

证书是双方在纸面上的会合点。挪方应对输华野生水产品实施检验检疫,并对每批货物出具经双方确认格式的卫生证书,证明该批产品符合中挪两国食品安全法律法规和国家标准的有关规定。证书至少用中文、英文印制,填写时英文为必选语言;卫生证书应完整填写输华生产企业的信息。

这段话里有三个细节,分量都比字面更重。

「至少中文、英文」是下限,不是上限。用挪威文、中文、英文三种语言印制的证书是合规的。只有中文的证书不合规,因为英文是明确必选。只有挪威文和英文、没有中文的证书同样不合规,因为下限是两种语言都在。这听起来琐碎,直到一批货抵达时,随附证书出自一套为别的市场配置的系统。

「完整填写生产企业信息」,是让证书能够与注册数据库对得上的前提。一份写清了出口商和收货人、却把生产企业笼统带过的证书,给不了口岸查验人员任何可比对的东西。实务上,承担比对功能的就是注册编号;而证书上的编号与报关单上的编号不一致,是这门生意里最常见的一处单证瑕疵。

「格式经双方确认」,意味着证书是一份受控文件,不是挪威某家实验室可以自行调整的模板。公告还把控制机制写明:格式如有变更,挪方应至少在证书生效前一个月向中方备案。这一个月的备案窗口,正是进口商应当把「证书版本」与「签发日期」一并记录的原因。在一处变更之后仍按旧格式签发、或者在通知中方之前就启用新格式的货物,即便其中每一项实质事实都真实,处境也不舒服。

证书还是逐批的,这正是证书与资格的区别。企业的注册,界定的是这家企业;证书界定的是一批货。一批货分成两个生产批次,证书就要与这个分法对应;一个集装箱由两家企业的货拼成,就要两份证书,或者一份确实覆盖了两家的证书。拼柜是小进口商最容易临场发挥的地方,也恰恰是临场发挥最藏不住的地方。

证书背后那六项产品条件

公告列出了挪方应确保的产品条件。把它当作一份合规文件来读,这六项就是证书所主张的内容,而每一项在挪方都有一个可被询问的操作对应物。

第一项是来源。产品须在本国或国际水域合法捕捞,或者来自中方已准入国家(地区)已获注册的水产品生产企业。后半句对转口原料很要紧:一家挪威加工商如果从第三国采购原料,只有在该第三国本身在中方准入名单上、且供货企业已获注册的前提下,才能把料做进输华产品。挪威供应商自己的供应链,从这里开始进入中国进口商的尽职调查范围。

第二项是三类被定义的问题均不存在:已输华或拟输华产品违反中国食品安全法律法规、国家标准以及议定书规定,已经严重影响输华产品安全;挪威境内发生中国进境动物检疫疫病名录列明的、或世界动物卫生组织规定必须通报的、与议定书产品相关的水生动物疫病,已经严重影响输华产品安全;挪威境内发生重大食品安全事件,或生产企业发生重大公共卫生事件,或捕捞水域受到污染物影响,已经严重影响输华产品安全。三项都带着「已经严重影响」这个限定,那是判断,不是阈值。进口商应当把它读成挪威行业出事时的对话触发器,而不是一次性打勾的选项。

第三项是药物与添加剂。不得直接或间接使用中挪双方禁用的药物或添加剂,限用或允许使用的须按规定使用。公告没有复述任何一方的清单,实务上的参照就是中方残留监控框架与挪方公布的要求。对涉及养殖后放流或资源增殖类货源的进口商,这一项的时间尾巴最长。

第四项是检验结果:未发现中挪法律法规列明的致病微生物、有毒有害物质和异物,未发现名录列明的水生动物疫病,未发现世界动物卫生组织规定必须通报的水生动物疫病。这正是证书所证明的内容,也是证书不能在检测完成之前签发的原因。

第五项是全链条过程控制:捕捞、生产加工、包装、贮存、运输、中转和出口,均应符合中挪相关安全卫生要求和可追溯要求。关键词是可追溯。一批货如果无法从中国口岸回溯到捕捞水域和注册企业,即便每一项检测结果都干净,这一项也没有满足。

第六项是包装与标签标识:应符合中国食品安全国家标准和进口食品安全管理相关规定。要注意它说了什么、没说什么。它是以援引的方式把中国标签标准纳进来,而不是复述标准内容;也就是说,标签问题由标准回答,不由公告回答。又因为这是食品,注册规定里另有一条要适用:已获注册的企业向中国境内出口食品时,应当在包装上标注在华注册编号,或者标注所在国家(地区)主管当局批准的注册编号。公告没有复述这一条,但到了口岸,公告与规章是放在一起读的,这也是为什么纸箱上的编号与报关单上的编号被期待为同一个编号。

这条链通常断在哪里

这门生意里的摩擦,六种情形占了大部分,其中五种是单证问题。

第一种是证书上的注册编号与报关单上的编号不一致。原因通常是供应商换用了兄弟工厂、工厂搬迁后重新注册、或者贸易商从上一批货里抄了一个编号。海关系统会把编号与注册库比对,对不上申报货物与原产地的编号,申报就卡住。

第二种是产品落在排除项而不是纳入项里。活水生动物、繁殖材料、水生哺乳动物是显性的例子。不那么显性的是「加工后性质已变」的产品——加工程度深到已经无法用企业注册时所属的水产品类别来描述。归类与注册是两个不同的问题,通过其中一个,不等于回答了另一个。

第三种是把藻类当成边角料处理。藻类等海洋植物产品及其制品明确在范围之内。以为「植物就不是动物产品、因此不在动物注册制度里」的贸易商,会发现公告并不作这个区分。注册义务挂在为输华而生产该产品的企业身上,与产品属于哪一界无关。

第四种是语言。只用一种语言印制的证书,或者用了两种语言但不含中文和英文的证书,是被形式挡下来的,不是被实质挡下来的。这是最便宜就能避免、也最高频的失败之一,因为证书模板通常在系统里集中配置,很少改动。

第五种是贮存这一环从未注册。中转枢纽的冷库、拼柜商的操作场地、负责改包的第二个厂区——其中任何一个都可能构成规则意义上的生产企业,而未注册的那一个,会让链条在捕捞与加工环节都很干净的情况下照样断掉。

第六种根本不是单证问题:是把准入当成永久的。公告规定,对发生严重问题的产品,或多次发生不合格问题的生产企业,中方可采取加强检验检疫或暂停进口等措施。暂停针对的是某家企业或某个类别,生效速度比一份供货合同改写得快得多。

申报,以及那口五年的钟

到了口岸,公告里的条件就变成数据字段。进口食品须在报关单「产品资质」项下「进口食品境外生产企业注册」证书栏,按许可证类别代码 519,规范填写与该食品「原产国(地区)」对应的境外生产企业在华注册编号;「商品填报」项下「用途」栏须规范填报「食用」。未按要求规范填报的,海关不接受申报。

关于这项要求,有两点值得强调。第一,它是校验,不是描述:编号要拿去与数据库比对,所以一个看上去像模像样的编号,和一个空着的编号,失败方式是一样的。第二,虚假填报不是笔误——配套公告写明,虚假填报信息、骗取海关单证的,将依法予以查处。对报关行来说,这意味着「供应商说没问题就照填」的成本被抬高了。

申报之外是常规那一套:合同、发票、装箱单、提单、需要享受优惠税率时的原产地证书,以及卫生证书。卫生证书是把其余单证串起来的那一份,因为它是唯一一份讲产品安全、而不讲货值或货物流向的文件。

这一切背后还有一口钟,多数进口商从不看它,那就是注册的钟。五年,自动延续,三类例外。要盯的是例外。最常见的是整改:中方发现已注册企业不再符合注册要求时,会责令其在规定期限内整改,通知其所在国家(地区)主管当局督促整改,整改期间暂停该企业相关食品进口;整改完成后,企业提交主管当局出具的监督整改报告和符合注册要求的书面声明,中方审查,符合要求的恢复进口。

对进口商来说,实际含义是:供应方的注册状态是一个有三种可能取值的活变量——有效、整改期间暂停、以及因变更事项而失效——而状态可以在不通知进口商的情况下改变。应对办法不是为谨慎而谨慎,而是一份供应方档案:链条上每一家企业,记录注册编号、最近一次核实日期、以及核实来源。

进口商身上还有第二项义务,而它很容易被忽略,因为公告没有提。按食品安全法及其实施制度的要求,进口食品的进口商应当建立并落实生产企业审核制度,审核不合格的,不得进口。海关在公开答复中重申了这一点,并补上一句操作性的提醒:进口商应当及时关注海关总署网站公布的未准入境食品信息,因为一家供应商的不合格模式,往往先出现在那里,之后才出现在合同纠纷里。

把两项义务合起来读,它们描述的是一种比单独看任何一项都更吃力的分工。监管方按标准核验企业并公布结果;进口商按自己的商务与合规要求核验企业并留存记录。这两件事都不能靠手里有一份证书来满足,因为证书讲的是某一批货,而义务讲的是某一段关系。

也值得留意公告在出事之后怎么做,因为它决定第一批货该怎么组织。不合格产品予以销毁、退运或其他处理——三种后果的成本结构差别很大,而进口商没有在其中挑选的权利。对首批货物而言,这种不对称本身就是论据:小批量、单一生产企业、单一生产批次,证书查两遍——一遍由供应商在发运前查,一遍由进口商或其报关行在申报前对照注册记录查。

第一批订单之前该做完的事

这项工作是前置的,而且做一次比做两次便宜。八件事可以覆盖。

确认产品在纳入项内,而不只是「来自挪威」和「是野生的」。把品种对照排除清单过一遍,把加工形态对照企业注册时所属的类别过一遍。遇到边界产品,先解决归类问题再解决商务问题,因为两个问题的答案不同,而只有一个还能重新谈。

把链条上的企业画出来,而不是只画供应商。捕捞、加工、包装、贮存、拼柜、出口,可能涉及不止一个注册主体。逐个记录在华注册编号,并确认其有效。

按证书实际到手的样子索取样张,核对语言。至少中文和英文。如果证书由系统生成,问清该系统当前输出的是哪个格式版本、上一次变更是什么时候。

确认证书以能与注册记录对应的方式写明生产企业。到了口岸会被比对的字段,就是注册编号。

核查供应商能拿出的可追溯材料:捕捞水域或渔船、卸货记录、加工记录、贮存记录。可追溯是公告写明的产品条件之一,也是唯一一项无法事后补做的条件。

在印刷之前,把标签稿对照中国标准核一遍。公告是以援引方式纳入标准,它不复述,也不会提醒你。

提前与报关行把字段定好:代码 519、注册编号、用途栏。确认报关行做的是校验,而不是照抄。

为每家企业记录注册到期日与延续状态,并设一个复核提醒。自动延续是默认,不是保证。

前景是缓慢变宽,而不是一次性事件。2026 年 10 月签署的议定书,双方都把它描述为挪威野生捕捞海产的框架性安排,本公告是它第一项落地成果。此前未获批准的品种——业内一直在谈的鲽鱼、狼鱼、海参这几条线——会沿同一条路径变得可批,而不需要一份新的文件。框架的好处就在这里:它让下一个品种变成一项文书工作,而不是一场谈判。

对进口商来说,战略上的读法是:约束已经挪了位置。对挪威野生水产品而言,市场准入不再是稀缺的东西。稀缺的是一批注册状态有效、证书版本正确、可追溯性能经得起追问的供应商。这些资产不体面,而这份公告真正奖励的,正是它们。

来源

  • 海关总署·海关总署公告 2026 年第 159 号(关于进口挪威野生水产品检验检疫要求的公告,2026-11-17 发布并实施):http://www.customs.gov.cn/customs/2026-11/18/article_2026111816055771888.html
  • 海关总署·关于《中华人民共和国海关进口食品境外生产企业注册管理规定》实施相关事宜的公告(公告〔2026〕27 号,含推荐注册目录、不予自动延续清单、许可证类别代码 519 填报要求):http://www.customs.gov.cn/customs/2026-03/18/article_2026032510113832914.html
  • 中国政府网·国务院公报《中华人民共和国海关总署令(第 280 号)——中华人民共和国海关进口食品境外生产企业注册管理规定》(2025-10-14 公布,2026-06-01 施行):https://www.gov.cn/gongbao/2025/issue_12426/202511/content_7049741.html
  • 海关总署·在线访谈「聚焦进口食品源头监管,释放注册改革惠企红利」(2026-11-11,注册有效期、整改与暂停口径):http://zxft.customs.gov.cn/online?i=1003807

How Beijing Reviews Duty-Free Eligibility for a Foreign-Funded R&D Centre’s Imported Equipment

  • Beijing’s four-department notice, signed 18 August 2026, converts the two central 2026 instruments into a local review procedure for Foreign-Funded R&D Centre Duty-Free Eligibility, valid from 1 January 2026 to 31 December 2030.
    北京市四部门联合印发的《关于做好外资研发中心采购进口设备免税资格审核有关工作的通知》,把中央两份 2026 年文件落成本市的一道审核程序,有效期自 2026 年 1 月 1 日至 2030 年 12 月 31 日。
  • Three thresholds apply at the same time: an investment total of at least USD 8 million for a centre that is an independent legal person, at least 80 full-time research and experimental development staff, and at least RMB 20 million in cumulative equipment at original value since establishment.
    三道门槛须同时满足:独立法人的投资总额不低于 800 万美元;专职研究与试验发展人员不低于 80 人;设立以来累计购置设备原值不低于 2000 万元人民币。
  • Six documents are required, bound in order in four copies, including a commitment letter that Imported Equipment contracts are already signed and delivery will fall before the end of the year of application.
    需提交六份材料,按顺序装订、一式四份,其中包含一份承诺函——进口设备合同已签订、设备将于申报当年年底前交货。
  • The Joint Review Meeting is convened by the municipal commerce bureau after acceptance and preliminary review, and a joint announcement follows on the bureau’s own website.
    市商务局受理并初审同意后牵头召开审核部门联席会议,通过后由审核部门在市商务局网站联合公告。
  • Forty-five working days is the outer limit for the announcement or the written refusal, counted from acceptance and excluding time the applicant spends correcting the file.
    自受理之日起 45 个工作日内须作出公告或者书面审核意见,申请人补正材料的时间不计入。
  • The exemption is list-managed and attaches to imported equipment procured for the centre’s own research activity, not to every good the centre imports.
    免税实行清单管理,落在研发中心为自身研发活动采购的进口设备上,而不是研发中心进口的所有货物。
  • Approval authority is provincial: the central notice names provincial commerce departments acting with provincial finance and tax authorities and the local customs district, so each province runs its own window and its own material list.
    核定权在省级:中央文件把核定主体写成省级商务主管部门会同省级财政、税务部门和所在地直属海关,因此各省有自己的窗口和材料清单。
  • A change of name or business scope must be reported, and the bureau then determines whether the centre keeps the treatment from the date of the change registration.
    名称或者经营范围发生变更须及时报送,由市商务局核定变更后自变更登记之日起能否继续享受政策。

How Beijing Reviews Duty-Free Eligibility for a Foreign-Funded R&D Centre’s Imported Equipment | 外资研发中心进口设备免税资格怎么核:北京执行细则逐条拆解

What the Beijing notice actually decides

A foreign-funded R&D centre that wants to import equipment without paying import duty and import-stage value-added tax needs two things that are easy to confuse: an entitlement under the central policy, and a determination by a provincial authority that this particular centre is one of the entities entitled. The central documents — Caiguanshui [2026] No. 7, which sets the policy for the 15th Five-Year Plan period, and Caiguanshui [2026] No. 8, which sets the administrative measures — describe who may qualify and what may be imported. They do not tell a company which window to walk into, which form to fill, or which official signs.

That gap is what a provincial notice fills, and Beijing’s is worth reading closely because it is unusually explicit. The notice is issued by the municipal commerce bureau together with the municipal finance bureau, Beijing Customs and the Beijing tax bureau of the State Taxation Administration. It is dated 18 August 2026 and the municipal science and technology platform posted it on 9 November 2026. Its stated validity runs from 1 January 2026 to 31 December 2030, which matches the five-year window of the central policy.

Three features of the document deserve attention before anything else.

First, it names the reviewing departments and, by doing so, fixes where the decision is made. The four departments review together. The commerce bureau convenes, receives and holds the file; the finance bureau, the customs district and the tax bureau sit in the same meeting. A centre does not apply to customs for a duty exemption and separately to tax for a waiver. It applies once, to a joint process, and the output is a joint announcement.

Second, it defines eligibility by reference rather than by restatement. The notice lists three thresholds and then says that the meaning of investment total, R&D input, full-time research and experimental development personnel, and equipment follows Caiguanshui [2026] No. 8. That cross-reference is the part most applicants skim and later regret. The thresholds look like round numbers. Whether a particular cost counts toward them is decided elsewhere, in a document the applicant has usually not read.

Third, it converts an entitlement into a file with a physical shape: six documents, stamped with the company seal, bound in order, four copies. The requirement to bind in order is not decorative. A joint review is conducted by people from four organisations reading one file in sequence, and a file assembled in the order the notice specifies can be read by four reviewers without anyone re-sorting it.

The notice also does something that is easy to miss on a first reading. It supplies six annexes as templates — the application letter, the review form, the asset list for a non-legal-person centre, the equipment summary, the delivery commitment letter and the personnel roster. Using them is not optional in practice. A submission that restates the content in a company’s own format gives the reviewers a document they must map back to the template before they can compare it with the criteria.

Status should be stated plainly. This is not a new incentive and not a pilot. It is the municipal operating procedure for an incentive that already exists at the central level, restated for the 2026–2030 period. A centre that qualified under the previous period and has continued to meet the thresholds is not required to rediscover the policy. It is required to check whether the file it holds still answers the questions this notice asks.

Finally, this is a municipal document about a national programme, and the two should not be conflated. Beijing’s thresholds are not Beijing’s invention; they come from the central measures. What Beijing adds is the meeting, the window, the clock, the annexes and the announcement. Those are the parts that differ from province to province, and they are the parts this article is about.

Three thresholds that must all be met

The thresholds read as follows. A centre that is an independent legal person must have an investment total of at least USD 8 million. A centre that is not an independent legal person — an internal department of a company or a branch — must have an R&D input of at least USD 8 million. Either way, the centre must have at least 80 full-time research and experimental development personnel. And it must have purchased, since establishment, equipment with an original value of at least RMB 20 million.

The conjunction matters. These are not alternative routes; a centre that clears two of the three does not qualify. In practice each threshold fails for a different reason and at a different stage of a company’s life, which is why a single yes-or-no answer to the question of eligibility is rarely useful.

Investment total is the threshold that trips up centres that grew by acquisition or by internal reorganisation. It is a stock measure tied to the legal person, and it is not the same as the amount the parent has actually funded into the entity. A centre set up with modest capital and since funded through intercompany service charges may have a large operating budget and a small investment total. The Beijing notice does not resolve that question; it points to the central measures, and the central measures define the term.

R&D input is the threshold that trips up non-legal-person centres, and it is the harder of the two to document because it is a flow rather than a stock. An internal department has no capital line. What it has is a history of spending on people, materials, depreciation and outside services that the parent has attributed to the centre. Assembling that history into an auditable figure requires the accounting to have been separated from the beginning, and the absence of that separation is the single most common reason a non-legal-person application is deferred rather than refused. The notice anticipates the problem by requiring, for such a centre, both the verification report of the foreign-invested enterprise it sits inside and an asset list of what was invested specifically to establish and build the centre.

The headcount threshold looks like the easiest and is often the hardest. Eighty full-time research and experimental development personnel is a number, but both “full-time” and “research and experimental development” are defined terms. A centre that counts everyone with an engineering title, including staff who spend most of their time on customer support or production engineering, will produce a roster that does not survive comparison with the payroll and social insurance records. The roster the notice requires is specific about its fields: name, post, labour contract signing date, contract term and contact details. Those fields are chosen so that the roster can be cross-checked, not so that it can be read as a headcount.

The equipment threshold is the one most often met and least often proved. Twenty million renminbi of original equipment value accumulated since establishment is a modest figure for a laboratory and a large one for a software centre. The difficulty is not the number but the aggregation: equipment bought by the parent and used by the centre, equipment bought by the centre and transferred to a sister company, equipment that has been fully depreciated and written off the books. The notice asks for a cumulative equipment summary, which implies a schedule that survives audit, and the central measures define what counts as equipment.

Two practical observations follow.

The first is that the three thresholds are assessed at the point of application, not at the point of establishment. A centre that met all three when it was set up and has since let its headcount fall below eighty holds no grandfather right. It applies on the facts it can prove at the date of filing, and the file has to show those facts.

The second is that the thresholds are the easy part of the conversation and the definitions are the hard part. A company that starts by asking whether it reaches USD 8 million will usually have an answer within an afternoon. A company that starts by asking how the central measures define investment total, R&D input and full-time personnel will usually discover that one of the three sits closer to the line than management believed.

Six documents and the reasons files come back

The six documents are listed in a fixed order, and the order is a reading order.

The first is the application letter in original form. Its content is prescribed: the centre’s basic situation, its research activity, and an explanation of how it meets the conditions. This is the document reviewers read first and the one most often written as a cover note. A letter that describes the group’s global business and its China strategy but not the centre’s research activity gives the joint meeting nothing to test against the thresholds.

The second is the review form in original form, which is the structured version of the same information.

The third is the evidence of funding: the most recent capital verification report and the prior year’s audit report, in copy. For a non-legal-person centre the requirement changes. The file needs the most recent capital verification report of the foreign-invested enterprise the centre sits inside, plus an original asset list of what was invested specifically to establish and build the centre. This is the document that most often arrives incomplete, because the asset list is not a standard accounting output and has to be constructed.

The fourth is the cumulative equipment summary in original form.

The fifth is the commitment letter: the centre confirms that contracts for the purchase of imported equipment have already been signed and that the equipment will be delivered before the end of the year in which the qualification is determined. This is the most distinctive document in the set and the one that reveals what the review is for. The exemption is not a general status that a centre holds and then draws on at leisure. It is tied to equipment that is on order and arriving within the year, which is why the commitment letter is a required document rather than a supporting one.

The sixth is the roster of full-time research and experimental development personnel in original form, with the five fields described above.

Four copies are required, each stamped with the company seal, bound in the order listed.

The reasons files come back are predictable, and most are failures of internal consistency rather than failures of substance.

The equipment summary and the delivery commitment letter disagree. A centre lists equipment in the summary that is not covered by a signed contract, or signs contracts whose delivery dates fall into the following year. The two documents are read together, and the mismatch is visible without any external check.

The capital verification report is out of date. The notice asks for the most recent one, which means the file has to reflect the position at filing rather than the position at establishment. A centre that has not had a verification since it was set up will submit a document describing a company that no longer exists.

The roster does not reconcile with the payroll or the social insurance record. This is the most consequential mismatch because it goes to the eighty-person threshold and because a reviewer can test it quickly. A roster with eighty-two names and a social insurance record with seventy-one is not a rounding difference; it is a different answer to the question.

The audit report covers the wrong entity. For a non-legal-person centre the audit is of the foreign-invested enterprise, and the centre’s own costs appear inside it as a segment rather than as a set of accounts. A file that submits the centre’s internal management accounts instead of the statutory audit has submitted something the reviewers cannot use.

The application letter argues a case the notice does not ask about. Letters that spend their length on the group’s commitment to China, the strategic importance of the centre or the technology transferred, and that do not address the three thresholds, read as though the applicant has not understood what is being decided.

None of these is a legal obstacle. Each is a reason for the joint meeting to defer, and deferral costs a quarter, because the review clock stops while the applicant corrects the file. The notice is explicit that time spent correcting materials is not counted within the forty-five working days. That is fair as a rule and expensive as a fact.

The review chain and its clock

The sequence has four steps and one deadline.

The file goes to the commerce bureau window at the municipal government service centre. The bureau accepts it and conducts a preliminary review. If the preliminary review is satisfactory, the bureau convenes the joint meeting of the reviewing departments, which examines the file and settles the list of centres meeting the conditions.

The result is published as a joint announcement on the commerce bureau’s website. The determination is then sent by letter to Beijing Customs, copied to the finance bureau and the tax bureau, reported to the Ministry of Commerce’s foreign investment department, and notified to the centre. A centre that does not qualify receives a written review opinion explaining why.

The clock is forty-five working days from acceptance, excluding time the applicant spends correcting materials. That is a real commitment and a useful planning input, but it should be read with the exclusion in mind. The forty-five days measure the reviewers, not the applicant, and a file needing two rounds of correction can take a season.

A second route exists, and it matters for companies outside the central districts. Several bodies have been delegated the municipal bureau’s authority to lead the review: the economic and technological development zone administrative committee and the commerce bureaus of Chaoyang, Haidian, Tongzhou, Shunyi, Changping and Daxing. Those bodies may follow the notice. Under the delegated route the process differs slightly. The delegated body convenes its own joint meeting, reports the qualifying list to the municipal bureau, and the municipal bureau then seeks the views of Beijing Customs, the finance bureau and the tax bureau. Once those views are unanimous, the delegated body publishes the announcement on its own government website and notifies the centre, while the municipal bureau writes to customs, copies the finance and tax bureaus, and reports to the Ministry of Commerce.

The difference is administrative rather than substantive — the criteria are the same — but it changes whom a company talks to and whose website carries the announcement. A centre in the economic and technological development zone that files at the municipal window has done nothing wrong; it has simply taken a longer route to the same meeting.

Two further procedural points are worth extracting.

The first is the treatment of tax already paid. Where duty and import-stage taxes have been levied on goods that should have been exempt, the centre may apply for a refund, and the refund carries interest at the bank demand deposit rate for the same period. A condition attaches to part of this. Where the imported goods have been taxed and the input value-added tax has not yet been credited, the centre must first obtain a statement from its competent tax authority confirming that the input tax on those imports has not been credited. That statement is an annex to the central measures, and obtaining it sits with the tax function rather than the customs function.

The second is the treatment of change. If the centre’s name or business scope changes, it must report the change to the commerce bureau, which then determines, following the same procedure, whether the centre continues to enjoy the treatment from the date of the change registration, and records that date. The determination is again written to customs, copied to finance and tax, reported to the ministry and notified to the centre. A centre that changes its scope without reporting holds a determination that no longer describes it.

What the exemption covers, and what it does not

Three boundaries are worth stating plainly, because each is a common source of overstatement.

The first is that the exemption is list-managed. The goods that may be imported free of duty and import-stage taxes are set out in a list issued by the Ministry of Finance together with the relevant departments. For the current period that list is published under Caiguanshui [2026] No. 9 and applies from 1 January 2026. A centre that holds the qualification and imports an item outside the list does not obtain the exemption for that item. Qualification is a condition for using the list; it is not a substitute for it.

The second is that the object of the exemption is imported equipment procured for the centre’s research activity, not everything the centre imports. The Beijing notice is framed around the purchase of imported equipment, and the commitment letter it requires concerns equipment contracts and delivery. A centre that imports samples, spare parts, consumables or office equipment should not assume the qualification reaches them. The central measures describe the covered categories and the list describes the covered goods; between the two, the room for assumption is small.

The third is that the exemption covers import duty, import-stage value-added tax and import-stage consumption tax. It is not an income tax measure. A centre holding the qualification does not thereby obtain a reduced corporate income tax rate, and the qualification does not interact with the high-technology enterprise regime or with the western development regime. Those are separate determinations on separate criteria, and a company that treats them as one package will misplan.

A fourth boundary is less obvious and more useful. The exemption attaches to the centre, and the centre is a defined applicant: either an independent legal person or a non-legal-person department or branch. That definition decides who signs the application, whose audit report is submitted and whose equipment summary is compiled. In a group with several research entities in the same city, which entity applies is a real question, and the answer is not always the largest one. It is the one whose books can produce the three thresholds cleanly.

For the finance function, the practical consequence is that the qualification has to be modelled before it is applied for. The benefit is duty and import-stage taxes on a defined set of equipment arriving within a defined year. The cost is the file, the meeting, the clock and the annual maintenance of the underlying records. A centre importing one instrument a year may find the arithmetic thin. A centre equipping a new laboratory in a single year will usually find it substantial.

Why Beijing is a sample rather than the rule

The central notice names the approving authority in a way that decides the geography of the whole exercise. Among the categories of eligible entities, foreign-funded R&D centres are approved by the provincial commerce department together with the provincial finance and tax authorities and the local directly subordinate customs. The authority is provincial. There is no national window, no single form and no single announcement.

That has three consequences.

The first is that the material list differs. Beijing asks for six documents. Another province may ask for the same six plus a statement on the centre’s research projects, or accept a copy where Beijing wants an original, or require a different number of copies. A group preparing a multi-province filing should not assume the Beijing file can be reused unchanged.

The second is that the timing differs. Beijing’s notice runs to the end of 2030 and sets a forty-five working day clock. Other provinces have published their own measures for the same period, and application windows and release dates of results are set locally. A centre planning around a national calendar will be surprised.

The third is that the delegated route differs. Beijing has delegated the authority to an economic development zone committee and six district bureaus. That is a Beijing arrangement, reflecting the size of the municipality and the concentration of foreign investment in those districts. It has no counterpart in a province where the commerce department reviews everything centrally.

For a group with research entities in several provinces, the sensible approach is to treat the central documents as the constant and everything else as a variable. The three thresholds come from the centre. The definitions come from the centre. The list of exempt goods comes from the centre. The window, the annexes, the number of copies, the meeting, the clock and the announcement come from the province, and they should be collected province by province rather than extrapolated from the first one.

A further point is easy to miss. Because the authority is provincial and the announcement is local, a qualification obtained in one province is not a qualification in another. A centre in Beijing that opens a second research facility elsewhere applies again, in that province, on that facility’s own facts. Groups that run a single national research entity alongside several smaller satellite teams should decide early which entity carries the qualification, because the answer determines where the file is built.

A working checklist for the 2026–2030 window

The window runs five years, and the useful way to use it is to work backwards from the equipment.

Start with the equipment plan. Which instruments, for which laboratory, ordered in which year, delivered in which year? The commitment letter makes delivery date a document, so the equipment plan and the filing year have to be the same plan. A centre that files in a year when nothing is arriving has nothing for the letter to say.

Then test the three thresholds against the accounting rather than the business plan. Take the investment total or the R&D input as the central measures define it, and check whether the figure is above USD 8 million on the books. Take the roster and reconcile it against payroll and social insurance before submitting it, not after. Take the equipment summary and check that it aggregates purchases rather than only those still carried on the books.

Then decide which entity applies. For a non-legal-person centre, confirm that the foreign-invested enterprise’s most recent capital verification report exists and is current, and that the asset list of what was invested to build the centre can be produced as an original document rather than reconstructed.

Then assemble the file in the prescribed order, in four stamped copies, using the six annex templates. Where a document is in a foreign language, attach a Chinese translation with the translator’s details, since the material specifications require it.

Then diarise the clock. Forty-five working days from acceptance, excluding correction time. A file submitted in a busy quarter should be expected to emerge in the next one, and a centre that needs the qualification for equipment arriving in December should file early enough that a deferral does not push the determination into the following year.

Finally, plan for maintenance. The qualification is not permanent in the sense of needing no attention. A change of name or business scope triggers a re-determination. Reporting obligations to the tax and commerce authorities continue. The equipment records have to keep supporting the summary. None of this is onerous, and all of it is easier if the centre’s books were separated from the parent’s at the outset.

A word on who should read this notice at all. It is written for a centre that is already established and already buying equipment. It does not help a company that is still deciding whether to set up a research centre in China, and it should not be used as a comparison tool between locations, because the thresholds it restates are national and the procedure it adds is local. A group comparing cities should compare the procedure — the window, the clock, the delegated route, the annexes — and treat the thresholds as a given.

One closing observation about how to read the notice. It is a procedural document, and procedural documents are usually read for their deadlines. The more useful reading is for their assumptions. This one assumes that the applicant has already read the central measures, already knows how investment total and R&D input are defined, already keeps a roster that reconciles, and already plans equipment purchases a year ahead. A company that meets those assumptions will find the notice a clear description of a manageable process. A company that does not will find that the forty-five days is not the constraint that matters.

Sources

  • 北京市人民政府·《北京市商务局 北京市财政局等 4 部门关于做好外资研发中心采购进口设备免税资格审核有关工作的通知》(2026-08-18 印发,2026-09 公开):https://www.beijing.gov.cn/zhengce/zhengcefagui/202609/t20260920_4872405.html
  • 北京市商务局·《关于做好外资研发中心采购进口设备免税资格审核有关工作的通知》全文及六个附件下载页:https://sw.beijing.gov.cn/zwxx/2024zcwj/202609/t20260920_4872284.html
  • 北京市科技创新服务平台·《关于做好外资研发中心采购进口设备免税资格审核有关工作的通知》(2026-11-09 公开发布):https://www.ncsti.gov.cn/kjdt/tzgg/202611/t20261109_262109.html
  • 财政部 海关总署 税务总局·《关于”十五五”期间支持科技创新进口税收优惠政策的通知》(财关税〔2026〕7 号)全文:http://hangzhou.customs.gov.cn/hangzhou_customs/fileDir/resource/cms/article/6010991/7183589/2026060908504267182.pdf
  • 北京市政务服务网·牵头组织对外资研发中心采购设备免、退税资格进行审核认定(办事指南,含财关税〔2026〕9 号免税清单说明):https://banshi.beijing.gov.cn/pubtask/task/1/110105000000/fb564bc7-be4d-458d-adcd-00b1575a1923_app.html

外资研发中心进口设备免税资格怎么核:北京执行细则逐条拆解

这份北京通知到底定了什么

一家外资研发中心想把设备运进境内而不缴进口关税和进口环节增值税,需要两样容易混淆的东西:一样是中央政策给的待遇,另一样是省级机关作出的”这家中心属于可享受主体”的认定。中央的两份文件——规定”十五五”期间政策的财关税〔2026〕7 号,和规定管理办法的财关税〔2026〕8 号——说的是谁可以够格、什么可以进口。它们不会告诉企业该进哪个窗口、该填哪张表、该由谁签字。

这道缝隙由省级通知来填,北京这一份值得细读,因为它写得少见地明确。文件由市商务局会同市财政局、北京海关、国家税务总局北京市税务局联合印发,落款日期 2026 年 8 月 18 日,市科技创新服务平台于 2026 年 11 月 9 日发布。文件写明的有效期自 2026 年 1 月 1 日至 2030 年 12 月 31 日,与中央政策的五年窗口对齐。

有三处特征,在别的之前先值得注意。

第一处,它点明了审核部门,因而把”决定在哪里作出”这件事固定下来。四个部门联合审核。商务局牵头、受理、持有卷宗;财政局、海关、税务局坐在同一场会议里。一家中心不是先向海关申请免关税、再单独向税务申请免税,而是提交一次,进入一个联合流程,产出是一份联合公告。

第二处,它用援引而不是复述来界定资格。通知列出三道门槛,随后写明”投资总额””研发总投入””专职研究与试验发展人员””设备”的含义按财关税〔2026〕8 号执行。这句援引,是多数申请人扫过去、后来后悔的地方。门槛看着都是整数,而某一笔成本算不算得进去,答案在另一份文件里,而那份文件申请人通常没读。

第三处,它把一项待遇变成一份有物理形态的卷宗:六份材料、加盖公章、按顺序装订、一式四份。按顺序装订不是形式要求。联席审核是四个机构的人依次读同一份卷宗,按通知规定的次序装好的卷宗,四个人读下来不需要任何人重新排序。

这份通知还做了一件初读容易漏掉的事:它给出六个附件作为模板——申请书、审核表、非独立法人的研发投入资产清单、设备汇总表、交货承诺函、人员名册。实务中用不用模板不是可选项。把同样内容按企业自己的格式重写一遍,等于让审核人员先把文件映射回模板,才能拿去和条件比对。

状态要说清楚。这不是一项新优惠,也不是试点。它是中央层面早已存在的优惠在本市的办事程序,为 2026 至 2030 年这一段重新写了一遍。上一期已经取得资格、并且一直满足门槛的中心,不需要重新发现这项政策;它需要核对的是:手里那份卷宗,是否还回答得了这份通知提出的问题。

最后一句,这是一份关于国家项目的市级文件,两者不宜混为一谈。北京的三道门槛不是北京发明的,来自中央的管理办法。北京加进去的是那场会议、那个窗口、那个时钟、那六个附件和那份公告。这些才是各省不同的部分,也是本文要谈的部分。

三道门槛,须同时满足

门槛原文是这样:作为独立法人的,投资总额不低于 800 万美元;作为公司内设部门或者分公司的非独立法人,研发总投入不低于 800 万美元。两种情形都要满足专职研究与试验发展人员不低于 80 人,以及设立以来累计购置的设备原值不低于 2000 万元人民币。

“同时”两个字要紧。这不是三条可选路径,过了两条不等于够格。实务中,三道门槛各有各的失败原因,也各在各的企业阶段上失败,所以一个笼统的”我们够不够”的答案,通常没什么用。

投资总额这道,绊住的是靠收购或者内部重组长起来的中心。它是一个挂在法人身上的存量指标,与母公司实际投进这家主体的金额不是一回事。一家以不大的注册资本设立、之后靠关联服务费维持运转的中心,可能运营预算很大,投资总额很小。北京这份通知不解决这个问题,它指向中央的管理办法,而术语的定义在那里面。

研发总投入这道,绊住的是非独立法人的中心,而且它是两者中更难举证的,因为它是一条流量而不是存量。内设部门没有资本项。它有的是一段历史:花在人、材料、折旧和外部服务上的支出,由母公司归集到中心名下。要把这段历史拼成一个经得起审计的数字,前提是会计从一开始就分开记;没有分开,正是非独立法人申报被”暂缓”而不是被”驳回”的最常见原因。通知预见到了这一点,所以要求这类中心同时提交其所在外商投资企业最近一次的验资报告,以及专门为设立和建设本研发中心而投入的资产清单原件。

人数这道看着最容易,往往最难。80 名专职研究与试验发展人员是个数字,但”专职”和”研究与试验发展”都是被定义过的词。把凡是挂工程师头衔的人都算进来,包括大部分时间在做客户支持或者生产工艺的人,做出来的名册经不起与工资表、社保名单的对照。通知对名册的字段写得很具体:姓名、工作岗位、劳动合同签约日期、劳动合同期限、联系方式。这几个字段是为了让名册可以被交叉核对,不是为了让人读一个总人数。

设备这道,是最常满足、最不常被证明的。设立以来累计购置设备原值 2000 万元,对一个实验室不算高,对一个软件中心不算低。难的不是数字,是归集:母公司买、中心用的设备;中心买、又转给兄弟公司的设备;已经提足折旧、从账上核销掉的设备。通知要一份累计购置设备汇总表,这句话的含义是一张经得起审计的明细,而什么算”设备”,由中央的管理办法定义。

由此有两个实务判断。

其一,三道门槛是在申报时点评估的,不是设立时点。一家设立时三条都够、后来人数掉到 80 以下的中心,没有祖父条款可援。它按申报日能够证明的事实申报,卷宗要呈现的就是那些事实。

其二,门槛是这场对话里容易的部分,定义是难的部分。从”我们够不够 800 万美元”问起的公司,通常一个下午就有答案。从”中央文件怎么定义投资总额、研发总投入、专职人员”问起的公司,通常会查出三者之中有一个比管理层以为的更贴近那条线。

六份材料,以及文件为什么会被退回

六份材料按固定顺序排列,而这个顺序就是一个阅读顺序。

第一份是申请书原件。内容是指定的:研发中心基本情况、研发活动情况、符合申报条件的说明。这是审核人员最先读的一份,也是最常被写成一张封面说明的一份。一封通篇讲集团全球业务和在华战略、却不讲这家中心在做什么研发的信,让联席会议没有东西可以拿去和门槛对照。

第二份是审核表原件,是同一组信息的结构化版本。

第三份是出资证明:最近一次的验资报告,加上一年度的审计报告,复印件。非独立法人的中心,要求变了——需要其所在外商投资企业最近一次的验资报告,外加一份专门为设立和建设本研发中心而投入的资产清单原件。这是最常不齐的一份,因为资产清单不是会计系统的标准产出,得专门做出来。

第四份是累计购置设备汇总表原件。

第五份是承诺函:中心确认已经签订购置进口设备的合同,并且设备将于申报资格认定当年年底前交货。这是整套材料里最有辨识度的一份,也是最能说明这次审核在审什么的一份。免税不是一项中心取得之后可以慢慢取用的身份,它绑定在已经下单、并且在当年内到货的设备上——所以承诺函是必备材料,不是辅助材料。

第六份是专职研究与试验发展人员名册原件,含前面说的五个字段。

一式四份,每份加盖企业公章,按列明次序装订。

退回的原因是可预期的,而且多数不是实质不合规,是内部不一致。

设备汇总表与交货承诺函对不上。汇总表里列了没有合同支撑的设备,或者合同签了、交货日期落在下一年。两份材料是放在一起读的,不用外部核查就能看出错位。

验资报告过期。通知要的是”最近一次”,意味着卷宗要反映申报时的状态,而不是设立时的状态。设立之后没再做验资的中心,交上去的是一份描述着一家已经不存在的公司的文件。

名册与工资表、社保名单对不上。这一处最要紧,因为它直接指向 80 人这道线,而且审核人员很容易验。名册上 82 个名字、社保名单上 71 个,不是四舍五入的差别,是对同一个问题给出的不同答案。

审计报告的主体错了。非独立法人的中心,审计的对象是它所在的外商投资企业,中心自身的成本在这份审计里表现为一个分部,而不是一套账。把中心内部的管理报表当法定审计交上去,交的是审核人员用不了的东西。

申请书写了通知没问的事。通篇讲集团对华承诺、中心的战略意义、转移了什么技术,却不谈三道门槛的信,读起来像是申请人没弄清这次在决定什么。

这些都不是法律障碍。每一条都是联席会议暂缓的理由,而暂缓的代价是一个季度,因为补正期间审核时钟是停的。通知明确写了申请人补正材料的时间不计入 45 个工作日。作为规则这是公道的,作为事实这是贵的。

审核链条与它的时钟

流程四步,一个截止点。

卷宗报到市政务服务中心综合窗口,也就是市商务局的窗口。商务局受理并作初步审核。初审同意后,商务局牵头召开审核部门联席会议,对材料进行审核,确定符合免税资格条件的研发中心名单。

结果以联合公告形式在市商务局网站发布。核定结果由商务局函告北京海关,抄送市财政局和市税务局,报送商务部外资司,并通知相关研发中心。不符合规定的,由商务局根据联席会议决定出具书面审核意见,说明理由。

时钟是自受理之日起 45 个工作日,扣除申请人补正材料的时间。这是一个实打实的承诺,也是有用的排期输入,但要连扣除项一起读。45 天量的是审核方,不是申请人;一份需要两轮补正的材料,前后可以跨一个季度。

还有第二条路径,对不在城中心的公司有意义。有几个主体承接了市商务局”牵头组织对外资研发中心采购设备免退税资格进行审核认定”的职权:经开区管委会,以及朝阳、海淀、通州、顺义、昌平、大兴六个区商务局。这些经赋权的单位可以参照这份通知执行。走这条路,程序略有不同:由赋权单位牵头召开联席会议审核,把符合条件的名单报市商务局;市商务局发函征求北京海关、市财政局、市税务局意见,一致同意后,由赋权单位在其政府官网发布公告并通知中心;市商务局再把核定结果函告北京海关、抄送两局、报送商务部。

差别是行政性的,不是实质性的——标准是同一套——但它改变了企业找谁、公告挂在谁的网站上。一家在经开区的中心跑到市级窗口递材料,并没有做错什么,只是绕了一段更长的路去开同一场会。

还有两处程序细节值得单独拎出来。

第一处是已缴税款的处理。对按规定应当免税而已经征税的进口货物,中心可以申请退还,并同时退还银行同期活期存款利息。其中有一项前置条件:已征税进口、且尚未申报增值税进项税额抵扣的,须先取得主管税务机关出具的《”十五五”期间支持科技创新进口税收优惠政策项下进口商品已征进口环节增值税未抵扣情况表》。这张表是中央管理办法的附件,取它这件事归税务条线,不归关务条线。

第二处是变更的处理。中心名称、经营范围发生变更的,应及时把变更情况说明报送市商务局;商务局按同一套程序核定,变更后的中心自变更登记之日起能否继续享受政策,并注明变更登记日期。核定结果同样函告海关、抄送财政与税务、报送商务部、通知中心。改了经营范围却不报送的中心,手里的核定结果已经不再描述它自己。

免的是什么,不免的是什么

有三条边界要说明白,因为每一条都常被放大。

第一条,免税实行清单管理。可以免税进口的货物,由财政部会同有关部门制定的清单确定;本期清单以财关税〔2026〕9 号公布,自 2026 年 1 月 1 日起实施。取得资格的中心进口清单之外的物品,那一件就不免税。资格是使用清单的前提,不是清单的替代。

第二条,免税的对象是为研发活动采购的进口设备,不是中心进口的所有东西。北京这份通知的标题就落在”采购进口设备”上,它要求的承诺函也关于设备合同和交货。进口样品、备件、耗材、办公设备的中心,不宜假定资格能覆盖到这些。中央的管理办法描述可覆盖的类别,清单描述可覆盖的货物;两者之间,留给假设的空间很小。

第三条,免的是进口关税、进口环节增值税和进口环节消费税。它不是所得税措施。取得资格不会因此得到更低的企业所得税税率,这项资格也不与高新技术企业、西部大开发等分项政策互动。那些是各自独立、各按各的条件作出的认定,把它们当成一个打包方案的公司在排期上会算错。

还有第四条边界,不那么显眼,但更有用。免税挂在”中心”身上,而”中心”是一个被定义过的申请人:要么是独立法人,要么是内设部门或者分公司。这个定义决定了谁签申请、交谁的审计报告、编谁的设备汇总表。一个在同一座城市里有好几个研发主体的集团,”哪家主体去申报”是一个真问题,答案往往不是最大的那家,而是账上能把三道门槛干净地做出来的那家。

对财务条线来说,实务上的结论是:资格要在申报之前先建模。收益是一批在特定年份到货的设备所对应的关税与进口环节税;成本是那份卷宗、那场会议、那个时钟,以及底层记录的常年维护。一年只进口一台仪器的中心,算下来可能很薄;一年之内把一个新实验室配齐的中心,通常算下来不小。

为什么北京只是样本,不是规则

中央文件对核定主体的写法,决定了整件事的地理分布。在可享受的主体类别里,外资研发中心由省级商务主管部门会同省级财政、税务部门和所在地直属海关核定。权限在省一级。没有全国统一的窗口、统一的表格、统一的公告。

这带来三个后果。

第一个是材料清单不同。北京要六份。另一个省可能要同样六份再加一份研发项目说明,或者在北京市要原件的地方接受复印件,或者要求不同的份数。准备多省申报的集团,不宜假定北京的卷宗可以原样复用。

第二个是时点不同。北京这份通知管到 2030 年底,并设了 45 个工作日的时钟。其他省份为同一时期发布了自己的办法,申报窗口和结果公布时间由地方定。按全国统一日历排期的中心,会遇上意外。

第三个是赋权路径不同。北京把权限下放给了一个经开区管委会和六个区商务局。这是北京特有的安排,反映的是这座直辖市的体量和外资在那几个区的集中度。在由省商务部门统一审核的省份,没有对应的结构。

对在多省设有研发主体的集团,合适的做法是把中央文件当成常量,把其余当成变量。三道门槛来自中央,术语定义来自中央,免税货物清单来自中央。窗口、附件、份数、会议、时钟、公告来自省里,而且应当一省一省去收集,而不是从第一个省外推。

还有一点容易漏。因为权限在省级、公告在本地,在一个省取得的资格,不是另一个省的资格。一家在北京的中心到别处再开一个研发场所,要在那个省、按那个场所自己的事实重新申报。把全国研发集中在一家主体、再带几个小型卫星团队的集团,应当早点决定资格挂在哪家主体上,因为这个答案决定了卷宗在哪里搭。

面向 2026—2030 年窗口的一张核对表

窗口五年,好用的用法是从设备倒推。

先看设备计划。哪些仪器、给哪个实验室、哪一年下单、哪一年到货?承诺函把交货日期变成了一份文件,所以设备计划和申报年份必须是同一个计划。在一个什么都不进来的年份申报,那封承诺函没有内容可写。

然后拿账去验三道门槛,不要拿商业计划去验。按中央管理办法的定义取投资总额或者研发总投入,看账上这个数字是否在 800 万美元之上。把名册拿去和工资表、社保名单对,对完再交,不是交完再对。把设备汇总表拿出来,确认它归集的是”购置”,而不是只剩”还在账上的”。

然后定主体。非独立法人的中心,要确认其所在外商投资企业最近一次的验资报告存在且是最新的,并且那份”为设立和建设本研发中心而投入”的资产清单能出原件,而不是事后拼出来。

然后按规定的次序把卷宗装好,一式四份、加盖公章,用那六个附件模板。文件是外文的,附中文翻译并附翻译单位或者翻译人员信息——材料规范是这么要求的。

然后把时钟记进日程。自受理起 45 个工作日,扣除补正时间。在忙的季度递进去,出结果大概率在下一个季度;需要为 12 月到货的设备取得资格的中心,递得足够早,才不会因为一次暂缓把核定推到下一年。

最后,把维护也排进去。这项资格不是”取得之后不用管”的那种。名称或者经营范围变更会触发一次重新核定;对税务、商务部门的报送义务继续存在;设备记录要持续支撑得住那张汇总表。这些都不重,而如果中心的账从一开始就和母公司分开记,全都会轻很多。

顺带说一句:这份通知是写给谁的。它是写给一家已经设立、已经在采购设备的中心的。它帮不上一家还在决定要不要在华设研发中心的公司,也不适合当作选址比较的工具——它复述的门槛是全国性的,它加上的程序是地方性的。比较不同城市的集团,应当比较程序:窗口、时钟、赋权路径、附件;门槛则当作既定条件。

最后一句,关于怎么读这份通知。它是一份程序性文件,程序性文件通常被读它的期限。更有用的读法是读它的假设。它假设申请人已经读过中央的管理办法,已经知道投资总额和研发总投入怎么定义,已经有一份对得上的名册,已经提前一年安排设备采购。满足这些假设的公司,会觉得这份通知把一件可管理的事写清楚了;不满足的公司,会发现 45 个工作日根本不是那道真正的约束。

来源

  • 北京市人民政府·《北京市商务局 北京市财政局等 4 部门关于做好外资研发中心采购进口设备免税资格审核有关工作的通知》(2026-08-18 印发,2026-09 公开):https://www.beijing.gov.cn/zhengce/zhengcefagui/202609/t20260920_4872405.html
  • 北京市商务局·《关于做好外资研发中心采购进口设备免税资格审核有关工作的通知》全文及六个附件下载页:https://sw.beijing.gov.cn/zwxx/2024zcwj/202609/t20260920_4872284.html
  • 北京市科技创新服务平台·《关于做好外资研发中心采购进口设备免税资格审核有关工作的通知》(2026-11-09 公开发布):https://www.ncsti.gov.cn/kjdt/tzgg/202611/t20261109_262109.html
  • 财政部 海关总署 税务总局·《关于”十五五”期间支持科技创新进口税收优惠政策的通知》(财关税〔2026〕7 号)全文:http://hangzhou.customs.gov.cn/hangzhou_customs/fileDir/resource/cms/article/6010991/7183589/2026060908504267182.pdf
  • 北京市政务服务网·牵头组织对外资研发中心采购设备免、退税资格进行审核认定(办事指南,含财关税〔2026〕9 号免税清单说明):https://banshi.beijing.gov.cn/pubtask/task/1/110105000000/fb564bc7-be4d-458d-adcd-00b1575a1923_app.html

Reading the 2026 National Cybersecurity Week Agenda as a Compliance List for Foreign Companies

  • The 2026 National Cybersecurity Week ran from 14 to 20 September under the theme of cybersecurity for the people in the intelligent age, with ten central departments as joint sponsors and the opening ceremony in Jinan.
    2026 年国家网络安全宣传周于 9 月 14 日至 20 日举行,主题为「网络安全为人民,网络安全靠人民——智能时代 网安护航」,由十部门联合主办,开幕式在济南举行。
  • The week is neither a compliance obligation nor a new rule; it creates no filing duty and moves no threshold, and treating it as a regulatory event is the first error to avoid.
    宣传周既不是合规义务,也不是新规;它不创设申报义务,也不移动任何门槛,把它当成一次监管事件是首先要避免的错误。
  • Its diagnostic value lies in the agenda: closed-door sessions on critical information infrastructure, open-source supply chain and cloud service security, plus forums on automotive data and personal information.
    它的诊断价值在议程里:关键信息基础设施、开源软件供应链、云计算服务安全三场闭门会,以及汽车数据与个人信息两场分论坛。
  • MLPS Grading is the entry ticket, and it runs as five steps — grade, file, build out, test, supervise — with testing expected from level two and annually from level three.
    等级保护是入场券,按定级、备案、建设整改、测评、监督检查五步推进,二级以上应测评,三级以上原则上每年一次。
  • A Supply Chain Review question arises from two triggers, not from foreign ownership: critical information infrastructure procurement, and a platform with more than a million users listing abroad.
    供应链安全审查的触发点是两个,而不是外资身份:关键信息基础设施采购,以及掌握超过一百万用户个人信息的平台赴国外上市。
  • Data Classification decides which export route is available, and a handler of important data owes an annual risk assessment report to the provincial authority.
    数据分类分级决定可走哪条出境路径,重要数据处理者还须每年向省级主管部门报送风险评估报告。
  • Incident Reporting is where a global playbook and a local clock diverge, because the local duty does not wait for a fully confirmed root cause.
    事件报告是全球预案与本地时限分岔之处,因为本地义务并不等到根本原因完全确认才启动。

Reading the 2026 National Cybersecurity Week Agenda as a Compliance List for Foreign Companies | 从 2026 年国家网络安全宣传周议程读出外资企业的合规清单

What the week is, and what it is not

The 2026 National Cybersecurity Week ran from 14 to 20 September, organised jointly by ten central departments — the Publicity Department, the Cyberspace Administration, the Ministry of Education, the Ministry of Industry and Information Technology, the Ministry of Public Security, the People’s Bank of China, the National Radio and Television Administration, the All-China Federation of Trade Unions, the Communist Youth League Central Committee and the All-China Women’s Federation. The opening ceremony was held in Jinan on 14 September. The theme was cybersecurity for the people, cybersecurity by the people, with the intelligent age and its escort appended as the 2026 suffix.

The first thing to say about the week, and the thing that compliance teams most often get wrong, is that it is a campaign and not an instrument. It creates no filing obligation, imposes no new duty, moves no threshold and repeals nothing. No system has to be registered because the week occurred, and no assessment has to be refreshed because a forum was held. A company that treats the week as a regulatory event will spend September producing documents that no one asked for, and will miss the thing the week is actually good for.

The confusion is understandable, because China’s cybersecurity regime does produce new instruments at a steady rate, and a company that has learned to scan the calendar for new rules will scan this one too. But the regime’s density is exactly why a campaign matters. When the instruments are numerous, the binding constraint on a compliance function is not information about the rules; it is a view of which rules the supervisor currently considers live. A rule that has been in force for four years without a public enforcement story is a different risk from a rule that has been in force for four years and appears twice on a forum agenda. The week is the cheapest available signal of that difference, and it is free.

There is also a practical asymmetry worth noting. A new rule arrives with a text, a commencement date and usually a transition period, which means it can be planned. A shift in supervisory emphasis arrives without any of those things. It shows up as a question in an inspection that the company did not expect, and by then the answer has to be produced on the spot. Reading the agenda is a way of anticipating the question rather than answering it under time pressure.

What the week is good for is diagnosis. The agenda is published, the topics are chosen, and the choice is made by the same departments that write the rules and run the inspections. A programme of fifteen thematic sub-forums, with three topics taken in closed session, is not a compliance requirement, but it is a reasonably reliable statement of what the regulators believe is currently unsolved. Read that way, the week is the one moment in the year when the enforcement agenda is visible in advance and in public.

There is a second, softer function. In practice the week is also the period in which agencies schedule briefings, publish guidance and, in some sectors, run inspection campaigns. Nothing in the campaign documents makes any of that a legal obligation, and a company should not treat the calendar as a deadline. But a company that has a genuine open item — an unfiled system, an overdue test, a data flow with no route — should expect that September is a month in which someone may ask about it.

The distinction matters because it determines what a compliance function should do with the week. It should not build a workstream. It should take the agenda, hold it against its own programme, and use the gaps as a prioritisation input for the year that follows.

Reading the agenda as a map

The 2026 programme is unusually legible. Alongside the opening ceremony and the technology summit, there were fifteen thematic sub-forums, and the titles describe the areas where the organisers think work remains.

Three were closed-door sessions rather than open forums: one for chief security officers of critical information infrastructure operators, one on open-source software supply chain security, and one on cloud service security. Closed-door format is itself information. These are the topics where the discussion is more useful than the announcement, and where the participants are expected to speak about their own arrangements rather than listen to a policy summary.

The open forums covered a wider set: collaborative defence, security insurance, government information system security management, awareness and literacy, personal information protection, artificial intelligence applied to cybersecurity, artificial intelligence security governance, electronic product information erasure, and automotive data security. The opening ceremony released the third version of the artificial intelligence security governance framework, together with the results of an application testing programme for AI-enabled security and a batch of registered security labels for consumer connected cameras.

The three closed-door topics are worth separating out, because each one maps onto a question that a foreign-invested group tends to answer too quickly. Critical information infrastructure is the question of status: whether the entity has been designated, whether it knows, and what follows if it has. Designation is not self-assessed and not published in a list that a company can read; the practical consequence is that a group should be able to say what it would do if the designation arrived, rather than assert that it will not. Open-source supply chain is the question of composition: whether the company can produce, for a system it operates, the list of components it contains and where they came from. Most enterprises cannot, and the reason is not negligence but the fact that modern software is assembled rather than written. Cloud service security is the question of boundary: whether the company knows which of its data sits in a service it does not control, in which region, under whose key management.

Taken together the three closed-door sessions describe a single shift. The supervisory question has moved from whether a company has deployed controls to whether it can describe what it runs. That is a harder question, because it cannot be answered by a policy document.

Two readings follow. The first is that artificial intelligence has moved from principle to instrumentation: a framework in its third version, a testing programme with published results, and a labelling scheme for a consumer device category are the marks of a regime that is being operated rather than declared. The second is that the supply chain has become a first-class subject. Open-source components, cloud services and connected devices are three faces of the same question — whether the thing a company runs is the thing it thinks it runs — and the programme put all three on the agenda in the same week.

For a foreign-invested company the practical move is a mapping exercise, not a reading exercise. Take the eight or nine topics, and for each one ask a single question: what would an inspector ask me for, and could I produce it this week. The answers sort themselves quickly into three groups — items that are documented and current, items that are documented and stale, and items that exist only as an intention. The third group is the work plan.

What follows is that mapping, organised as the five things a foreign-invested enterprise is expected to have, with the single most common point of failure attached to each. The list is not novel. What is useful is the pairing: each obligation has one failure mode that recurs often enough to be worth naming.

Step one and two: the inventory and the grade

The first thing an enterprise is expected to have is an inventory of its systems and data, and the second is a grade for each system that matters. They are treated together because the second is worthless without the first.

An inventory that satisfies an inspector is not an asset register maintained by the finance function and not a service catalogue maintained by the information technology function. It is a list of systems that process data, each with an owner, a hosting location, a network boundary, a data category and a downstream recipient. The reason the distinction matters is that the three lists diverge, and the divergence is exactly where the findings are. A marketing automation tool bought by a regional office, a dealer portal run by a distributor, a plant execution system connected to a group manufacturing platform, a customer service tool embedded in a global licence — none of these necessarily appear in a service catalogue, and all of them process data.

A usable inventory has five fields per system, and the fifth is the one companies omit. The first is the operator, meaning the legal entity that runs the system and would be named in a filing. The second is the hosting location, physical or cloud, including the region. The third is the network boundary, meaning what the system can reach and who can reach it. The fourth is the data category, at the level of granularity the company actually uses. The fifth is the downstream recipient: the group platform, the vendor, the distributor or the overseas service that receives data from the system, and the mechanism by which it does so. Without the fifth field the inventory answers a cybersecurity question and not a data question, and the two regimes described below cannot be joined.

The most common single failure at this step is an inventory built from the organisation chart rather than from the network. The consequence is predictable: the systems that are missed are the ones that were adopted without central approval, and those are also the ones with the weakest controls. A second, related failure is a mismatch between the entity on the registration and the entity that operates the system. A group may file under its China holding company while the operating subsidiary is the actual operator, which produces a filing that is formally complete and substantively wrong.

Grading follows. The level assigned to a system turns on the object that would be harmed by its compromise and the degree of harm, and the discipline of the exercise is that the answer has to be argued rather than assumed. Once a system is graded at level two or above, it is filed with the public security authority, and the filing is the point at which the grade becomes a matter of record.

The most common single failure at this step is grading too low. A manufacturing execution system, an ERP instance holding production and cost data, or a logistics platform serving a large number of counterparties is often graded as an internal system when its compromise would reach beyond the enterprise. The second failure is grading once and never again. A system that was level two when it held internal records may be level three after it is connected to a customer portal, and the change is not always accompanied by a re-grading. The grade is a statement about the current system, not about the system as originally built.

The legal floor matters here. The Regulations on the Administration of Network Data Security require a network data processor to strengthen data security protection on the basis of cybersecurity graded protection. That phrase does the work of connecting two regimes that companies often run as separate programmes: the grade is not only a cybersecurity filing, it is the baseline on which the data security obligations are built. A company that treats graded protection as an information technology chore and data security as a legal chore ends up with two programmes that disagree about which systems exist.

Step three: testing that produces evidence

The third thing an enterprise is expected to have is a test result, and the fourth is evidence that the findings were closed. These are usually described as one step. They are two, and the second is the one that is examined.

Testing is a measurement at a point in time, carried out by an institution with the relevant qualification, against a defined set of control requirements. The output is a report with a score and a list of findings, sorted by severity. The report has a limited life, and a company that is asked for it in September and produces one from two years ago has produced evidence of a problem rather than evidence of compliance.

Remediation is the part that is examined, and it is examined through the file rather than through the system. The question an inspector asks is not whether a finding was fixed but whether the company can show that it was fixed: a change record, a configuration screenshot with a date, a re-test result, a compensating control with a documented rationale and a review date. A finding closed by assertion is a finding that will be reopened.

What an inspector samples is worth describing, because it shapes what the file should contain. In practice the sample is small and deep: two or three findings, followed from the report through the change record to the re-test. A company with a thin file on three findings looks worse than a company with a thick file on ten, because the inspection is testing whether the process works rather than whether the score is high. The implication is counter-intuitive and worth stating: it is better to have a small number of fully evidenced closures and an honest list of open items than a clean report with no trail behind it.

The most common single failure at this step is a remediation file that contains a plan but no closure. The second is a gap analysis that was never costed, so the high-severity items were deferred year after year and the test report became a document that everyone signs and no one acts on. The third, less obvious, is that the remediation is scoped to the tested system rather than to the class of systems. A finding about an unpatched component on one server is, almost always, a finding about the patch process, and fixing the server leaves the process untouched.

There is a sequencing point worth making because it is often reversed in practice. Build-out comes before testing in the standard order of the five steps, and the order is not decorative. A test run before remediation measures the old design and produces findings that will be found again next year. Companies that test early because a procurement deadline requires a report, and then remediate afterwards, end up paying for two tests and one report that describes a system that no longer exists.

Step four: the Supply Chain Review question

The fourth thing an enterprise has to be able to answer is whether a procurement or a listing triggers a cybersecurity review. This is the item most often answered by assertion rather than by analysis.

The review measures of 2022 set out two triggers. The first is procurement of network products and services by an operator of critical information infrastructure, where the procurement affects or may affect national security; the operator is required to make that prediction itself and to file where the answer is affirmative. The second is a network platform operator holding personal information of more than one million users that seeks to list abroad, which must file. The definition of network products and services is broad and specific at the same time: core network equipment, important communications products, high-performance computers and servers, large-capacity storage, large databases and application software, cybersecurity equipment, cloud computing services, and other items with a significant bearing on critical information infrastructure, cybersecurity and data security.

Two features of the regime are worth stating plainly. The review is not a prohibition, and it is not a general licence for procurement. It is a process with a defined sequence and defined clocks: a decision on whether review is needed, an initial review with a stated period, a special review procedure for cases where the member departments disagree, and a set of risk factors that the review is required to assess. And the review is not the only screening that may apply: the measures provide expressly that where data security review or foreign investment security review is separately regulated, those requirements apply as well.

The uncertainty that companies find hardest to manage is designation. Critical information infrastructure is identified by the protection departments of the State Council under the regulations, and the identification is communicated to the operator rather than published as a register. A foreign-invested company therefore often cannot prove a negative, and the honest position is not “we are not designated” but “we have not been notified, and here is what we would do if we were”. That position is defensible, and it is more useful than an assertion, because it forces the company to answer the follow-up question in advance: who would run the review, what would the procurement inventory look like, and how long would a filing take.

The most common single failure at this step is to answer the trigger question with the wrong fact. Foreign ownership is not a trigger; it is a reason to expect more questions, but the legal test is the operator status and the procurement. A company that concludes it is outside the regime because it is foreign-invested has answered a question that was not asked. The second failure is to treat the review as a vendor due diligence exercise and to stop there. The measures require the operator to require, through procurement documents and contracts, that the provider commit not to use its position to obtain data unlawfully, not to control or manipulate the customer’s equipment without cause, and not to interrupt supply or necessary technical support without justification. Those three commitments are contract terms, they cost nothing to insert, and they are frequently absent from the template agreements that a group uses globally.

A third failure is more subtle and worth flagging for groups that procure centrally. Where the contracting entity is outside China and the operating entity inside China uses the service, the Chinese operator may not see the procurement at all. The trigger test attaches to the operator’s use of the product, not to who signed the purchase order.

Step five: Data Classification and the export route

The fifth thing an enterprise needs is a classification of its data, because every export route depends on it.

The statutory base is the Data Security Law, which establishes a system of classified and graded data protection and a catalogue of important data. The 2025 network data regulation fills in the operating detail: the national coordination mechanism coordinates the making of the important data catalogue, regions and departments determine their own catalogues, processors identify and declare important data, and where data is confirmed as important the responsible region or department must notify the processor or publish the determination. That last provision matters in both directions. A processor is not required to guess, but a processor that has been told, or that operates in a sector where the catalogue has been published, cannot claim ignorance.

The regulation adds two obligations that a foreign-invested group should read together. A handler of important data must carry out an annual risk assessment of its network data processing activities and submit a report to the provincial-level competent authority, and the report must cover, among other things, the data export situation. And a processor handling personal information of ten million individuals or more is made subject to the provisions that apply to handlers of important data on security officers, management bodies and reporting. The second provision catches consumer-facing businesses that would not describe themselves as handling important data at all.

The export route then follows from the classification. Personal information may leave under a security assessment, a certification, a standard contract, or one of the statutory exemptions; important data must go through a security assessment. The free trade zone negative lists discussed elsewhere in this collection change the routing inside a zone, not the classification.

The classification exercise itself is usually run on three axes, and the order matters. The first axis is the object: whether the data concerns national security, economic operation, public health, personal rights or commercial interest. The second is the degree of harm, which is where most arguments occur, because harm is easy to assert and hard to bound. The third is the scale, which for personal information is a count of individuals and for other data is a matter of coverage and precision. A classification that starts from the third axis — counting records — will produce a defensible answer for personal information and a poor one for everything else, which is the shape of most classification failures in industrial and trading companies.

The most common single failure at this step is to equate personal information with data. A company that has correctly concluded that a flow carries no personal information may still be exporting important data, because important data is defined by its significance rather than by whether an individual is identifiable. Production yields, capacity utilisation, inventory levels, pricing positions and supply chain maps are the standard examples, and they travel in the same integration feeds as the innocuous data around them. The second failure is to treat a transfer within the group as not a transfer. The test is whether the data leaves the territory, not whether it leaves the legal entity.

Incident Reporting, and a calendar for the year

The obligation that ties the other five together is reporting, because it is the moment at which everything else becomes visible to someone outside the company.

The duties sit in several instruments and they are cumulative. The Cybersecurity Law, as amended with effect from 1 January 2026, provides the general framework and the penalties; the amendment raised the penalties for a range of network security violations and extended the law’s reach. The network data regulation adds specifics: where a defect or vulnerability in a product or service endangers national security or the public interest, the processor must report to the competent authority within twenty-four hours; where a data security incident occurs, the processor must activate its contingency plan, take measures to prevent the harm from spreading and report as required, and where the incident harms the rights of individuals or organisations it must notify the affected parties. The regulation also requires a processor to report suspected criminal conduct to the public security and state security authorities.

The most common single failure at this step is a collision of clocks. A group that operates under a seventy-two-hour notification regime, and under an internal rule that notification follows confirmation of a root cause, will find that the local duty is not built that way: the trigger is the occurrence of the incident, the report is due as required by the authority’s rules, and waiting for confirmation is a decision to report late. The second failure is authority. In many groups the local entity cannot report without clearance from a regional or global function, and the clearance process takes longer than the local window. The fix is a pre-agreed protocol that names who may report, on what trigger, without further approval.

Such a protocol has four elements and takes about a day to write. It names the reporting entity and a deputy. It defines the trigger in local terms rather than in the group’s terms, so that an incident is reportable when it meets the local threshold and not only when the group’s severity scale says so. It states that the local report may go out before the group’s own assessment is complete, and that the group will be informed in parallel rather than in advance. And it records who is authorised to speak to the authority, because a report is a statement by the company and a second, inconsistent statement is worse than a late one.

Set against those five items, the year has a shape. Graded protection testing recurs annually for level three systems and on the cycle applicable to level two. The important data risk assessment is annual. The personal information compliance audit is periodic. Filings have to be updated when a system changes grade or a new system is brought into scope. Contingency plans have to be exercised, not merely written. A company that lays these on a calendar discovers that the workload is uneven and that September, the month of the campaign, is a poor month to add anything.

Which returns the argument to where it began. The week does not create the obligations and does not change them. It publishes, once a year, the list of things the supervisors are still worried about, and it does so in a form that a compliance function can use as a check rather than as a task. The value is in the check. A company that reads the agenda, holds its own programme against it, and finds that four of the five items are documented and current has learned something worth knowing. A company that reads it as a deadline has learned nothing and will spend the month producing paper.

Sources

  • 国家网络安全宣传周官网·2026 年国家网络安全宣传周开幕式在山东济南举行(2026-09-14):https://www.gjwlaqxcz.cn/web/news/detail.html?id=235
  • 中国网信网·中华人民共和国主席令(第六十一号)——《全国人民代表大会常务委员会关于修改〈中华人民共和国网络安全法〉的决定》自 2026-01-01 起施行(2025-10-28):https://www.cac.gov.cn/2025-10/29/c_1763461514706581.htm
  • 新华网·引导人工智能发展——网络安全法完成修改(2025-10-28,含罚则调整要点):https://www.news.cn/politics/20251028/b3400522666a4b3f8c844b9fa6c31131/c.html
  • 中国网信网·《网络安全审查办法》(国家互联网信息办公室等十三部门令第 8 号,2022-02-15 施行):https://www.cac.gov.cn/2022-01/04/c_1642894602182845.htm
  • 中国政府网·《网络数据安全管理条例》(国务院令第 790 号,2025-01-01 施行):https://www.gov.cn/gongbao/2024/issue_11646/202410/content_6980863.html

从 2026 年国家网络安全宣传周议程读出外资企业的合规清单

宣传周是什么,不是什么

2026 年国家网络安全宣传周于 9 月 14 日至 20 日举行,由中央宣传部、中央网信办、教育部、工业和信息化部、公安部、中国人民银行、国家广播电视总局、全国总工会、共青团中央、全国妇联十部门联合主办。开幕式于 9 月 14 日在济南举行。主题是「网络安全为人民,网络安全靠人民——智能时代 网安护航」。

关于这一周,第一句要说清、也是合规团队最常搞错的一句是:它是一场宣传活动,不是一件监管工具。它不创设申报义务,不新增任何职责,不移动任何门槛,也不废止任何规定。没有哪个系统因为宣传周举行而需要登记,也没有哪份评估因为开了论坛而需要重做。把宣传周当成一次监管事件的企业,会在九月生产一批没人要的文件,同时错过这一周真正有用的地方。

这种混淆可以理解,因为中国的网络安全制度确实在以稳定的节奏出文件,而一家已经养成「扫日历找新规」习惯的企业会把这个日历也扫一遍。但正因为制度密集,宣传活动才有意义。当规范多到一定程度,合规职能的约束条件就不再是「规则的信息」,而是「监管方当下认为哪些规则是活的」这一判断。一条施行四年、却从未有过公开执法故事的规定,与一条施行四年、在议程上出现两次的规定,风险不一样。宣传周是能拿到这个差别的最便宜信号,而且免费。

还有一处实际的错位值得注意。新规到来时带着文本、施行日,通常还有过渡期,这意味着它可以被规划。监管重心的移动不带这些东西中的任何一样。它表现为检查中一个企业没预料到的问题,而到那时答案必须当场给出。读议程,是提前想到那个问题,而不是在压力下回答它。

它真正有用的地方是诊断。议程是公开的,议题是选出来的,而选议题的人就是写规则、跑检查的那几个部门。十五场专题分论坛、其中三个议题以闭门会形式进行,这不是合规要求,但它是一份相当可靠的陈述:监管方认为哪些问题目前尚未解决。这样读,宣传周就是一年之中唯一一次,执法议程提前且公开可见的时刻。

还有一层更软的功能。实践中,宣传周也是各部门集中安排宣讲、发布指引,并在某些行业开展检查的时间段。宣传周的文件里没有任何一条把这变成法律义务,企业也不该把日历当成截止日。但一家手里确实有未了事项的企业——一个没备案的系统、一次逾期的测评、一条没有路径的数据流——应当预期九月是可能有人来问的月份。

这个区分之所以要紧,是因为它决定合规职能该拿这一周做什么。它不该为此立项。它该把议程拿过来,对一遍自己的方案,把差距当作下一年度的排序输入。

把议程当地图读

2026 年的议程异常清晰。除开幕式与技术高峰论坛之外,共有十五场专题分论坛,标题本身就说明了主办方认为哪些地方还有活要干。

其中三场是闭门会而不是公开论坛:一场面向关键信息基础设施运营者的首席安全官,一场谈开源软件供应链安全,一场谈云计算服务安全。闭门这个形式本身就是信息。这些议题上,讨论比宣布更有用,与会者被期待讲自己的安排,而不是听政策综述。

公开论坛覆盖面更宽:协同防御、网络安全保险、政务信息系统安全管理与实践、意识和素养提升、个人信息保护、人工智能技术赋能网络安全应用、人工智能安全治理、电子产品信息清除、汽车数据安全。开幕式发布了《人工智能安全治理框架》第三版,同时发布了人工智能技术赋能网络安全应用的测试结果,以及一批消费类网联摄像头的网络安全标识备案产品。

三场闭门会的议题值得单拎出来,因为每一个都对应一个外资集团往往回答得太快的问题。关键信息基础设施是「身份」问题:主体有没有被认定、知不知道、被认定了之后要做什么。认定不是自我评估,也不在一份企业可以查的名单上;实际的推论是,集团应当能说清「认定真的来了我会怎么做」,而不是断言它不会来。开源软件供应链是「构成」问题:对一个自己在运行的系统,公司能不能列出它包含哪些组件、这些组件从哪里来。多数企业列不出来,而原因不是疏忽,是现代软件是装配出来的、不是写出来的。云计算服务安全是「边界」问题:公司知不知道自己的哪些数据放在一个它不控制的服务里、在哪个区域、由谁做密钥管理。

三场闭门会合起来描述的是一次位移。监管的问题,从「企业有没有部署控制措施」移到了「企业能不能描述自己在运行什么」。后一个问题更难,因为它无法用一份制度文件回答。

由此可以读出两件事。第一,人工智能已经从原则走到了工具层面:一个出到第三版的框架、一个有公开结果的测试项目、一套针对某一消费设备类别的标识备案,这些都是一套制度在被运行、而不只是被宣告的标志。第二,供应链成了一等议题。开源组件、云服务、联网设备是同一个问题的三张面孔——企业运行的东西,是不是它以为的那个东西——而议程把这三张面孔放在同一周里。

对外资企业而言,实际动作是一次映射,而不是一次阅读。把八九个议题拿过来,对每一个只问一个问题:检查的人会向我要什么,我这周拿得出来吗。答案很快会分成三堆:有文件且是最新的、有文件但已过时的、只停留在打算上的。第三堆就是工作计划。

下面就是这次映射,按外资企业被期待具备的五件事组织,每一件配一项最常见的失分。这份清单并不新颖。有用的是配对:每一项义务都配一个反复出现到值得命名的失败方式。

第一步与第二步:清单与定级

企业被期待具备的第一件东西是系统与数据的清单,第二件是每个重要系统的等级。两者放在一起讲,因为离开第一件,第二件没有意义。

一份经得起检查的清单,不是财务口维护的资产台账,也不是信息技术口维护的服务目录。它是一份处理数据的系统清单,每个系统带负责人、托管位置、网络边界、数据类别与下游接收方。

一份能用的清单,每个系统有五个字段,而第五个是企业最常漏掉的。第一是运营者,即运行该系统、并会在备案中被点名的法律主体。第二是托管位置,物理机房或云端,含区域。第三是网络边界,即该系统能触达什么、谁能触达它。第四是数据类别,颗粒度按企业实际使用的来。第五是下游接收方:接收该系统数据的集团平台、供应商、经销商或境外服务,以及接收所经由的机制。缺了第五个字段,清单回答的是网络安全问题,而不是数据问题,下面要讲的两套制度也就接不上。

这个区分之所以要紧,是因为三份清单会分叉,而分叉处正是问题所在。区域办公室自行采购的营销自动化工具、经销商运营的门户、连着集团制造平台的工厂执行系统、嵌在全球授权里的客服工具——这些都不一定出现在服务目录里,而它们全都在处理数据。

这一步最常见的单项失分,是按组织架构而不是按网络来建清单。后果可以预料:被漏掉的系统,正是那些未经集中审批就被采用的系统,而它们也是控制最弱的系统。第二种相关的失分,是备案主体与系统运营主体不一致。集团可能以中国控股公司名义备案,而实际运营者是业务子公司,于是得到一份形式上完整、实质上错误的备案。

定级随之而来。一个系统定在哪一级,取决于它被侵害时会损害什么客体、损害到什么程度,而这项功课的纪律在于:答案要论证,不能假定。系统定在二级及以上,要向公安机关备案,而备案正是等级成为记录的那一刻。

这一步最常见的单项失分,是定级偏低。制造执行系统、装着生产与成本数据的核心业务系统、服务大量交易对手的物流平台,常被当作内部系统定级,而它们一旦出事,影响会超出企业自身。第二种失分是定一次、此后不再动。一个装着内部记录时是二级的系统,在接上客户门户之后可能应当是三级,而这个变化并不总伴随着重新定级。等级是关于当前系统的陈述,不是关于系统最初建成时样子的陈述。

这里有一条法律底线值得点出。《网络数据安全管理条例》要求网络数据处理者在网络安全等级保护的基础上加强网络数据安全防护。这句话的作用,是把企业常常分开运行的两套制度接上:等级不只是一份网络安全备案,它是数据安全义务赖以搭建的地基。把等级保护当成信息技术的杂活、把数据安全当成法务的杂活的企业,最后会有两套对「哪些系统存在」意见不一致的方案。

第三步:能产出证据的测评

企业被期待具备的第三件东西是一份测评结果,第四件是问题项已关闭的证据。这两件常被说成一步。它们是两步,而被查的是第二步。

测评是一次时点测量,由具备相应资质的机构,对照一套确定的控制要求完成。产出是一份带分数和问题清单的报告,问题按严重程度排序。报告有寿命,一家在九月被要求提供、却拿出一份两年前报告的企业,提供的是问题的证据,而不是合规的证据。

整改才是被查的部分,而查的方式是查卷宗,不是查系统。检查者问的不是问题有没有修好,而是企业能不能证明它修好了:一条变更记录、一张带日期的配置截图、一次复测结果、一项写明了理由与复核日期的补偿性控制。以断言方式关闭的问题项,是会被重新打开的问题项。

检查取样什么,值得描述,因为它决定了卷宗里该放什么。实践中取样量小、深度大:两三个问题项,从报告顺着变更记录追到复测。三个问题项上卷宗很薄的企业,看起来比十个问题项上卷宗很厚的企业更糟,因为检查考的是流程是否运转,不是分数是否漂亮。由此有一个反直觉、但值得说清的推论:少数几项证据完整地关闭、加上一份诚实的未关闭清单,比一份干净却没有痕迹的报告更好。

这一步最常见的单项失分,是有整改计划而没有关闭。第二种是差距分析从未被计价,于是高危项年复一年地被推后,测评报告变成一份人人签字、无人执行的文档。第三种不那么显眼:整改的范围被限定在被测系统,而不是那一类系统。一台服务器上未打补丁的组件,几乎总是一个关于补丁流程的问题,修好那台服务器,流程原封不动。

有一个次序问题值得说,因为实践中常被倒过来。按五步的标准顺序,建设整改在测评之前,而这个顺序不是装饰。在整改之前做的测评,测的是旧设计,产出的问题明年还会再被发现一次。因采购节点需要一份报告而提前测评、事后再整改的企业,最后付两次测评的钱,其中一份报告描述的是一个已经不存在的系统。

第四步:供应链审查这道题

企业必须能回答的第四件事,是某次采购或某次上市是否触发网络安全审查。这一项最常被用断言回答,而不是用分析回答。

2022 年的审查办法给出两个触发点。第一是关键信息基础设施运营者采购网络产品和服务,且该采购影响或者可能影响国家安全;运营者须自行预判,答案为肯定时申报。第二是掌握超过一百万用户个人信息的网络平台运营者赴国外上市,必须申报。网络产品和服务的定义既宽又具体:核心网络设备、重要通信产品、高性能计算机和服务器、大容量存储设备、大型数据库和应用软件、网络安全设备、云计算服务,以及其他对关键信息基础设施安全、网络安全和数据安全有重要影响的品类。

这套制度有两个特征值得直说。审查不是禁止,也不是采购的通行证。它是一个有既定顺序和既定时钟的流程:先确定是否需要审查,再进入有明确期限的初步审查,成员单位意见不一致时进入特别审查程序,并配有一组必须评估的风险因素。审查也不是唯一一道程序:办法明确规定,数据安全审查、外商投资安全审查另有规定的,应当同时符合其规定。

企业最难处理的不确定性是认定本身。关键信息基础设施由国务院保护工作部门依规认定,认定结果告知运营者,而不以名录形式公布。因此外资企业往往无法证明一个否定命题,而诚实的立场不是「我们没被认定」,而是「我们尚未收到告知,若收到,我们的动作是这样的」。这个立场站得住,也比一句断言更有用,因为它逼着企业提前回答后续问题:谁来牵头审查、采购清单长什么样、申报要走多久。

这一步最常见的单项失分,是用错误的事实回答触发问题。外资身份不是触发点;它是预期会被多问几句的理由,但法律上的检验是运营者身份与采购行为。一家因为自己是外资就断定不在制度范围内的企业,回答的是一个没有被问的问题。第二种失分是把审查当成供应商尽职调查,做到那里就停。办法要求运营者通过采购文件、协议,要求产品和服务提供者承诺:不利用提供产品和服务的便利条件非法获取用户数据,无正当理由不非法控制和操纵用户设备,无正当理由不中断产品供应或必要的技术支持服务。这三项承诺是合同条款,写进去不花钱,而集团全球模板协议里常常没有。

第三种失分更微妙,值得提示给集中采购的集团。当签约主体在中国境外、而使用服务的主体在中国境内时,中国的运营者可能根本看不到这次采购。触发检验附着于运营者对产品的使用,不附着于谁签了采购单。

第五步:数据分类分级与出境路径

企业需要的第五件东西是数据分类分级,因为每一条出境路径都取决于它。

法律基础是《数据安全法》,它确立了数据分类分级保护制度与重要数据目录制度。2025 年施行的《网络数据安全管理条例》把操作细节补齐:国家数据安全工作协调机制统筹协调有关部门制定重要数据目录,各地区、各部门确定本地区、本部门以及相关行业、领域的重要数据具体目录,数据处理者按规定识别、申报重要数据,确认为重要数据的,相关地区、部门应当及时告知数据处理者或者公开发布。最后这一句在两个方向上都有意义。企业不需要猜,但一家已经被告知、或在目录已发布的行业里经营的企业,不能主张不知情。

条例还加了两项义务,外资集团应当连起来读。重要数据的处理者应当每年度对网络数据处理活动开展风险评估,并向省级以上有关主管部门报送风险评估报告,报告内容中明确包含网络数据出境情况。处理一千万人以上个人信息的网络数据处理者,还要遵守条例对重要数据处理者在安全负责人、管理机构与报告方面的规定。后一条会圈住那些自认为完全不碰重要数据的消费类业务。

出境路径随后由分类分级推出。个人信息可以走安全评估、认证、标准合同或法定豁免之一;重要数据必须走安全评估。其他地方讨论的自贸试验区负面清单,改变的是区内走哪条路,不是分类分级本身。

分类分级这项功课通常按三个轴做,而顺序有讲究。第一轴是客体:数据涉及国家安全、经济运行、公共卫生、个人权益,还是商业利益。第二轴是危害程度,争论多半发生在这里,因为危害容易主张、难以界定。第三轴是规模,对个人信息是个自然人计数,对其他数据是覆盖面与精度。从第三轴起步——先数记录条数——的分类,会给个人信息一个站得住的答案,给其他一切一个糟糕的答案,而这正是工业与贸易企业里多数分类失败的样子。

这一步最常见的单项失分,是把个人信息等同于数据。一家正确判断出某条数据流不含个人信息的企业,仍可能在出境重要数据,因为重要数据的定义看的是重要性,而不是能否识别到某个自然人。产量、产能利用率、库存水平、报价头寸、供应链图谱是标准例子,而它们和周围无害的数据坐在同一条集成管道里。第二种失分是把集团内部传输当成不是传输。检验的是数据是否出境,不是是否出门。

事件报告,与一年的日历

把其余五项串起来的那项义务是报告,因为它是其他一切被公司外部看见的那一刻。

相关义务分散在几部规范里,而且是累加的。《网络安全法》经修改后自 2026 年 1 月 1 日起施行,提供总体框架与罚则;此次修改提高了一系列网络安全违法行为的处罚,并扩大了法律的域外适用情形。条例补上具体:网络产品、服务存在安全缺陷、漏洞等风险,涉及危害国家安全、公共利益的,应当在二十四小时内向有关主管部门报告;发生网络数据安全事件,应当立即启动预案、采取措施防止危害扩大并按规定报告,对个人、组织合法权益造成危害的,还应当通知利害关系人。条例还要求,处置过程中发现涉嫌违法犯罪线索的,向公安机关、国家安全机关报案。

这一步最常见的单项失分,是时钟相撞。一家同时运行七十二小时通报制度、并在内部规定「确认根本原因后再通报」的集团,会发现本地义务不是这么设计的:触发点是事件发生,报告按主管部门规定的时限报送,等待确认等于决定迟报。第二种失分是权限。很多集团里,本地实体无权在未获区域或全球职能放行前报告,而放行流程比本地窗口更长。解法是一份事先约定的规程:写明谁可以报告、在什么触发点报告、无需再行审批。

这样一份规程有四个要素,写下来大约一天。它点名报告主体与一名替补。它用本地口径而不是集团口径定义触发点,使事件在满足本地门槛时就应当报告,而不是只在集团的严重度分级说应当时才报告。它写明本地报告可以在集团自身评估完成之前发出,集团同步获知而不是事先批准。它还记录谁有权对主管部门陈述,因为报告是公司的陈述,第二份不一致的陈述比一份迟到的陈述更糟。

把五项摆在一起,一年就有了形状。三级系统的等级测评按年进行,二级按适用周期;重要数据风险评估按年;个人信息合规审计按周期;系统定级变化或新系统纳入范围时,备案要更新;应急预案要演练,而不只是写出来。把这些排在日历上的企业会发现工作量并不均匀,而九月——宣传周所在的月份——并不是适合再加事情的时候。

这就把论证带回起点。宣传周不创设义务,也不改变义务。它一年一次地公布监管方仍在担心的事情,形式是一份合规职能可以拿来当检查表、而不是当任务单的东西。价值在检查。一家读了议程、把自己的方案对上去、发现五项里有四项有文件且是最新的企业,知道了值得知道的事。一家把它读成截止日的企业,什么也没学到,还会花掉一个月生产纸张。

来源

  • 国家网络安全宣传周官网·2026 年国家网络安全宣传周开幕式在山东济南举行(2026-09-14):https://www.gjwlaqxcz.cn/web/news/detail.html?id=235
  • 中国网信网·中华人民共和国主席令(第六十一号)——《全国人民代表大会常务委员会关于修改〈中华人民共和国网络安全法〉的决定》自 2026-01-01 起施行(2025-10-28):https://www.cac.gov.cn/2025-10/29/c_1763461514706581.htm
  • 新华网·引导人工智能发展——网络安全法完成修改(2025-10-28,含罚则调整要点):https://www.news.cn/politics/20251028/b3400522666a4b3f8c844b9fa6c31131/c.html
  • 中国网信网·《网络安全审查办法》(国家互联网信息办公室等十三部门令第 8 号,2022-02-15 施行):https://www.cac.gov.cn/2022-01/04/c_1642894602182845.htm
  • 中国政府网·《网络数据安全管理条例》(国务院令第 790 号,2025-01-01 施行):https://www.gov.cn/gongbao/2024/issue_11646/202410/content_6980863.html

Zhejiang and Anhui Free Trade Zones Publish 2026 Data Export Negative Lists Two Days Apart

  • Zhejiang issued its 2026 Negative List on 9 November 2026 and Anhui followed on 11 November, each replacing an earlier version while preserving filing results already notified.
    浙江于 2026 年 11 月 9 日印发 2026 版负面清单,安徽于 11 月 11 日跟进,两省均废止旧版,已取得备案结果通知书的备案结果继续有效。
  • The list decides which outbound data still needs a security assessment, a standard contract or certification; data outside the list may leave the zone without any of the three.
    清单决定哪些出境数据仍需走安全评估、标准合同或认证;清单之外的数据出境,可以免予申报这三条路径。
  • Zhejiang kept the e-commerce business-to-business and clearing fields and added eleven Commodity Trading entries covering seven scenarios and thirty-eight data items.
    浙江保留电子商务(企业对企业)与清结算两个领域,新增十一条大宗商品条目,覆盖七个场景、三十八个数据项。
  • Zhejiang compressed the sequence into Filing In Two Steps — submit a filing application, then transfer lawfully — and replaced the earlier three-stage route.
    浙江把流程压缩为「提交备案申请—合规开展数据出境」两步办理,取代原先的三段式路径。
  • A clause that is easy to miss permits Cross-Regional Reference to another zone’s list only through a five-stage provincial approval ladder ending in national filing.
    一条容易被忽略的条款规定,参照适用其他自贸区清单须经五级省级审批链条,最终报国家层面备案。
  • A filing result is not a permit to export: notification, separate consent and impact assessment survive the list, and export control or technology export rules apply on top.
    备案结果不是出境许可:告知、单独同意与个人信息保护影响评估并不因清单而免除,出口管制与技术出口管理规定另行叠加。
  • Anhui’s measures, published two days later, keep the same two-list architecture, the same two-step filing route and a ten-working-day first review at the municipal tier.
    安徽两天后出台的办法沿用同样的双层清单结构、同样的两步备案路径,并由市级管理部门在十个工作日内完成初审。

Zhejiang and Anhui Free Trade Zones Publish 2026 Data Export Negative Lists Two Days Apart | 浙江、安徽自贸试验区相隔两日发布 2026 版数据出境负面清单

Two provinces, two days, one architecture

On 9 November 2026 the Cyberspace Administration of Zhejiang Province, the Zhejiang Provincial Department of Commerce and the Zhejiang Provincial Data Bureau jointly issued two documents: trial measures for managing the negative list for outbound data transfers in the China (Zhejiang) Pilot Free Trade Zone, and the negative list itself in its 2026 edition. The pair was published on the Zhejiang cyberspace portal on 10 November at four minutes past four in the afternoon, and it carries the provincial cyberspace office’s 2026 document number three.

Two days later, on 11 November, the Cyberspace Administration of Anhui Province, the Anhui Provincial Department of Commerce and the Anhui Provincial Data Resources Administration issued the mirror-image pair for the China (Anhui) Pilot Free Trade Zone. That notice appeared on the Anhui cyberspace portal at ten minutes to four in the afternoon on the same day it was signed, with both attachments — the trial measures and the list — posted as separate files.

The two provinces are not coordinating for the first time, and the timing is less remarkable than the architecture. Both instruments are built as a pair: a management measure that governs who may use the list, how a filing is made, how the list is revised and how compliance is supervised, plus a list that names the data. Both rest on Article 6 of the Provisions on Promoting and Regulating Cross-Border Data Flows, the March 2024 regulation that authorised pilot free trade zones to draw up their own lists of data requiring a security assessment, a standard contract or certification, subject to approval by the provincial cyberspace committee and filing with the national cyberspace and data authorities. Both cite the same political mandate: the fourth plenary session of the twentieth Central Committee and its call to promote efficient, convenient and secure cross-border data flows.

Where they differ is in economic content and in administrative detail, and those differences are what a company actually has to work with.

Two provinces publishing within a week is a reminder of how this layer of the regime is built. The national provisions set the frame and the thresholds; the zones set the sector lists. There is no national negative list for outbound data, and there is not going to be one, because the instrument was designed to be local. That has a consequence for how a group should staff the question: the person who can answer it sits close to the zone, not at the centre. A head office that maintains one China data export position is maintaining a position that is, in a strict sense, not true of any particular entity.

The Zhejiang notice contains a transitional provision that is easy to read past. The 2024 negative list and the original management measures are repealed from the date of the notice, but filing results for which a notification has already been issued remain valid. The repeal is therefore not retroactive in effect. A company holding a 2024 filing result does not have to re-file on 10 November; it holds a valid result until something changes in its own data flows. Anhui’s notice does not carry an equivalent sentence, which is consistent with the fact that this is its first list rather than a replacement.

What the list decides, and what it does not

The most common misreading of a free trade zone negative list is that it is a permission to export. It is not. It is a routing instrument, and the routing has only two outcomes.

Under Article 6 of the national provisions, a processor inside a pilot free trade zone that transfers data outside the list may be exempted from filing a security assessment, from concluding a standard contract and from obtaining certification. That is the exemption. A processor transferring data that is on the list falls back into the ordinary regime: the assessment route for the categories the list assigns to assessment, and the standard contract or certification route for the categories it assigns to those. The list is therefore drawn as two sub-lists, and both provinces say so explicitly. Zhejiang’s list is organised as a list requiring assessment and a list requiring a standard contract or certification. Anhui’s measures state that the list shall contain at least those two parts, and describe what each part covers.

The second common misreading is that the list settles the question of important data. It partly does, and only inside the zone. Article 2 of the national provisions already provides that a processor need not declare data as important data if no department or region has notified it or published it as such. The zone list adds a local determination: within the named industries and the named scenarios, data of the kind described is treated as requiring the stated route. Outside those industries and scenarios, the national rules apply unchanged. Zhejiang’s list says this in a note — that each entry is limited to specified scenarios, and that data outside them follows the ordinary rules.

The mechanics of the split deserve a paragraph of their own, because the two sub-lists are not two degrees of the same test. The assessment sub-list describes data whose transfer requires the national security assessment: important data in the named fields, and personal information above the volume band that the list sets for that field. The standard contract and certification sub-list describes personal information in the band below. Read together, the two sub-lists form a staircase. At the top sits assessment; in the middle, contract or certification; at the bottom, a band in which data may move after filing, and only within the named scenarios. The staircase is the facilitation. It does not lower the obligation that attaches at each step. It fixes where each step begins, which is precisely what a company could not previously determine for a commodity flow or for business-to-business e-commerce data. A trader arguing from analogy had to guess where the first step started; a trader reading the list can see the riser.

That has a corollary that companies tend to resist. Because the staircase is drawn by scenario and by data item, a flow that sits one field outside a described scenario does not inherit the band. It falls to the national thresholds, which are set by volume alone, take no account of the commercial context and may sit higher or lower than the band the list draws for that field. Purpose, not data type, is what moves a flow between one route and another, and purpose is documented in the contract and in the system design rather than in the data itself.

The third misreading concerns what survives. A filing result confirms which route applies. It does not discharge the obligations that attach to the transfer itself. Notification, separate consent for sensitive personal information, a personal information impact assessment and the record-keeping that goes with them are obligations under the Personal Information Protection Law and the Regulations on the Administration of Network Data Security, and nothing in either provincial instrument touches them. Nor does the list displace export control or technology export administration: both provinces provide that where a transfer involves controlled items or technology export matters, those regimes apply on their own terms. Zhejiang’s list carries the point as a note; Anhui’s measures carry it as a substantive article.

There is a fourth point, smaller but practically important. A zone list applies to processors registered in the zone. Anhui defines the processor as an enterprise, public institution, organisation, group or other entity, or an individual, registered in the Anhui pilot free trade zone and carrying out cross-border data activities. A group that operates in the province but is registered outside the zone is not covered by the zone’s list, and must use the national pathways. Registration address, not operational footprint, is the test.

Zhejiang: a commodity chapter and a shorter sequence

Zhejiang’s 2026 edition keeps the frame of the 2024 edition. The two original fields remain: e-commerce between businesses, and clearing and settlement. On top of that, the province added a commodity trading chapter of eleven entries covering seven scenarios and thirty-eight data items, and the provincial question-and-answer describes this as filling a gap — before the addition, commodity traders had no directly applicable list and had to reason by analogy from the e-commerce or clearing entries.

The scenarios named in the published text are specific rather than generic: raw material procurement, equipment maintenance and technical support, commodity sales, digital services for bulk trade, bunkering, offshore transit trade, and insurance claims. Each entry is confined to its scenario, which means a commodity trading company has to match its actual flow to a named scenario rather than to the industry as a whole. A company that moves contract and settlement data for a transaction that is none of the seven is outside the list and falls back on the national rules.

Thirty-eight data items across eleven entries is a ratio worth pausing on. Roughly three and a half items per entry means the entries are not broad categories; they are enumerated fields. That granularity is what allows the list to be used as a checklist rather than as an occasion for judgement, and it is also what makes it brittle. A field that is not enumerated is not covered, and reading an entry by its title rather than by its items is the most reliable way to reach the wrong answer. The provincial text says as much in a note: each entry is limited to the scenarios listed, and the scenarios are described by the fields they contain.

The commercial consequence is a change in who does the work. Under an analogical regime, the analysis sat with the company’s lawyers and produced a memorandum that argued a position. Under an enumerated list, the analysis sits with whoever owns the data dictionary, and it produces a mapping. A company that has not maintained a field-level inventory of its outbound feeds cannot use the list well, because the list asks a question — is this field in this scenario — that only the inventory can answer. The list has therefore made data governance, in a fairly literal sense, a precondition for using a compliance channel.

The commodity chapter matters for a reason that is easy to miss in a summary of entry counts. Zhejiang’s free trade zone is built around a commodity resource allocation hub, and the data that moves with a cargo of oil, iron ore or grain is not consumer data. It is price data, inventory data, counterparty data and logistics data, and much of it sits close to the definition of important data without ever being named as such. Putting it into a list with named scenarios gives a trader something it did not have: a documented answer to the question of whether a particular feed needs an assessment, rather than an argument from analogy conducted during an inspection.

The procedural change is the other half. The question-and-answer states that the filing sequence has been compressed from three stages — submit an application, submit a filing, transfer lawfully — into two: submit a filing application, then transfer lawfully. The province describes the resulting regime as a closed loop of filing in advance, monitoring during the activity and assessment afterwards, and it says the list is subject to dynamic management, with revisions carried out as implementation and risk are tracked.

Two things follow for a company already operating under the 2024 regime. First, the shorter sequence changes the internal approval path, not the evidentiary standard. A filing application still has to be supported by a description of the data, the scenario and the volume, and the provincial text is explicit that implementing guidelines and detailed rules are still to be issued. Until those appear, the practical content of the two steps is the same as before, with one gate removed. Second, the service channel named in the question-and-answer is the data cross-border service centre in the Binjiang district of the Hangzhou zone, which handles receipt of materials, feedback of results, inspection and filing coaching. For a company in another zone — Ningbo, Jinyi or Zhoushan — the centre is a template rather than a counter.

The reference clause

Buried in the ninth article of Zhejiang’s measures, and in the eighth article of Anhui’s, is a provision that no summary of the lists tends to mention, and that will decide a number of real cases.

Both provinces allow reference to a negative list formally published by another pilot free trade zone, free trade port, reform and opening-up pioneer area or national pilot for comprehensive opening-up of the services sector. Both then attach conditions that turn an apparent shortcut into an administrative process. The reference must be for a specific field in which the province genuinely needs it. It must sit inside the national framework for classified and graded data protection. The provincial departments must carry out a needs analysis and a risk assessment for each list proposed for reference, one by one. The result must pass review by the provincial data security coordination mechanism, then be approved by the provincial cyberspace and informatisation committee, and then be filed jointly by the provincial cyberspace office and the provincial data authority with the national cyberspace department and the national data department.

Five stages, two of them at committee level, and a national filing at the end. The design is not obstruction for its own sake. A zone list is an exercise of a delegated power, and a decision to adopt another province’s exercise of the same power has to be traceable to the body that holds the delegation. The ladder is the trace. It also explains why the clause is written as a permission for the province rather than as a right for the company: what is being approved is the province’s adoption of a standard, not the company’s choice of a forum.

The clause is best understood as an answer to a question that would otherwise be answered informally. A company in Zhejiang whose sector is covered by Shanghai’s list, or a company in Anhui whose sector is covered by Zhejiang’s new commodity chapter, cannot simply adopt the other province’s entries. It can ask its provincial authorities to consider reference, and the authorities have a defined route for doing so, but the route has no published timetable and no obligation to reach a positive conclusion.

For planning purposes the clause has a clear implication. A group with operations in several zones should not assume that a list obtained in one zone travels to another. It should map, zone by zone, which list covers which entity and which flow, and treat a gap as a gap until a reference decision is actually published. The alternative — building a compliance position on an assumed reference — is the kind of assumption that survives until the first inspection.

Anhui: same skeleton, different municipal tier

Anhui’s package is structurally identical and administratively more explicit in places, which makes it a useful comparison.

The measures define the provincial tier as five bodies: the cyberspace office, the data resources administration, the commerce department acting as the office of the free trade zone, the public security department and the state security department. The municipal tier is defined by geography: the management bodies of the Hefei, Wuhu and Bengbu zones, together with the relevant functional departments at the same level. That two-tier split is where the practical work sits, because the municipal tier receives the filing, organises the review and issues the initial opinion.

The published text gives the municipal tier ten working days from receipt of the filing materials to organise the review and produce an initial opinion on whether the outbound data falls within the list. That opinion goes to the provincial tier for confirmation, and the result is then fed back to the processor. The outcome has three branches, and they are worth separating because companies tend to collapse them. Where the data is on the list, the municipal tier directs the processor to file a security assessment, conclude a standard contract or obtain certification. Where the data is off the list, the municipal tier tells the processor in terms that the data may flow safely and in an orderly manner. Where the list does not apply at all — because the entity or the activity falls outside its scope — the ordinary law applies.

A first list behaves differently from a revision, and the difference is not only transitional. A revision inherits a body of decided cases: filings granted, questions answered, edge cases settled between the zone and its processors over a year of operation. A first list starts without that sediment, and the published text compensates by writing procedure at a finer grain — five provincial bodies named, three municipal zones named, ten working days fixed, three outcomes enumerated, two supervisory mechanisms described. Anhui has written a manual where Zhejiang could write a revision note. For a company, the practical reading is that Anhui’s answers are likely to be more predictable at the first encounter and less settled at the margin, because there is no accumulated practice to consult when a case does not fit the enumerated outcome.

Anhui also writes down two mechanisms that Zhejiang handles more briefly. The first is consistency sampling: during the validity of a filing, the municipal tier compares the data actually transferred against the filed materials, and where a processor has concealed or misreported, or has deliberately caused the actual flow to diverge from the filing, the consequences run from suspension and a deadline for rectification, through a fresh filing after rectification, to termination of the outbound activity and enhanced supervision. The second is a credit-and-risk approach to supervision, under which processors with stronger data security capability, a better credit record and lower-risk outbound flows may face reduced inspection frequency.

Two boundaries are stated as substantive rules rather than notes. Data concerning national security, the lifelines of the national economy, important livelihood matters or major public interest is core data, is subject to a stricter regime and is not brought within the negative list at all. And where an international treaty or agreement to which China is a party provides otherwise, the treaty applies, except for provisions on which China has entered a reservation. The second is unusual enough to be worth flagging: it does not create an exemption, but it does mean that a transfer covered by a treaty is analysed under the treaty first.

Anhui’s list content is published as an attachment, and the notice itself does not summarise the fields or the entry count. That is a difference in drafting habit rather than in substance, and it has one practical consequence: a company working on Anhui has to open the attachment, whereas a company working on Zhejiang can begin from the question-and-answer. Anhui’s measures also attach a reference rule for classifying and grading data, which is used to identify important data where a national sectoral authority has not published its own standard. The rule works from the scale of personal information held, from personal information held by operators of critical information infrastructure, and from high-value sensitive data generated in research, design and production.

Four steps for a company

The two lists reward a sequence, and the sequence is the same in both provinces.

The first step is to locate the entity, not the business. The question is which legal person is registered in which zone, because the list attaches to the registered processor. A group with a trading company in Ningbo, a manufacturing company in Hefei and a regional headquarters in Shanghai has three different answers to the same question, and the answer for one entity says nothing about the others.

The second step is to read the list for that zone, in its attachment form, and to read the field definitions rather than the field names. Zhejiang’s e-commerce chapter is expressly confined to business-to-business e-commerce and lists the scenarios it covers; a business-to-consumer flow is not inside it merely because the words e-commerce appear in the title. The same discipline applies to the commodity chapter’s seven scenarios.

The third step is to characterise the flow. A single business process usually produces several flows with different characters: order data, payment data, logistics data, after-sales data, employee data on secondment, and system telemetry sent to a group platform. Each has to be tested against the list separately, because the list is written by scenario and by data item, not by business unit. The most frequent error at this stage is to characterise the whole relationship instead of each flow, and to conclude either that everything is covered or that nothing is.

The fourth step is to choose the route and document the choice. If the flow is on the list, the route is the one the list assigns, and the work is an assessment, a standard contract or a certification, with the timelines those processes carry. If the flow is off the list, the route is a filing in the zone, and the deliverable is a filing result. If the entity is outside the zone, neither applies and the national thresholds govern. Whichever branch is taken, the file should record the reasoning, because the consistency sampling described in Anhui’s measures and the inspection practice in both provinces test the record against the actual flow, not against the conclusion.

An example makes the third step concrete. A European manufacturer seconds three engineers to its Chinese subsidiary for eighteen months. Payroll data, performance reviews and expense records for those engineers flow back to the group’s human resources platform in the home country. Three characterisations are available. The flow may be human resources management under the national provisions, which carry their own exemption. It may be personal information inside an e-commerce or clearing scenario, which it plainly is not. Or it may be employee data outside every listed scenario, in which case the list does not apply and the national thresholds decide. The third reading is correct in most cases, and it is the reading a company reaches only if it has separated the secondment flow from the commercial flows around it. Groups that characterise by relationship rather than by flow usually miss the secondment entirely, because no one in the commercial organisation owns it.

One sequencing point deserves emphasis. A filing is not a substitute for the substantive route, and the substantive route is not a substitute for the transfer-level obligations. Companies that treat the three as alternatives tend to arrive at an inspection with a filing result, no impact assessment and an unresolved question about whether the receiving entity is a separate controller. The list does not answer that question and does not pretend to.

What to watch next

Three things will determine how much these lists change in practice.

The first is the implementing guidance. Zhejiang’s question-and-answer states that implementation guidelines and management rules are to be issued, covering filing review, changes and termination of filings, supervision and inspection, with time limits for each stage. Until those appear, the two-step sequence is a direction rather than a procedure, and companies should expect the material requirements to be restated in the guidance.

The second is the reference decisions. The clause exists in both provinces, and the first published reference decision will show how the five-stage ladder works in practice: how long it takes, what a needs analysis looks like, and whether a reference is granted field by field or in a block. A group with a sector that is listed in one zone and unlisted in another has a direct interest in that answer.

The third is the revision cycle. Both provinces describe dynamic management, and Anhui adds that where the content of a list changes, a fresh filing is required. That is a reasonable rule and a demanding one: it means the filing is tied to a version of the list, and a revision reopens the file. Companies should keep the version they filed against, and should expect that the 2026 editions are not the last.

A fourth question sits behind those three, and it is the one a group with several Chinese entities will ask first: whether the lists converge. There is reason to expect convergence in architecture and divergence in content. The architecture is fixed by Article 6 and by a common drafting template — a management measure, two sub-lists, a filing route, a reference clause. The content is fixed by local industry, and local industry differs. A commodity chapter in Zhejiang, a first list in Anhui and a medical device cluster in Beijing will produce lists that do not resemble one another, and a group operating across all three will hold three compliance positions rather than one. That is not a defect in the design. It is what a sector-based instrument does when it is applied province by province, and the reference clause is the mechanism the drafters left for closing the gaps that result.

None of this changes the underlying structure, which is now visible across the two provinces and is likely to hold: a pair of documents, a two-part list, a filing route inside the zone, the national pathways outside it, and a provincial approval ladder for borrowing a neighbour’s list. For a foreign-invested company, the value of the arrangement is not that the obligations are lighter. It is that the question of which route applies now has a documentary answer, and a documentary answer is something a compliance function can audit.

Sources

  • 浙江省互联网信息办公室、浙江省商务厅、浙江省数据局·《关于印发〈中国(浙江)自由贸易试验区数据出境负面清单管理办法(试行)〉〈中国(浙江)自由贸易试验区数据出境负面清单(2026版)〉的通知》(浙网办〔2026〕3 号,2026-11-09):https://www.zjwx.gov.cn/col/col1673564/art/2026/art_b4db8400a4064cf2ba76d47935cd2836.html
  • 浙江网信网·《中国(浙江)自由贸易试验区数据出境负面清单管理办法(试行)》和《中国(浙江)自由贸易试验区数据出境负面清单(2026版)》政策问答(2026-11-10):https://www.zjwx.gov.cn/col/col1673564/art/2026/art_a5f9bb9ee9fa4437bce8f1dda49cbb37.html
  • 安徽省互联网信息办公室、安徽省商务厅、安徽省数据资源管理局·《关于印发中国(安徽)自由贸易试验区数据出境负面清单管理办法及负面清单(2026版)的通知》(2026-11-11,含两份附件):https://www.ahwx.gov.cn/gzzc/bbdt/202611/t20261105_9720746.html
  • 中国政府网·《促进和规范数据跨境流动规定》(国家互联网信息办公室令第 16 号,2024-03-22 公布施行):https://www.gov.cn/gongbao/2024/issue_11366/202405/content_6954192.html

浙江、安徽自贸试验区相隔两日发布 2026 版数据出境负面清单

两省、两天、一套结构

2026 年 11 月 9 日,浙江省互联网信息办公室、浙江省商务厅、浙江省数据局联合印发两份文件:《中国(浙江)自由贸易试验区数据出境负面清单管理办法(试行)》与《中国(浙江)自由贸易试验区数据出境管理清单(负面清单)(2026 版)》。两份文件于 11 月 10 日 16 时 04 分在浙江网信网公布,文号为浙网办〔2026〕3 号。

两天之后,11 月 11 日,安徽省互联网信息办公室、安徽省商务厅、安徽省数据资源管理局印发了安徽自贸试验区的对应文件。该通知于签署当日 15 时 50 分在安徽网信网发布,管理办法与负面清单分别以两份附件形式挂出。

两省并非首次协同,时点也不是最值得注意的部分,值得注意的是一套结构。两省都是「一对文件」:一份管理办法,管谁能用清单、怎么备案、清单怎么修订、怎么监督;一份清单,点名哪些数据。两份文件都建立在《促进和规范数据跨境流动规定》第六条之上——这条 2024 年 3 月的规章授权自贸试验区自行制定需要纳入安全评估、标准合同或认证管理范围的数据清单,经省级网络安全和信息化委员会批准后,报国家网信部门、国家数据管理部门备案。两份文件援引的政治依据也相同:党的二十届四中全会关于「推进数据高效便利安全跨境流动」的部署。

差别在经济内容与行政细节,而这两处差别才是企业真正要用的东西。

两省一周之内相继发文,也提醒人这套制度是怎么搭起来的。国家规定给框架、给门槛,片区给行业清单。出境数据没有全国统一负面清单,也不会有,因为这件工具从设计上就是本地的。这对集团如何配置这个问题有影响:能回答它的人坐在片区附近,不在中心。一个只维护一套「中国数据出境立场」的总部,维护的是一套严格说来对任何一个具体主体都不成立的立场。

浙江的通知里有一条容易被读过去的过渡条款:2024 版负面清单及原管理办法自本通知印发之日起废止,但已取得备案结果通知书的,其备案结果继续有效。也就是说,废止不溯及既往。持有 2024 版备案结果的企业,不必在 11 月 10 日重新备案;只要自身数据流没有变化,那份结果就是有效的。安徽的通知没有对应表述,这与安徽此次是首份清单而非替换版这一事实相符。

清单决定什么,不决定什么

对自贸试验区负面清单最常见的误读,是把它当成一张出境许可证。它不是。它是一套分流装置,而分流只有两个出口。

按国家规定第六条,自贸试验区内数据处理者向境外提供清单之外的数据,可以免予申报安全评估、订立标准合同、通过认证——这是豁免的一侧。向境外提供落在清单之内的数据,则回到常规制度:清单指派给安全评估的类别走评估,指派给标准合同或认证的类别走那两条路。因此清单本身被画成两张子清单,两省都明确写了出来。浙江的清单按「需要通过数据出境安全评估的数据清单」与「需要通过个人信息出境标准合同备案、个人信息保护认证出境的数据清单」两部分编排;安徽的管理办法则规定,负面清单至少应当包含这两方面,并分别说明各自覆盖什么。

第二种常见误读,是以为清单把重要数据的问题一并解决了。它只解决了一部分,而且只在区内解决。国家规定第二条本来就已经写明:未被相关部门、地区告知或者公开发布为重要数据的,数据处理者不需要作为重要数据申报安全评估。区内清单在此之上加了一层本地判断:在所列行业、所列场景之内,符合描述的数据按清单指派的路径办理。出了这些行业与场景,国家规则原样适用。浙江清单用注释把这一点写明——每一条都限定于特定场景,场景之外按现行规定执行。

两张子清单的分工值得单独说一句,因为它们不是同一道门槛的两档严厉程度,而是一道阶梯。评估清单一侧,描述的是必须走国家层面安全评估的数据:所列领域中的重要数据,以及超过该领域量级档位的个人信息。标准合同与认证清单一侧,描述的是档位之下的个人信息。合起来读,就是一道阶梯:顶上是评估,中间是合同或认证,底下是经备案即可在所列场景内流动的一段。阶梯本身就是便利化所在。它不降低任何一级上的义务,它做的是把每一级从哪里起步钉死——而这恰恰是企业此前对一条大宗商品数据流、一条企业对企业电子商务数据流无法自行确定的。靠类推论证的贸易商只能猜第一级台阶在哪;读清单的贸易商看得见那块竖板。

由此有一个企业往往不愿接受的推论。既然阶梯是按场景、按数据项画的,落在被描述场景之外一格的数据流并不继承那个档位,而是回到国家门槛——那套门槛只看量级,不看商业语境,且可能高于也可能低于清单为该领域划出的档位。决定一条流落在哪一级的变量是目的,不是数据类型;而目的写在合同与系统设计里,不写在数据本身。

第三种误读关乎「什么被保留了下来」。备案结果确认的是走哪条路,它并不免除数据出境行为本身的义务。告知、处理敏感个人信息取得单独同意、个人信息保护影响评估,以及与之配套的记录留存,都是《个人信息保护法》与《网络数据安全管理条例》下的义务,两省文件对此一字未动。清单也不取代出口管制与技术出口管理:两省都规定,出境数据涉及管制物项相关技术资料或技术出口管理事项的,按相关法律法规另行执行。浙江把这一点写成清单注释,安徽写成实体条款。

还有第四点,较小但很实际。区内清单适用于在区内登记注册的数据处理者。安徽把数据处理者定义为在安徽自贸试验区内登记注册并开展数据跨境流动等相关活动的企业、事业单位、机构、团体、其他组织或个人。一家在安徽经营但注册在区外的集团,不适用区内清单,仍须走国家路径。判断标准是注册地址,不是经营足迹。

浙江:一个大宗商品章节,一段更短的流程

浙江 2026 版保留了 2024 版的框架。原有的两个领域仍在:电子商务(企业对企业)与清结算。在此之上,浙江新增了大宗商品领域十一条负面清单条目,覆盖七个具体场景、三十八个数据项。省里的政策问答把这一定位为「填补空白」——在增补之前,大宗商品企业没有直接对应的清单,只能从电子商务或清结算的条目类推。

公开文本点名的场景很具体,而不是笼统的行业名:原材料采购、设备维护与技术支持、大宗商品销售、大宗贸易数字化服务、燃油加注、离岸转口贸易、保险理赔。每一条都限定在自己的场景里,这意味着大宗商品贸易企业必须把实际数据流对上某个被点名的场景,而不是对上「大宗商品」这个行业。一笔交易若七个场景都不落,就在清单之外,回到国家规则。

大宗商品章节之所以要紧,原因在一份条目数摘要里看不出来。浙江自贸试验区围绕大宗商品资源配置枢纽建设,而随一船油、一船铁矿或一船粮食流动的数据并不是消费者数据。它是价格数据、库存数据、交易对手数据与物流数据,其中相当一部分贴着重要数据的边界,却从未被点名。把这些数据放进一份点名场景的清单,给贸易商的是一件此前没有的东西:对「某条数据流要不要评估」这个问题的书面答案,而不是一次检查中临场展开的类推论证。

程序变化是另一半。政策问答写明,企业申报流程已由「提交申请—提交备案—合规出境」三段,压缩为「提交备案申请—合规开展数据出境」两步。省里把由此形成的机制描述为事前备案、事中监测、事后评估的全链条闭环,并说明负面清单实行动态管理,跟踪评估实施情况和安全风险,统筹开展修订。

对已在 2024 版制度下运行的企业,有两点随之而来。第一,更短的流程改变的是内部审批路径,不是证据标准。备案申请仍要说明数据类型、场景与量级,而省级文本明确表示实施指南和管理细则仍待出台。在那些文件面世之前,两步的实际内容与从前相同,只是少了一道闸门。第二,政策问答点名的服务通道是杭州片区滨江区的数据跨境服务中心,承担资料接收、结果反馈、监督检查与申报辅导。对宁波、金义、舟山片区的企业而言,这个中心是模板,不是柜台。

参照条款

浙江管理办法第九条、安徽管理办法第八条里,藏着一条清单摘要通常不会提到的规定,而它将决定相当一批真实案例。

两省都允许参照适用全国其他省市自由贸易试验区、自由贸易港、改革开放先行区、国家服务业扩大开放综合试点示范地区正式发布的数据出境负面清单。紧接着,两省都附上了把这条看似捷径变成行政流程的条件。参照必须针对本省确有需要的特定领域;必须在国家数据分类分级保护制度框架下进行;省级管理部门须对拟参照的清单逐一开展需求分析和风险评估;结果须经省数据安全工作协调机制审核通过,报省委网络安全和信息化委员会批准,再由省网信办与省数据局联合报国家网信部门、国家数据管理部门备案。

五道环节,其中两道在委员会层级,末尾还有一次国家层面备案。这么设计不是为了刁难。区内清单是一项被授予的权力,而决定采纳另一个省对同一项权力的行使,必须能追溯到持有授权的机关。这条阶梯就是那条追溯线。它也解释了为什么条款被写成省里的许可,而不是企业的权利:被批准的是省里采纳一套标准,不是企业选择一个管辖地。

这条款最好被理解为对一个本来会被非正式回答的问题的正式回答。一家浙江企业所处的行业被上海清单覆盖,或一家安徽企业所处的行业被浙江新出的大宗商品章节覆盖,它都不能直接把别省的条目拿来用。它可以请本省主管部门考虑参照,主管部门也有既定路径可走,但这条路径没有公布时限,也没有必须给出肯定结论的义务。

从规划角度看,这条款的含义很清楚。在多片区布局的集团,不应假定在一个片区取得的清单可以平移到另一个片区。它应当按片区逐一梳理:哪份清单覆盖哪个主体、哪条数据流,并把空白当作空白,直到参照决定真正公布。反其道而行——把合规立场建立在一个假定的参照之上——是那种能撑到第一次检查之前的假定。

安徽:同一骨架,不同市级层级

安徽的文件在结构上完全一致,在若干处行政表述上更细,因此是个有用的对照。

管理办法把省级层面定义为五个部门:省互联网信息办公室、省数据资源管理局、省商务厅(省自贸办)、省公安厅、省国家安全厅。市级层面按地理定义:合肥、芜湖、蚌埠片区管理机构及属地相关职能部门。两级分工正是实际工作的落点,因为受理备案、组织审核、出具初审意见都在市级。

公开文本给出市级管理部门的时限:收到数据处理者提交的备案材料后,于十个工作日内组织审核,提出出境数据是否适用负面清单的初审意见。初审意见报省级管理部门确认后,向数据处理者反馈备案审核结果。结果分三支,值得分开看,因为企业往往把它们揉成一团。数据在清单内的,市级指导企业申报安全评估、订立标准合同或通过认证;数据在清单外的,市级明确告知企业可以安全有序流动;清单不适用的——因为主体或活动不在其范围内——按现行法律法规执行。

首份清单与修订版的行为方式不同,而这种不同不只是过渡性的。修订版继承一批既成案例:批过的备案、答过的问题、片区与企业之间在过去一年里磨出来的边缘情形。首份清单没有这层沉积,公开文本便以更细的程序颗粒度来补偿——五个省级部门点名、三个片区点名、十个工作日钉死、三种结果枚举、两套监督机制写明。安徽写的是一本手册,浙江可以只写一份修订说明。对企业而言,实际读法是:安徽的答案在第一次接触时更可预期,在边缘处则更不确定,因为当个案不落入被枚举的结果时,没有累积的做法可查。

安徽还写下了两套浙江处理得较简的机制。一是备案有效期内的一致性抽验:市级就数据实际出境情况与备案材料进行比对;对未严格履行承诺、瞒报虚报、故意造成实际出境数据与备案数据不一致的,处理从责令暂停出境、限期整改、整改后重新履行备案程序,到终止出境活动并采取重点监管措施。二是「信用+风险」分类分级监管:数据安全保障能力较强、信用较好、出境数据风险较低的数据处理者,可合理降低监督检查频次。

两条边界被写成实体规则,而不是注释。关系国家安全、国民经济命脉、重要民生、重大公共利益等数据属于国家核心数据,实行更严格的管理制度,不纳入负面清单管理。我国缔结或者参加的国际条约、协定与国内法律法规有不同约定的,适用该国际条约、协定,但我国声明保留的条款除外。后一条足够少见,值得标出:它不创设豁免,但确实意味着受条约覆盖的出境,先按条约分析。

安徽的清单内容以附件形式发布,通知本身没有概述领域或条目数。这是起草习惯的差异,而非实质差异,但它有一个实际后果:做安徽的企业必须打开附件,做浙江的企业可以从政策问答起步。安徽的管理办法还附了一份数据分类分级参考规则,用于在国家行业主管部门未发布本行业标准时识别重要数据。该规则从所持个人信息的规模、关键信息基础设施运营者掌握的个人信息,以及研发设计与生产制造过程中产生的高价值敏感数据入手。

企业侧四步

两份清单都奖励一种次序,而两省的次序相同。

第一步,定位主体,而不是定位业务。要问的是哪个法律主体在哪个片区注册,因为清单附着于注册的数据处理者。一家在宁波设有贸易公司、在合肥设有制造公司、在上海设有区域总部的集团,对同一个问题有三个不同答案,而其中一个主体的答案对另外两个什么也说明不了。

第二步,读那个片区的清单,读附件形态的清单,并且读领域定义而不是领域名称。浙江的电子商务章节明确限于企业对企业电子商务,并列出所覆盖的场景;一条企业对消费者的数据流,不会仅仅因为标题里有「电子商务」四个字就落进去。同样的纪律适用于大宗商品章节的七个场景。

第三步,给数据流定性。一个业务流程通常产出若干性质不同的流:订单数据、支付数据、物流数据、售后数据、外派员工数据,以及发往集团平台的系统遥测。每一条都要单独对照清单,因为清单是按场景、按数据项写的,不是按事业部写的。这一步最常见的错误,是给整段关系定性而不是给每条流定性,于是要么得出「全都在内」,要么得出「全都不在」。

第四步,选路径并留下选择的记录。数据流在清单内,路径就是清单指派的那条,工作是评估、标准合同或认证,附带各自的时间成本。数据流在清单外,路径是区内备案,交付物是一份备案结果。主体在区外,两者都不适用,由国家门槛管辖。无论走哪一支,卷宗都应当记录推理过程,因为安徽办法里的一致性抽验与两省的检查实践,比对的是记录与实际数据流,而不是与结论。

有一个例子可以把第三步讲实。一家欧洲制造商把三名工程师派驻中国子公司十八个月。这三人的薪酬数据、绩效评价与报销记录回流到母国集团的人力资源平台。三种定性都可能被提出:这可能属于国家规定下的跨境人力资源管理,那有它自己的豁免;可能属于电子商务或清结算场景内的个人信息,而它显然不是;也可能属于落在所有被列场景之外的员工数据,此时清单不适用,由国家门槛决定。多数情况下正确的读法是第三种,而只有把派驻数据流与周围的商业数据流分开的企业才会得到这个读法。按关系定性而不是按数据流定性的集团,通常会整个漏掉派驻这一块,因为商业组织里没有人负责它。

有一个次序问题值得强调。备案不是实体路径的替代品,实体路径也不是出境层面义务的替代品。把三者当成可选项的企业,往往在检查时手里有一份备案结果、没有影响评估,还悬着一个「境外接收方是不是独立控制者」的未决问题。清单不回答这个问题,也不假装回答。

接下来看什么

有三件事将决定这两份清单在实践中改变多少。

第一是实施指引。浙江的政策问答写明将加快出台实施指南和管理细则,细化各责任部门职责,梳理备案审核、备案变更终止、监督检查等内部流程,明确各环节办结时限。在这些文件面世之前,两步办理是方向而不是程序,企业应当预期实质要求会在指引里被重新表述一遍。

第二是参照决定。条款在两省都存在,而第一份公开的参照决定会显示五道环节在实践中如何运转:要多久、需求分析长什么样、参照是按领域逐项批准还是整块批准。行业在一个片区被列入、在另一个片区未被列入的集团,对这个问题有直接利益。

第三是修订周期。两省都写了动态管理,安徽还加了一句:负面清单内容发生变化的,应当重新申请备案。这是一条合理而苛刻的规则——它意味着备案绑定在清单的某个版本上,一次修订就重新打开卷宗。企业应当留存自己备案时所依据的版本,并预期 2026 版不是最后一版。

这三件事背后还有第四个问题,也是在中国设有多个实体的集团会最先问的:各省清单会不会趋同。有理由预期结构趋同、内容分化。结构由国家规定第六条与一套共同的起草模板固定下来——一份管理办法、两张子清单、一条备案路径、一条参照条款。内容由本地产业决定,而本地产业各不相同。浙江写出了大宗商品章节,安徽拿出了首份清单,北京经开区把医疗器械纳入清单,三者不会彼此相像;跨三地经营的集团,手里会是三套合规立场而不是一套。这不是设计缺陷。以行业为基础的工具,一旦按省适用,结果就是这样;而参照条款,正是起草者留给填补由此产生的空白的机制。

以上都不改变底层结构,而这个结构在两省之间已经显现,且很可能延续:一对文件、两张清单、区内一条备案路径、区外国家路径,以及一条借用邻省清单的省级审批阶梯。对外资企业而言,这套安排的价值不在于义务变轻了。它在于「走哪条路」这个问题现在有了书面答案,而书面答案,是合规职能可以审计的东西。

来源

  • 浙江省互联网信息办公室、浙江省商务厅、浙江省数据局·《关于印发〈中国(浙江)自由贸易试验区数据出境负面清单管理办法(试行)〉〈中国(浙江)自由贸易试验区数据出境负面清单(2026版)〉的通知》(浙网办〔2026〕3 号,2026-11-09):https://www.zjwx.gov.cn/col/col1673564/art/2026/art_b4db8400a4064cf2ba76d47935cd2836.html
  • 浙江网信网·《中国(浙江)自由贸易试验区数据出境负面清单管理办法(试行)》和《中国(浙江)自由贸易试验区数据出境负面清单(2026版)》政策问答(2026-11-10):https://www.zjwx.gov.cn/col/col1673564/art/2026/art_a5f9bb9ee9fa4437bce8f1dda49cbb37.html
  • 安徽省互联网信息办公室、安徽省商务厅、安徽省数据资源管理局·《关于印发中国(安徽)自由贸易试验区数据出境负面清单管理办法及负面清单(2026版)的通知》(2026-11-11,含两份附件):https://www.ahwx.gov.cn/gzzc/bbdt/202611/t20261105_9720746.html
  • 中国政府网·《促进和规范数据跨境流动规定》(国家互联网信息办公室令第 16 号,2024-03-22 公布施行):https://www.gov.cn/gongbao/2024/issue_11366/202405/content_6954192.html

Cross-Border Guarantee Validity and FX Registration: Reading Two Published Cases Together

  • Cross-Border Guarantee contracts remain valid where the foreign exchange filing was late or absent, because the filing is a supervisory step and not a condition of effectiveness.
    跨境担保合同不因外汇登记迟延或者未办理而无效,因为登记是监管步骤,不是生效要件。
  • Mandatory Provision analysis under Article 153 of the Civil Code asks what the rule requires when it is broken, not simply whether a rule was broken.
    《民法典》第一百五十三条的强制性规定分析,问的是规范在被违反时要求什么后果,而不只是问规范有没有被违反。
  • Contract Validity turns on the purpose of the rule and the substance of the transaction, which is why two cases published together reach opposite results.
    合同效力取决于规范目的与交易实质,这正是同批公布的两起案件结论相反的原因。
  • FX Registration must still be completed on time: an administrative penalty does not void the guarantee, but it brings consequences of its own.
    外汇登记仍须按期办理:行政处罚不会使担保无效,但它有自己的后果。
  • The distinction that decides cases is between approval, which can be made a condition of effectiveness, and registration or filing, which ordinarily is not.
    决定案件走向的分界,是批准与登记、备案之别:前者可以被设定为生效条件,后者通常不是。
  • A transaction whose real design is to route around customs supervision fails on substance, not on paperwork, and the court refers the leads to the enforcement authority.
    交易实质是绕开海关监管的,输在实质而不是手续,法院还会把违法线索移送行政执法部门。

Cross-Border Guarantee Validity and FX Registration: Reading Two Published Cases Together | 跨境担保的效力与外汇登记:把两起公开案例放在一起读

Two cases, one question

In November 2026 the newspaper of the legal establishment published a long commentary on how courts should draw the line between lawful innovation and regulatory avoidance in free trade zone transactions. The vehicle was a pair of cases from the Supreme People’s Court’s published series on free trade zone disputes. One upheld a cross-border guarantee that had been attacked on foreign exchange grounds. The other struck down a contract for bonded cross-border e-commerce purchases that had been built on invented consumer orders.

Read separately, each is a tidy illustration of a familiar rule. Read together, they answer the question that foreign-invested companies actually ask: if a regulator’s procedure was not followed, does the contract die?

The answer is that it depends on what the procedure is for, and on what the transaction actually does. That sounds like a lawyer’s hedge. It is not. It is a test with three steps, and the two cases sit at opposite ends of it.

The guarantee case is the one that matters most for treasury and legal teams, because cross-border guarantees are routine. A Chinese company supports its overseas subsidiary’s borrowing; a foreign parent supports its Chinese subsidiary’s borrowing; an offshore holding company provides security for an onshore project. The foreign exchange rules require these arrangements to be registered, and sometimes to be approved. Deals close, filings slip, and years later a counterparty argues that the guarantee was never effective because the filing was late. The guarantee case says that argument fails, and it says so by construing the foreign exchange rules rather than by weighing the equities.

What the commentary adds is a method rather than a result. It does not say that foreign exchange rules are unimportant, and it does not say that procedural defects are always curable. It says that a court should first identify what the provision is, then ask whether the provision sets the procedure as a condition of effectiveness, and then examine the purpose of the provision against the way the transaction actually ran. Applied to a guarantee, the method produces one answer; applied to a bonded import scheme built on invented orders, it produces the opposite answer. That is why the two cases were published together, and why reading either one alone is a mistake.

The bonded e-commerce case is the counterweight. There, the operator used a facilitation regime designed for individual consumers — bonded imports against real personal orders — to bring in bulk infant formula in small parcels and sell it offline. The court held the sale contracts void for breach of the Customs Law and the customs supervision rules, and sent the leads to the enforcement authority. The paperwork in that case may well have been complete. What failed was the transaction.

What Article 153 actually asks

The starting point is Article 153 of the Civil Code. A civil juristic act that violates a mandatory provision of a law or administrative regulation is void, except where that mandatory provision does not lead to the invalidity of the act. An act that offends public order and good morals is void.

The exception in the first paragraph is the whole game, and it is deliberately open-textured. Article 16 of the Supreme People’s Court interpretation on the general provisions of the contract part fills it in. Where a contract violates a mandatory provision, the court may hold that the contract is not void on that ground where the legislative purpose of the provision can be achieved by imposing administrative or criminal liability on the party, in any of five situations: the provision aims to protect public order but the actual performance has a manifestly minor effect on it, so that invalidity would make the outcome unfair; the provision protects state interests such as tax revenue or land transfer fees, or the lawful interests of others rather than the contracting parties, and upholding the contract does not frustrate that purpose; the provision requires one party to strengthen risk control or internal management, and the other party has no ability or duty to check compliance, so that invalidity would visit the consequences on the wrong party; the party was in breach at formation but has since become able to cure and refuses to do so in bad faith; or other situations provided by law or judicial interpretation.

Article 16 adds a second paragraph that is easy to miss and often decisive. Where the mandatory provision is aimed at regulating performance after formation, a party cannot rely on the breach to argue invalidity — unless performance necessarily entails the breach.

For a cross-border guarantee, two of the five situations do the work. The foreign exchange filing regime is a risk-control and statistical regime imposed on the guarantor and administered through banks; the foreign creditor has no practical ability to verify whether the Chinese guarantor filed, and no statutory duty to do so. Invalidating the guarantee would therefore impose on the creditor the consequences of the guarantor’s own administrative lapse, which is the third situation. And the purpose of the regime — monitoring cross-border contingent liabilities and the balance of payments — is achieved by requiring registration, supervising it, and penalising failure; it is not achieved by releasing the guarantor from the promise it made.

The second paragraph then closes the escape route. Filing is a step that occurs around formation and performance; it is not the performance itself. A guarantee does not necessarily breach the filing rules merely by being performed.

There is a further provision that a careful reader should keep in view, because it cuts the other way. Article 13 of the same interpretation says that where a contract is void or voidable, a party cannot argue that it is valid on the ground that it has been filed with an administrative department, approved by the approving authority, or used to complete a transfer registration. Filing does not cure invalidity, just as the absence of filing does not create it. The two propositions are symmetrical, and together they describe a regime in which registration is neither a validator nor an invalidator. It is a separate track.

Article 15 supplies the bridge between the rule and the facts, and it is the provision the bonded case really turns on. A court determining the rights and obligations between the parties is not to be bound by the name the parties gave the contract; it is to look at the content. Where the rights and obligations asserted differ from those the content supports, the court determines the actual legal relationship by reference to the background, the purpose, the structure, the performance, and whether the parties fabricated the subject matter of the transaction. In other words, the label on the file does not decide which regime applies; the flow of goods and money does. A scheme that calls itself retail but moves bulk goods offline is examined as what it is.

That provision also explains why the guarantee case is not an invitation to be careless. The court upheld the guarantee because the underlying transaction was real: a genuine loan to a genuine subsidiary, secured by a genuine account arrangement. Had the loan been a device to move money offshore outside the capital account rules, Article 15 would have allowed the court to look through the label, and Article 153 would then have had something to bite on. The filing question never arises in that scenario, because the transaction fails on substance before it reaches procedure.

What the foreign exchange rules actually require

The operative instrument is the State Administration of Foreign Exchange notice of 2014 issuing the Provisions on Foreign Exchange Administration of Cross-Border Guarantees, which took effect on 1 June 2014 and repealed twelve earlier normative documents in the field.

The provisions define a cross-border guarantee as a written undertaking by a guarantor to a creditor, legally binding, promising to perform payment obligations under the guarantee and capable of producing cross-border receipts and payments or cross-border transfers of asset ownership. They then classify guarantees by the place of registration of the parties into three types: onshore guarantee for offshore lending, where the guarantor is registered in China and both debtor and creditor are registered abroad; offshore guarantee for onshore lending, where the guarantor is registered abroad and both debtor and creditor are registered in China; and other forms.

What the 2014 reform did was to move the regime from approval to registration. The regulator’s own summary describes the change in five points: administrative approvals relating to cross-border guarantees were cancelled or substantially simplified, with only the part of a guarantee that creates a new resident claim on or liability to a non-resident after performance brought into the scope of transaction-by-transaction registration; the scope of foreign exchange administration was defined to cover all types of cross-border guarantee that meet the formal legal requirements, are performed by payment, and may materially affect the balance of payments; onshore and foreign-invested enterprises were placed on the same footing in offshore-guarantee-for-onshore-lending, allowed to sign on their own and to perform within one times net assets subject to the applicable limits; supervision was shifted towards off-site verification and foreign exchange inspection; and twelve normative documents were repealed to improve transparency. A separate question-and-answer published by the regulator confirms that approval of guarantee performance was cancelled, and that banks may handle performance of a foreign guarantee themselves.

The phrase the regulator uses for the resulting model is that registration became the principal method of administration. That is the phrase a legal team should keep, because it is more accurate than the shorthand “registration first, approval second” and because it is the regulator’s own characterisation. Registration is the ordinary route; approval survives only in the residual situations the rules still carve out.

Two features of the regime matter for the validity question. First, the obligation is imposed on the guarantor and on the bank that processes the business, not on the foreign creditor. Second, the consequence of non-compliance is administrative: penalties under the foreign exchange regulations, and restrictions on the guarantor’s foreign exchange business. Neither feature is a statement about whether the promise is enforceable between the parties.

The two main categories behave differently, and the difference affects how much diligence a foreign party should do. In an onshore guarantee for offshore lending, the guarantor is Chinese, the bank is Chinese, and the registration is a domestic step between them. A foreign lender can ask for the registration receipt, and should, but it has no independent means of producing one. In an offshore guarantee for onshore lending, the guarantor is foreign and the borrower is a Chinese subsidiary; the regime permits the parties to sign on their own and to perform within one times the borrower’s net assets, subject to the applicable limits, and the registration is handled on the onshore side. The practical point is that the party best placed to know whether a filing happened is rarely the party that will be arguing about validity years later. That asymmetry is precisely why the court’s reasoning lands where it does: the risk of the lapse sits with the party that controls it.

The three-step test, and the other direction

The commentary that accompanied the two cases reduced the reasoning to three steps, and the formulation is worth adopting because it is both accurate and usable.

The first step is to ask whether the provision in question is a mandatory provision of a law or administrative regulation at all. This is a question about the source and character of the rule, not about its importance. A circular issued by a ministry is not a law or administrative regulation for the purposes of Article 153; a provision addressed to the regulator’s own procedures is not a prohibition on the parties.

The second step is to ask whether the rule sets the procedure as a condition of the contract’s effectiveness. Some rules do, expressly. Where a law or administrative regulation provides that a contract takes effect only upon approval, approval is a condition, and until it is obtained the contract is not effective — a different outcome from void, with different consequences, which Article 12 of the interpretation addresses by giving the other party a claim to compel the party under the duty to apply. Where the rule says that parties shall register, or shall file, without saying that effectiveness depends on it, the second step is not satisfied.

The third step is to ask what the rule is for and what the transaction actually does. This is where Article 16’s five situations operate, and where the two cases diverge. A filing regime for balance-of-payments monitoring is served by the filing and by penalties for its absence. A customs supervision regime for bonded imports is served by keeping bonded goods inside the supervision chain; a scheme that extracts them into ordinary domestic sale through fabricated orders does not merely miss a filing, it defeats the regime.

The value of stating the test this way is that it prevents two opposite errors. One is to treat every procedural requirement as a validity requirement, which would make a late filing fatal to a guarantee that everyone intended to be binding. The other is to treat every procedural requirement as merely regulatory, which would let a scheme that exists only to circumvent supervision hide behind the word “formality”.

Applying the test to a guarantee that exceeds the permitted amount is a useful exercise, because it is the most common fact pattern and the least clearly resolved. The amount limit is a prudential rule addressed to the guarantor’s balance sheet and to the regulator’s view of systemic exposure. It is not addressed to the creditor’s conduct, and the creditor is in no position to measure the guarantor’s net assets. On the reasoning of the guarantee case, exceeding the limit should produce an administrative consequence rather than invalidity. But the analysis does not stop there: if the excess is not a computational slip but a structure designed to move value offshore without a channel, Article 15 permits the court to look at the structure, and the outcome may differ. The line is between a guarantee that was too large and a guarantee that was a vehicle.

The test also runs the other way, and because the guarantee case is usually cited for the proposition that procedure does not matter, it is worth being explicit about the cases where it does. Three boundaries are worth marking.

The first is the difference between approval and registration. Approval is a decision by an authority that the transaction may proceed; it is the classic case in which the legislature can and sometimes does make effectiveness conditional. Registration and filing are records of a transaction that has already been decided. The revised foreign exchange regime moved cross-border guarantees from the first category to the second, which is precisely why the guarantee survived.

The second is the treatment of a contract that is not yet effective. Where approval is required and not obtained, the contract is not void; it is ineffective, and the party under the duty to seek approval can be ordered to perform that duty, with damages available if it refuses or delays. That is Article 12 of the interpretation, and it produces a very different negotiating posture from invalidity: the counterparty’s remedy is to push the transaction through, not to walk away.

The third is the rule that filing cannot rescue a bad contract. Article 13 exists because parties routinely assume that an official stamp cures a substantive defect. It does not. A contract that is void because its subject matter is prohibited does not become valid because someone filed it. The guarantee case and the bonded e-commerce case are consistent on this point: in the first, the filing was missing and the contract was valid; in the second, whatever filings existed did not save a transaction whose substance was prohibited.

For a legal team, the practical translation is a single question to ask of every procedural requirement in the deal: does the instrument say that effectiveness depends on it? If yes, put it in the conditions precedent and give someone a specific obligation to obtain it. If no, put it in the completion covenant with a deadline and a named owner, and stop treating it as a validity issue.

Why the bonded e-commerce case failed

The bonded case is worth setting out in some detail, because the facts are the reason it lost.

Cross-border e-commerce retail import in the bonded model works through a chain: goods enter a bonded supervision zone in bulk, and are then sold to individual consumers against real orders, with the platform and the logistics provider transmitting the order, payment and logistics data. The regime’s tax and licensing treatment depends on the sales being retail sales to individuals. The supervision is what keeps the goods inside the chain until they reach a consumer.

In the case, the operator fabricated individual consumer orders and used the bonded retail channel to import bulk infant formula in small parcels, then sold it offline in volume. The court’s characterisation was that the transaction’s substance was the use of a trade facilitation arrangement to circumvent customs supervision. It held the sale contracts void for breach of the Customs Law and the related customs supervision provisions, and referred the leads to the administrative enforcement authority.

Three points about that reasoning are worth extracting.

First, the defect was not documentary. The orders existed; they were simply not real. A regime that depends on the authenticity of the underlying retail transaction is defeated by fabricated retail transactions even where every field in every form is filled in. This is the mirror image of the guarantee case, where the missing document did not defeat a real transaction.

Second, the consequence of invalidity is not limited to the contract. Under Article 157 of the Civil Code, property obtained under a void act is returned, or where return is impossible or unnecessary, compensation is paid at the value; the party at fault compensates the other’s loss, and where both are at fault each bears its own share. In a scheme of this kind, the parties are typically both at fault, and the restitution calculation is unattractive for everyone. The referral of leads to the enforcement authority adds a separate exposure that no contract clause can allocate away.

Third, the case shows where the line sits for structured trade. Using a facilitation regime for the purpose for which it was designed is protected. Using it because it is the cheapest route to a result the regime was not designed to permit is not. The distinction is about the substance of the flow of goods and money, which is why the case file, and not the contract, tends to decide these disputes.

The mechanics of that regime are worth knowing because they define what “real” means. The bonded retail import channel relies on three data streams — order, payment and logistics — transmitted by the platform and the logistics provider, matched against a customs declaration for the parcel. The regime’s treatment of the goods as retail sales to individuals depends on those three streams describing the same small transaction. Fabricating orders does not create a documentary gap that can be filled later; it corrupts the data on which the whole treatment rests. That is why the court could characterise the substance so directly, and why a defence built on the completeness of the paperwork had nowhere to go.

What to do before and after signing

For a company that gives or takes cross-border security, the following is the working list.

Read the effectiveness clause against the regulation. Identify every procedural requirement in the transaction and classify it: is it stated to be a condition of effectiveness, or is it an obligation to complete a step? The classification belongs in the contract’s conditions precedent section, not in a schedule of undertakings.

Give the filing a deadline and an owner. Registration is a guarantor-side obligation, and in practice it is performed by the bank. Name the responsible entity, name the bank, name the deadline, and require evidence of completion to be delivered to the creditor. A creditor that receives the registration receipt at closing has a much better position than one that discovers the gap during enforcement.

Do not make the filing a condition of drawdown unless you mean it. A creditor that insists on prior registration as a condition precedent gets certainty at closing and a delay on every subsequent drawdown under a revolving facility. A creditor that accepts a covenant gets speed and takes the risk of a lapse it cannot easily detect. Both are legitimate; the mistake is to draft one and expect the other.

Keep the authenticity file. For any transaction that uses a facilitation regime — bonded imports, cross-border e-commerce, processing trade — the documents that matter are the ones that show the underlying flow was real. Contracts, orders, payments, logistics and inventory records should be retained as a single chain, because that chain is what a court will examine if the structure is challenged.

Watch the four failure points. In practice, guarantees come under attack on four grounds: the amount exceeds what the regime permits; the use of funds changes after registration; the registration is not completed within the required period; and the guarantor’s internal approvals are missing. The first three are administrative and, on the reasoning above, do not by themselves void the contract. The fourth is a corporate authority problem, not a foreign exchange problem, and it is the one that most often succeeds.

Separate the penalty from the promise. Where a filing was late, the right response is to regularise it and to assess the administrative exposure, not to assume the guarantee has failed. Conversely, where a filing was properly made, do not assume the guarantee is safe: the corporate approval, the authority of the signatory, and the substance of the underlying debt all remain live issues.

Treat a challenge to validity as a factual enquiry. The three-step test is applied to facts. A party arguing that a guarantee is void because of a foreign exchange irregularity will need to show that the rule was intended to make the promise unenforceable, and that outcome is hard to reach where the creditor had no role in the lapse. A party arguing that a structured trade is void will need to show what the flow actually was, and that enquiry is documentary.

Build the security package around a real flow, and check the corporate authority behind each leg. The guarantee case involved more than a promise: it involved an account opened with an onshore bank under a supervision arrangement, funds paid into it, and a claim to priority over those funds. The court upheld the guarantee and confirmed the pledge. That combination is the lesson for structuring: a guarantee that depends on a filing for its force is fragile, while a security package that combines a contractual promise with a possessory or account-based interest gives the creditor something it can enforce without first winning an argument about administrative procedure. The second leg, however, is where most packages fail — not on foreign exchange, but on authority. A company providing security for another’s debt needs the resolution the company law requires, taken by the right body, minuted, and available in the data room. A creditor that holds a guarantee signed by a manager without the resolution has a document, not a security.

Outlook

The direction is towards more registration and less approval, and towards more judicial willingness to separate the two. The 2014 reform moved cross-border guarantees onto the registration track. The interpretation of the contract part of the Civil Code gave courts a structured way to ask what a mandatory provision is for. And the published cases give the lower courts a pair of anchors at each end of the spectrum.

For foreign-invested companies, the practical consequence is a shift in where risk sits. A company that used to worry about whether a missing approval would void its security now worries about whether its filing was timely, whether its internal approvals were complete, and whether the underlying flow can be documented. Those are all things a legal and treasury function can control before closing. The risk that remains is the one the bonded case illustrates: a structure that works only because a supervisory regime is being used against its purpose will not be saved by careful paperwork, and the exposure that follows is not confined to the contract.

One caution belongs at the end, because it is where clients most often over-read a favourable judgment. A holding that the guarantee is valid is not a holding that the money can be paid. Validity is a question between the parties; payment across the border is a question between the guarantor, its bank and the regulator, and it still requires the transaction to satisfy the foreign exchange rules applicable at the time of performance. A creditor that wins on validity and then finds that performance cannot be processed has won a judgment it cannot convert. The two questions should be planned together: the clause, the filing and the payment channel are one design, and the case law on validity settles only the first of them.

Sources

  • 法治日报·以司法裁判确定性护航自贸试验区制度创新(2026-11-11,含长城控股公司案与网购保税商品合同纠纷案):https://epaper.legaldaily.com.cn/fzrb/content/20261111/Articel11004GN.htm
  • 最高人民法院·《最高人民法院关于适用〈中华人民共和国民法典〉合同编通则若干问题的解释》(法释〔2023〕13号,2023-12-05 施行,含第十二条、第十三条、第十六条):https://www.court.gov.cn/zixun/xiangqing/419382.html
  • 国家外汇管理局·关于发布《跨境担保外汇管理规定》的通知(汇发〔2014〕29 号,2014-06-01 施行):https://www.safe.gov.cn/hubei/2014/0519/51.html
  • 国家外汇管理局·《跨境担保外汇管理规定》中关于内保外贷的主要管理内容(以登记为主要管理方式、取消履约核准):http://m.safe.gov.cn/big5/big5/www.safe.gov.cn/safe/2014/0714/2503.html
  • 中国人大网·《中华人民共和国民法典》全文(第一百五十三条,2020 年公报版):http://www.npc.gov.cn/wxzlhgb/c27214/gb2020/202006/P020230313538731037747.pdf
  • 中国审判·最高法发布人民法院服务保障自由贸易试验区建设典型案例(含中德西拉子公司担保合同纠纷案典型意义):http://www.chinatrial.net.cn/news/29764.html

跨境担保的效力与外汇登记:把两起公开案例放在一起读

两起案件,一个问题

2026 年 11 月,法治日报刊发长文,讨论法院应当如何在自由贸易试验区的新型交易中划出“合法创新”与“规避监管”的界线。文章的载体,是最高人民法院发布的服务保障自由贸易试验区建设典型案例中的两起案件。一起支持了一份被以外汇为由攻击的跨境担保;另一起否定了一组建立在虚构消费者订单之上的网购保税进口买卖合同。

分开读,两起案件各是一条熟悉规则的干净示例。放在一起读,它们回答的是外资企业真正会问的那个问题:监管程序没有走,合同是不是就死了?

答案是:取决于这个程序是用来做什么的,以及这笔交易实际上做了什么。这句话听起来像法律人的推托,其实不是。它是一套三步检验,而两起案件恰好落在这套检验的两端。

担保案对资金与法务团队最重要,因为跨境担保是日常业务。中国公司为境外子公司的借款提供支持;外国母公司为中国子公司的借款提供支持;离岸控股公司为境内项目提供担保。外汇管理规定要求这些安排办理登记,个别情形还要求核准。交易先交割,手续后补,几年之后对手方主张:登记迟了,担保从未生效。担保案说这个主张不成立,而且它是通过解释外汇管理规定得出这个结论的,不是通过权衡公平。

随案刊发的评论给出的是一套方法,不是一个结论。它没有说外汇管理规定不重要,也没有说程序瑕疵总能补正。它说的是:法院应当先认清这条规范是什么,再问该规范有没有把这道手续设定为生效条件,然后拿规范的目的去对照交易实际是怎么跑的。用在担保上,这套方法得出一个答案;用在建立在虚构订单之上的保税进口结构上,它得出相反的答案。两起案件之所以被放在一起发布,读其中任何一起都是误读,原因就在这里。

网购保税案是另一端的配重。经营者利用一套为个人消费者设计的便利化制度——凭真实个人订单保税进口——把批量奶粉化整为零运进来,再线下销售。法院认定买卖合同因违反海关法及有关海关监管规定而无效,并将违法线索移送行政执法部门。那起案件里的单据很可能是齐的。失败的不是手续,是交易本身。

第一百五十三条到底在问什么

起点是《民法典》第一百五十三条:违反法律、行政法规的强制性规定的民事法律行为无效,但是该强制性规定不导致该民事法律行为无效的除外;违背公序良俗的民事法律行为无效。

第一款的但书才是全部战场,而它刻意留有弹性。最高人民法院关于适用民法典合同编通则若干问题的解释第十六条把这个但书填实:合同违反法律、行政法规的强制性规定,有下列情形之一,由行为人承担行政责任或者刑事责任能够实现该强制性规定的立法目的的,人民法院可以认定合同不因违反强制性规定无效——强制性规定旨在维护社会公共秩序,但合同实际履行对社会公共秩序造成的影响显著轻微,认定无效将导致处理结果有失公平公正;强制性规定旨在维护税收、土地出让金等国家利益或者其他民事主体的合法利益而非合同当事人的民事权益,认定合同有效不会影响该规范目的的实现;强制性规定旨在要求一方加强风险控制、内部管理,对方无能力或者无义务审查合同是否违反该规定,认定无效将使其承担不利后果;一方订立合同时违反强制性规定,但订立后已具备补正条件却违背诚信原则不予补正;以及法律、司法解释规定的其他情形。

第十六条还有第二款,容易被略过,却常常决定结果:法律、行政法规的强制性规定旨在规制合同订立后的履行行为的,当事人以合同违反该规定为由请求认定无效的,人民法院不予支持;但合同履行必然导致违反该规定或者法律、司法解释另有规定的除外。

对跨境担保而言,五项情形里有两项在起作用。外汇登记是一套施加于担保人、经由银行落实的风险控制与统计制度;境外债权人既没有实际能力核实中国担保人是否办理登记,也没有法定的核查义务。因此认定担保无效,等于把担保人自身的行政疏漏后果转嫁给债权人,这正是第三项情形。而该制度的目的——监测跨境或有负债与国际收支——通过要求登记、实施监督、对不登记予以处罚即可实现,并不需要通过免除担保人已经作出的承诺来实现。

第二款随后堵住了另一条退路。登记是围绕合同订立与履行发生的一个步骤,它本身不是履行。担保不会仅仅因为被履行就必然违反登记规定。

还有一条规定,细心的读者应当同时记住,因为它指向相反方向。同一部解释第十三条规定:合同存在无效或者可撤销情形,当事人以该合同已在有关行政管理部门办理备案、已经批准机关批准或者已依据该合同办理财产权利的变更登记、移转登记等为由主张合同有效的,人民法院不予支持。备案不能治愈无效,正如没有备案不能制造无效。两个命题是对称的,合起来描述的是一套制度:登记既不赋予效力,也不否定效力,它走的是另一条轨道。

第十五条则在规则与事实之间搭了一座桥,网购保税案真正落在这条上。人民法院认定当事人之间的权利义务关系,不应当拘泥于合同使用的名称,而应当根据合同约定的内容;当事人主张的权利义务关系与根据内容认定的不一致的,应当结合缔约背景、交易目的、交易结构、履行行为以及当事人是否存在虚构交易标的等事实,认定实际民事法律关系。换句话说,卷宗上贴的标签不决定适用哪套制度,货物与资金的流向才决定。一个自称零售、却把批量货物搬到线下销售的结构,就按它的实际样子被审查。

这一条也说明,担保案并不是一张“可以粗心”的通行证。法院之所以支持那份担保,是因为底层交易是真的:一笔真实的、对真实子公司的借款,配上一套真实的账户安排。假如那笔借款只是为了绕开资本项目规则把资金挪到境外的装置,第十五条就允许法院看穿标签,第一百五十三条也才有可咬的东西。在那个剧本里,登记问题根本不会出现,因为交易在走到程序之前,就已经在实质上倒下了。

外汇管理规定到底要求什么

现行依据是国家外汇管理局 2014 年发布《跨境担保外汇管理规定》的通知,自 2014 年 6 月 1 日起实施,同时废止该领域十二项规范性文件。

规定把跨境担保界定为:担保人向债权人书面作出的、具有法律约束力、承诺按照担保合同约定履行相关付款义务并可能产生资金跨境收付或者资产所有权跨境转移等国际收支交易的担保行为。随后按担保当事各方的注册地分为三类:担保人注册地在境内、债务人和债权人注册地均在境外的内保外贷;担保人注册地在境外、债务人和债权人注册地均在境内的外保内贷;以及其他形式跨境担保。

2014 年这次改革做的事情,是把制度从核准搬到登记。监管机关自己的概括有五点:取消或者大幅度简化与跨境担保相关的行政审批,只将“担保履约后新增居民对非居民负债或者债权”的部分跨境担保纳入逐笔登记范围;合理界定外汇管理范围,将符合法律形式要求、以付款为担保履约方式、对国际收支可能产生重要影响的所有类型跨境担保纳入调整;在外保内贷领域实现中外资企业统一待遇,符合相关限制性条件的,允许企业自行签约,并在净资产一倍以内办理担保履约;监管理念转向非现场核查与外汇检查;清理整合法规,废止十二项规范性文件。监管机关另行发布的问答进一步确认,担保履约核准已取消,银行可以自行办理对外担保履约。

监管机关对这套模式的用语是:以登记为主要管理方式。这个表述值得法务团队记住,因为它比“登记为主、核准为辅”的简称更准确,而且它是监管机关自己的定性。登记是常规路径,核准只在规则仍然保留的少数情形中存续。

这套制度有两个特征与效力问题直接相关。第一,义务落在担保人和经办银行身上,不落在境外债权人身上。第二,不遵守的后果是行政性的:依外汇管理条例处罚,并可能限制担保人的外汇业务。这两个特征都不是关于“当事人之间这项承诺是否可执行”的陈述。

两大类别的行为方式不同,这个差异决定境外一方该做多少尽职调查。内保外贷里,担保人是中国企业,银行是中国银行,登记是它们之间的境内步骤。境外贷款人可以索取登记凭证,也应当索取,但它没有独立产出凭证的手段。外保内贷里,担保人在境外,借款人是中国子公司;制度允许当事人自行签约,并在借款人净资产一倍以内办理担保履约,登记在境内一侧完成。实务上的要点是:最清楚有没有办成登记的,往往不是多年以后要就效力争论的那一方。这种不对称,恰恰解释了法院的推理为什么落在它落下的地方:疏漏的风险,配置在能够控制它的一方身上。

三步检验,以及相反的方向

随案刊发的评论把裁判思路归纳为三步,这个表述值得沿用,因为它既准确又可操作。

第一步,问该规定到底是不是法律、行政法规的强制性规定。这是关于规范来源与性质的问题,不是关于规范重要性的问题。部门发的通知,对第一百五十三条而言不是法律或者行政法规;写给监管机关自身工作程序的规定,也不是对当事人的禁止。

第二步,问该规范是否把这道手续设定为合同生效条件。有些规范确实这样设定,而且是明示的。法律、行政法规规定合同经批准才生效的,批准就是条件,取得之前合同不生效——这与无效是不同结果,后果也不同,司法解释第十二条为此给另一方一项请求履行报批义务的请求权。规范只说当事人应当登记、应当备案,而没有说效力取决于此的,第二步不成立。

第三步,问规范的目的何在,交易实际上做了什么。第十六条的五项情形在这一步运作,也是两起案件分道的地方。为国际收支监测而设的登记制度,靠登记本身和对不登记的处罚即可实现。为保税进口而设的海关监管制度,靠把保税货物留在监管链条内实现;一个用虚构订单把货物抽出链条、转入普通境内销售的结构,不是漏了一道手续,而是使制度落空。

把检验这样写出来,价值在于避免两个相反的错误。一个错误是把一切程序要求都当成效力要求,那会让一次迟延的登记杀死一份所有人都想让它有约束力的担保。另一个错误是把一切程序要求都当成纯监管事项,那会让一个只为绕开监管而存在的结构躲在“手续”两个字后面。

拿一笔超过允许金额的担保来跑一遍这套检验,是很有用的练习,因为它既是最常见的事实形态,也是结论最不清晰的一种。金额上限是一条审慎规则,指向担保人的资产负债表与监管机关对系统性敞口的判断,它不指向债权人的行为,而债权人也无从测量担保人的净资产。按担保案的推理,超额应当产生行政后果,而不是无效。但分析不在这里停下:如果超额不是计算失误,而是为在没有通道的情况下把价值挪到境外而设计的结构,第十五条允许法院看结构,结论就可能不同。界线在于:这是一笔金额过大的担保,还是一辆载具。

这套检验也朝相反方向跑。担保案常被引来论证“程序无所谓”,所以更有必要说清楚哪些情形程序确实有所谓。三条边界值得标出。

第一,批准与登记之别。批准是行政机关对交易可以进行的决定,是立法者可以、有时也确实会把效力系于其上的典型情形。登记与备案,是对已经决定的交易作出的记录。修订后的外汇制度把跨境担保从第一类挪到了第二类,这正是担保活下来的原因。

第二,未生效合同的处理。需要批准而未获批准的,合同不是无效,而是未生效;负有报批义务的一方可以被判令履行该义务,拒绝或者迟延时可以判赔。这是司法解释第十二条,它产生的谈判姿态与无效完全不同:对手方的救济是推动交易完成,而不是抽身走人。

第三,备案救不了坏合同。第十三条之所以存在,是因为当事人经常以为盖了章就治好了实质瑕疵。治不好。标的被禁止的合同,不会因为有人备了案就变成有效。担保案与网购保税案在这一点上是一致的:前者的手续缺失而合同有效,后者纵有手续也救不回一个实质被禁止的交易。

对法务团队而言,实务上的转化就是一个问题,对交易里每一项程序要求都要问:文件有没有说效力取决于它?说有,就把它放进先决条件,并指定具体的人承担取得义务。说没有,就把它放进交割承诺,配上期限与责任人,并且不要再把它当成效力问题。

网购保税案为什么输了

网购保税案值得展开,因为事实正是它输掉的原因。

保税模式下的跨境电商零售进口走一条链条:货物以批量方式进入保税监管区域,再凭真实订单销售给个人消费者,由平台与物流企业传输订单、支付、物流三单数据。这套制度在税收与准入上的待遇,建立在“向个人零售”这一前提上。监管的作用,是让货物在到达消费者之前始终留在链条里。

该案中,经营者虚构个人消费者订单,借保税零售渠道把批量奶粉化整为零进口,再线下成批销售。法院的定性是:交易实质是借贸易便利化安排规避海关监管。据此认定买卖合同违反海关法及有关海关监管规定而无效,并将违法线索移送行政执法部门。

这套推理有三点值得提取。

第一,瑕疵不在单据。订单是有的,只是不真。一套依赖底层零售交易真实性的制度,会被虚假的零售交易击穿,即使每一张单据的每一个字段都填满了。这与担保案恰好互为镜像:那里缺一份文件,却没有击穿一笔真实的交易。

第二,无效的后果不限于合同。依《民法典》第一百五十七条,民事法律行为无效后,行为人因该行为取得的财产应当返还;不能返还或者没有必要返还的,应当折价补偿;有过错的一方应当赔偿对方由此所受到的损失,各方都有过错的,应当各自承担相应的责任。在这类结构中,双方通常都有过错,返还与折价的计算对谁都不好看。线索被移送行政执法部门,又叠加了一层任何合同条款都无法分配的敞口。

第三,这起案件标出了结构化贸易的界线。把便利化制度用于它被设计出来的用途,受保护;因为它是达到某种制度本不打算允许的结果的最便宜路径而使用它,不受保护。这条界线关乎货物与资金流动的实质,这也是为什么这类争议往往由卷宗而不是由合同来决定胜负。

这套制度的运行机理值得了解,因为它定义了“真实”是什么意思。保税零售进口渠道依赖三股数据——订单、支付、物流——由平台与物流企业传输,并与包裹的报关单相互匹配。制度把货物当作向个人零售来对待,前提就是这三股数据描述的是同一笔小额交易。虚构订单不是留下一个日后可以补上的单据缺口,而是污染了整套待遇所依赖的数据本身。这就是法院能够如此直接地给交易定性的原因,也是一份以“单据齐备”为支点的抗辩无处落脚的原因。

签前签后各做什么

对提供或者接受跨境担保的企业,以下是可用的清单。

把效力条款对着规范读一遍。 找出交易里的每一项程序要求并分类:它是被写成生效条件,还是被写成完成某一步的义务?分类结果应当写进合同的先决条件部分,而不是写进承诺事项附表。

给登记一个期限和一个责任人。 登记是担保人一侧的义务,实务中由银行办理。写明责任主体、经办银行、期限,并要求把完成的凭证交付债权人。交割时拿到登记凭证的债权人,比在执行阶段才发现缺口的人位置好得多。

不要随口把登记设成提款条件。 坚持把事先登记作为先决条件的债权人,换来的是交割的确定性,以及循环额度下每一次提款的延误。接受承诺条款的债权人,换来的是速度,并承担自己不易察觉的疏漏风险。两种选择都正当;错误在于写了一种,却期待另一种。

把真实性卷宗留住。 凡是使用便利化制度的交易——保税进口、跨境电商、加工贸易——真正要紧的文件,是能证明底层流真实的那些。合同、订单、支付、物流、库存记录应当作为一条链保存,因为一旦结构被质疑,法院看的就是这条链。

盯住四个失效点。 实务中担保被攻击通常有四个理由:金额超出制度允许的范围;登记之后资金用途发生改变;未在规定期限内办理登记;担保人内部决议缺失。前三项是行政性的,按前面的推理,本身不使合同无效。第四项是公司授权问题,不是外汇问题,也是最常被攻破的那一项。

把处罚与承诺分开。 登记迟了的正确反应,是补办并评估行政敞口,而不是假定担保已经失效。反过来,登记办得齐的,也不要假定担保就安全:公司决议、签署人权限、底层债务的实质,都仍然是活的问题。

把效力之争当成事实调查来做。 三步检验是适用于事实的。主张担保因外汇瑕疵而无效的一方,需要说明该规范的意图是让承诺不可执行,而在债权人于疏漏中毫无角色的情形下,这个结论很难达到。主张结构化贸易无效的一方,需要说明资金与货物的实际流向,而这项调查是单据性的。

围绕真实的流搭建担保组合,并逐条检查每一段背后的公司授权。 担保案里不只有一项承诺:还有在境内银行依监管安排开立的账户、划入其中的资金,以及对这笔资金主张优先受偿。法院支持了担保,也确认了质权。这一组合正是结构设计上的教训:一份要靠登记才有力气的担保是脆弱的,而把合同承诺与占有性权益或者账户权益组合起来的担保包,能让债权人不必先赢下一场关于行政程序的争论就能执行到东西。然而第二段恰恰是多数担保包失手的地方,而且失手点不在外汇,在授权。为他人债务提供担保的公司,需要公司法所要求的决议,由有权机构作出,记录在案,并能放进资料室。债权人手上只有一份经理签署、背后没有决议的担保,那是文件,不是担保。

展望

方向是登记更多、核准更少,以及法院更愿意把两者分开。2014 年的改革把跨境担保搬上了登记轨道。合同编通则解释给了法院一套结构化的方法去追问强制性规定的目的。公开的案例则在光谱两端各给了下级法院一个锚点。

对外资企业而言,实际后果是风险位置的移动。过去担心的是“缺一个核准会不会让担保无效”,现在要担心的是“登记是否及时、内部决议是否齐备、底层流能否被单据还原”。这些都是法务与资金部门在交割前可以控制的事情。剩下的那部分风险,正是网购保税案所展示的:一个只有把监管制度用在其目的反面才成立的结构,靠细致的手续救不回来,而随之而来的敞口并不限于合同。

最后要留一句提醒,因为客户最常在这里过度解读一份有利判决。认定担保有效,不等于认定这笔钱付得出去。效力是当事人之间的问题;跨境付款是担保人、其经办银行与监管机关之间的问题,而且仍然要求交易满足履约时适用的外汇管理规定。一个在效力上赢了、随后发现履约办不下来的债权人,赢到的是一份换不成钱的判决。两个问题应当一起设计:条款、登记与付款通道是一套方案,而关于效力的判例只解决了其中第一个。

来源

  • 法治日报·以司法裁判确定性护航自贸试验区制度创新(2026-11-11,含长城控股公司案与网购保税商品合同纠纷案):https://epaper.legaldaily.com.cn/fzrb/content/20261111/Articel11004GN.htm
  • 最高人民法院·《最高人民法院关于适用〈中华人民共和国民法典〉合同编通则若干问题的解释》(法释〔2023〕13号,2023-12-05 施行,含第十二条、第十三条、第十六条):https://www.court.gov.cn/zixun/xiangqing/419382.html
  • 国家外汇管理局·关于发布《跨境担保外汇管理规定》的通知(汇发〔2014〕29 号,2014-06-01 施行):https://www.safe.gov.cn/hubei/2014/0519/51.html
  • 国家外汇管理局·《跨境担保外汇管理规定》中关于内保外贷的主要管理内容(以登记为主要管理方式、取消履约核准):http://m.safe.gov.cn/big5/big5/www.safe.gov.cn/safe/2014/0714/2503.html
  • 中国人大网·《中华人民共和国民法典》全文(第一百五十三条,2020 年公报版):http://www.npc.gov.cn/wxzlhgb/c27214/gb2020/202006/P020230313538731037747.pdf
  • 中国审判·最高法发布人民法院服务保障自由贸易试验区建设典型案例(含中德西拉子公司担保合同纠纷案典型意义):http://www.chinatrial.net.cn/news/29764.html

Arbitration or Litigation for Foreign-Invested Company Governance Disputes: A Shanghai Forum Guide

  • Arbitration Clause Design decides the case before the dispute exists, because the Arbitration Law requires three elements and treats an unclear reference to the institution as a defect that voids the clause.
    仲裁条款的设计在争议发生之前就决定了案件走向:仲裁法要求三项内容齐备,机构约定不明将被认定为无效缺陷。
  • Exclusive Jurisdiction under Article 279 of the Civil Procedure Law tells a Chinese court that it must hear a case; it does not tell the parties that they must litigate, and a valid clause still keeps the court out.
    民事诉讼法第二百七十九条的专属管辖是告诉中国法院必须受理,而不是告诉当事人必须诉讼;有效的仲裁条款仍将法院挡在门外。
  • Shareholder Deadlock has a contractual half and a corporate half: the exit price and the buy-out can be arbitrated, while the judicial dissolution that ends the company’s legal existence sits with the court.
    股东僵局一半是合同问题、一半是组织法问题:退出价格与股权收购可以仲裁,而终结公司法人资格的司法解散属于法院。
  • Awards travel under the New York Convention, so the seat and the institution written into the clause matter more than the city where hearings happen to be held.
    裁决依《纽约公约》流转,因此条款写明的仲裁地与机构,比开庭所在城市更重要。
  • Shanghai now offers a full chain: a local regulation on building an international commercial arbitration centre, a 2024 institutional rulebook, and a dedicated commercial court that hears applications to set aside or enforce awards.
    上海已形成完整链条:一部建设国际商事仲裁中心的地方性法规、一套 2024 年机构仲裁规则,以及集中审理撤销与执行申请的专门商事法庭。
  • The three-month window to apply to set aside an award runs from receipt of the award, and it is shorter than the period it replaced.
    申请撤销裁决的三个月期限自收到裁决书起算,比此前适用的期限更短。

Arbitration or Litigation for Foreign-Invested Company Governance Disputes: A Shanghai Forum Guide | 外资公司治理争议走仲裁还是诉讼:上海的程序选择指引

The dispute this piece is about

A foreign investor and a Chinese partner set up a joint venture. A few years later the relationship has run into one of four familiar walls: the Chinese partner blocks a capital increase the foreign side needs in order to keep its ratio; the Chinese partner proposes to transfer its equity to a third party and the foreign side says its pre-emptive right has been ignored; the board passes a resolution on a related-party transaction that the foreign-nominated directors say was never properly convened; or the two sides cannot agree on anything and the company stops making decisions altogether.

Each of these is a company law problem. Each is also a contract problem, because the joint venture contract and the shareholders’ agreement will have something to say about all four. The question here is narrow and practical: where the parties wrote an arbitration clause, which of these disputes can go to arbitration, which must go to a Chinese court, and what should the clause say so that the answer is not decided by the other side’s procedural objection?

The scope is deliberately limited. This is not a survey of the revised Arbitration Law, not a general guide to enforcing foreign awards, and not a note on arbitration of sale-of-goods or construction disputes. It is about governance disputes inside a foreign-invested company — the capital increase, the pre-emptive right, the resolution, the deadlock — and about the forum that hears them.

Two framing points before the detail.

First, the revised Arbitration Law has been in force since 1 March 2026. It was adopted on 12 September 2025, runs to eight chapters and 96 articles, and replaces a statute that had served for three decades. Most of what matters for a governance clause is not new: the three required elements of an arbitration agreement, the separability doctrine, and the rule that a valid clause keeps a court out all predate the revision. What changed matters at the edges, and the edges are exactly where governance clauses sit. The terminology moved from “arbitration commission” to “arbitration institution”. The seat became a statutory concept. Ad hoc arbitration opened a narrow door for foreign-related disputes involving enterprises registered in free trade zones and the Hainan free trade port. The window for applying to set aside an award was shortened to three months.

Second, Shanghai is not a neutral backdrop. It has a local regulation on building an international commercial arbitration centre, in force since 1 December 2023. It has an institution whose current rulebook took effect on 1 January 2024. And it has a dedicated international commercial court, established at the end of December 2024 and hearing cases from 1 January 2025, which concentrates foreign-related commercial trials and applications for review of arbitral awards. For a company with operations in the Yangtze delta, the real choice is rarely “arbitration or litigation” in the abstract. It is which of two Shanghai institutions to name, on what terms, and what happens afterwards if the losing side does not pay.

Three elements, read strictly

Article 27 of the revised law says an arbitration agreement may be a clause in a contract or a separate agreement concluded before or after the dispute arises, and that it shall contain three things: an expression of intent to arbitrate, the matters to be arbitrated, and the chosen arbitration institution. Article 29 then supplies the consequence that gives the three elements their bite: where the matters or the institution are not agreed or are agreed unclearly, the parties may supplement the agreement, and if they cannot, the agreement is invalid.

Read together, those two articles are the single most expensive drafting lesson in this field. A clause that says disputes “shall be resolved by arbitration in Shanghai” names a city, not an institution. A clause that says disputes shall go to “the arbitration commission in Shanghai” is ambiguous, because Shanghai has more than one. A clause that names an institution which does not exist, or names a body that administers mediation rather than arbitration, fails the third element. In each case the failure is not cured by the parties’ obvious intention. Intention is not the test; the institution is.

The revised law softened the vocabulary without softening the requirement. Article 89 defines “arbitration institution” to include arbitration commissions and arbitration institutes established according to law, which closes an old argument about whether a body calling itself a centre or an institute could administer a case. It does not rescue a clause that names nothing. Article 27 also adds a route that did not exist in this form before: where one party asserts an arbitration agreement and the other does not deny it before the first hearing, the tribunal, having noted the point on the record, may treat the agreement as existing. That is a useful safety net for a sloppy clause where the respondent chooses not to fight, and a trap for a respondent that assumes silence preserves its position.

Two further provisions complete the picture. Article 30 restates separability: the arbitration agreement stands independently, and whether the contract was formed, varied, ineffective, terminated, revoked or void does not affect the validity of the agreement already reached. The tribunal may rule on the validity of the contract. This matters in governance disputes more than in ordinary sale contracts, because the usual defence to a governance claim is that the very instrument containing the clause — a capital increase agreement, a share transfer agreement, an amended joint venture contract — was itself invalid or never approved. Separability means that argument is heard by the tribunal, not by a court asked to dismiss the arbitration.

Article 31 governs challenges to the agreement itself: a party may ask the institution or the tribunal to decide, or ask a court to rule, and where one side goes to the institution and the other to a court, the court prevails. The objection must be raised before the first hearing. The practical translation is that forum fights in these cases are short, front-loaded and decided early. There is no long preliminary phase in which to test the waters.

Exclusive jurisdiction is about courts, not about arbitration

Article 279 of the Civil Procedure Law lists three categories of case over which Chinese courts have exclusive jurisdiction. The first covers disputes arising from the establishment, dissolution or liquidation of a legal person or other organisation established within Chinese territory, and from the validity of resolutions made by that legal person or organisation. The second covers disputes over the validity of intellectual property rights granted after examination within Chinese territory. The third covers disputes arising from the performance within Chinese territory of a Chinese-foreign equity joint venture contract, a Chinese-foreign cooperative joint venture contract, and a contract for Chinese-foreign cooperative exploration and development of natural resources.

That first item is the one that lands directly on governance disputes, and it is often read as an answer to the forum question. It is not. Exclusive jurisdiction is a rule addressed to courts: it tells a Chinese court that it must take the case and cannot decline it in favour of a foreign court. Article 5 of the Arbitration Law addresses a different question: where the parties have reached an arbitration agreement and one of them brings a suit, the court shall not accept the case, unless the agreement is invalid or the law provides otherwise.

Those two rules operate on different planes. Exclusive jurisdiction decides the allocation of adjudicatory power among courts, including the question whether a foreign court’s judgment could be recognised. The arbitration agreement decides whether the parties have removed the dispute from adjudication altogether. A valid clause therefore does not offend Article 279; it operates one step earlier. What Article 279 does mean, in practice, is that the residual route — litigation — is not a route the parties can steer to a foreign court by agreement, and that if the clause fails, the case stays in China.

This distinction has a concrete consequence for drafting. A clause that provides for arbitration in Singapore or Hong Kong is not invalid because the underlying contract is a joint venture contract. But the fallback matters. If the clause is later held invalid, the dispute does not go to the chosen foreign forum; it goes to a Chinese court with exclusive jurisdiction. A party that negotiated hard for a foreign seat and accepted a vague institution name has, in effect, negotiated for Chinese litigation.

There is a genuine grey area, and it should be named rather than smoothed over. Where the claim attacks the validity of a shareholders’ resolution as an act of the company — a claim that the resolution never came into existence, or was adopted without the required majority — the subject matter looks like the first item of Article 279, and Chinese courts have tended to treat such organisational claims as belonging to the court. Where the claim is that a shareholder breached the shareholders’ agreement by voting a certain way, the same facts produce a contractual claim that is plainly arbitrable. The two claims can arise from one meeting. The statute does not resolve the boundary in terms, and no clause can eliminate the risk. What a clause can do is allocate the claims expressly, so that the arbitrable ones are pleaded as contract claims and the organisational ones are left where they belong.

One further rule completes the picture, and it is the one that decides how much a foreign forum is worth in a governance fight. Article 280 of the Civil Procedure Law addresses parallel proceedings: where the same dispute is brought before a foreign court by one party and before a Chinese court by the other, a Chinese court that has jurisdiction may accept the case. Where the parties have concluded an exclusive jurisdiction agreement choosing a foreign court, and that agreement does not offend the exclusive jurisdiction rules and does not involve sovereignty, security or public interest, the Chinese court may rule not to accept, or may dismiss a case already accepted. Article 281 then governs a stay application based on the foreign court having accepted the case first, and Article 282 sets out the conditions for declining jurisdiction in favour of a clearly more convenient foreign court — none of which is available where the dispute falls within exclusive jurisdiction.

The consequence for a governance clause is that a foreign court judgment on the dissolution of a Chinese company, or on the validity of its resolutions, has no route into China. Recognition and enforcement of foreign judgments is not governed by a multilateral convention in the way awards are; it rests on treaties and reciprocity. An arbitration award does have a route, and it is the treaty route. That asymmetry, rather than any preference for arbitration as a process, is the strongest practical argument for putting these disputes into a clause.

The four governance scenarios

With that framework, the four scenarios can be separated.

Capital increase. A capital increase is usually documented twice: as a shareholders’ resolution, and as a subscription agreement or an amendment to the joint venture contract. The contractual claim — that a party promised to subscribe, or promised to vote in favour, or promised not to dilute — is arbitrable. The corporate act of increasing registered capital is a registration matter that runs through the market supervision system, and no tribunal can order the registry to act. In practice an arbitral award in this scenario awards damages or specifically requires a party to vote and to execute documents; the registration follows, or it does not, and enforcement of the second step is where these cases become difficult. A well-drafted clause anticipates this by making the voting obligation and the execution obligation explicit, and by providing that a party may sign on behalf of a recalcitrant counterparty only where the law allows it — which, for company law instruments, is a narrow permission.

Pre-emptive right on a share transfer. Where the joint venture contract gives a shareholder a right of first refusal, a transfer to a third party in breach of that right generates a claim that is contractual in form and arbitrable. The remedy is typically an order restraining the transfer, damages, or both. Timing decides everything: an arbitration commenced after the transferee has been registered as shareholder faces a third party whose interests the tribunal cannot dispose of, and the remedy collapses into damages. This is the scenario in which the interval between discovering the breach and filing the request for arbitration is the whole case, and in which interim relief — available under Article 39, with the institution transmitting the application to a court — is worth more than the final award.

Setting aside a board or shareholder resolution. As noted, the organisational claim sits with the court. But the same meeting usually generates contract claims: that directors nominated by one side were not given notice as the shareholders’ agreement requires, that a reserved matter was decided without the consent the agreement demanded, that a related-party transaction breached a contractual approval mechanism. Those claims are arbitrable, and an award finding that the consent mechanism was breached can support a damages claim even where the resolution itself remains on the register. Practitioners sometimes describe this as arbitration achieving indirectly what it cannot achieve directly. That is accurate, and it is worth planning for rather than discovering.

Deadlock and dissolution. This is the scenario where the limits are clearest. A petition for judicial dissolution of the company asks a court to end a legal personality; it is a statutory remedy available to shareholders holding a specified proportion of voting rights, and it is not a claim the parties can put to a tribunal. What arbitration can do is enforce the contractual exit that the shareholders’ agreement should have provided: a put option, a buy-out at a formula price, a right to appoint a third-party valuer, a right to require the other side to purchase at fair value determined by an expert. Companies that have litigated a deadlock to judgment often discover that the judgment dissolves the company and distributes proceeds at liquidation values, which is rarely what either side wanted. The contractual exit is usually the better commercial outcome and is the one an arbitration clause can deliver.

The pattern across all four is the same. Arbitration reaches the money and the promises. It does not reach the register, the legal personality, or the organisational act. Clauses that ignore the second half generate awards that are correct and hard to use.

Drafting the clause

The clause is short, so every word carries weight. Six elements are worth writing out.

The institution, in full. Use the registered name and, where the institution publishes one, the name it accepts as equivalent. Article 89 confirms that arbitration commissions and arbitration institutes both count, but the safer course is the exact name on the institution’s own website, in the language in which the institution is registered, with the Chinese name included where the counterparty is Chinese. Where a clause is bilingual, add a provision that the Chinese text prevails for the purpose of identifying the institution, so that a translation error does not become an invalidity argument.

The seat. Article 81 allows the parties to agree the seat in writing, and provides that, unless they have agreed a different law for the procedure, the seat determines the law applicable to the procedure and the court with supervisory jurisdiction, and that the award is deemed made at the seat. Naming a seat is not the same as naming a venue for hearings. A clause that says hearings will be held in Shanghai but names Singapore as the seat has chosen Singaporean curial law and Singaporean courts for set-aside purposes. That is a legitimate choice, but it must be a deliberate one.

The law governing the contract. Distinct from the seat, and frequently omitted. For a joint venture contract, the substantive law is Chinese law in most cases, and a clause that leaves it open invites an argument that the tribunal must first decide which law applies before it can decide anything else.

Language and the number of arbitrators. Article 42 allows one or three. For a governance dispute with documents in two languages and directors from three jurisdictions, three is the usual choice, and the clause should say whether the language of the arbitration is Chinese, English, or both, and whether documents may be submitted untranslated.

Consolidation and joinder. Governance disputes rarely involve two parties only. A capital increase may involve the company, two shareholders and a new investor; a share transfer may involve a transferee; a deadlock may involve a holding company above the joint venture. Article 27 permits one agreement covering several matters but says nothing about consolidating several arbitrations. Institutional rules fill the gap: the 2024 rulebook of the Shanghai institution provides for consolidated arbitration and for the addition of a party, on conditions the institution administers. A clause that does not address consolidation leaves the parties to run parallel arbitrations with inconsistent outcomes.

The scope of the submission. The most common defect after a misnamed institution is a scope that is either too narrow or too vague. A clause covering “disputes arising under this agreement” may not reach a claim under the joint venture contract, because they are different instruments. A clause covering “all disputes in connection with the company” may be attacked as covering matters that cannot be arbitrated. The workable formula is to list the instruments — the joint venture contract, the shareholders’ agreement, the articles of association to the extent a breach claim is contractual, any equity transfer or capital increase agreement — and then to add a residual clause covering disputes connected with the establishment, operation or termination of the company that the parties are legally entitled to refer to arbitration.

Two smaller points. First, service. Article 41 provides that arbitration documents are served in the manner agreed, failing which the institutional rules apply. A governance dispute frequently involves a party that has stopped answering; a clause that specifies an address for service, and provides that a change of address must be notified, converts a default into a routine step. Second, confidentiality. Article 52 makes arbitration private by default, with public hearings only by agreement and never where state secrets, trade secrets or personal privacy are involved. For a joint venture dispute involving pricing, customer lists and board minutes, the default is an advantage — but it is worth confirming that the institutional rules chosen do not permit disclosure in circumstances the parties would regret.

Shanghai: institution, seat and judicial review

Three layers make up the Shanghai offering, and they are often confused with one another.

The first is the policy and legal layer. Shanghai was among the first locations designated for the national pilot on building international commercial arbitration centres, and in December 2023 a local regulation on advancing that construction took effect. The regulation addresses the governance of arbitration institutions, the opening of the sector, temporary arbitration and the seat, and support policies. It is a local regulation, not a national statute, but it is the reason several institutional features appear in Shanghai before they appear elsewhere.

The second is the institution. The Shanghai institution’s current rulebook took effect on 1 January 2024, replacing the 2014 free trade zone rules. It is a package rather than a single text: a main rulebook, separate rules for aviation and for data disputes, an online arbitration guideline, and a guideline on assisting ad hoc arbitration. The features relevant to governance disputes are consolidation, joinder, interim measures and emergency arbitrator relief, a widened ability to choose arbitrators, and provision for a Hong Kong centre. Its free trade zone arbitration court, the aviation court and the specialised platforms sit under the same institution, and the rules treat a reference to a platform as a reference to the institution — which is a useful antidote to the naming problem described earlier.

The third is the court. The Shanghai International Commercial Court was established at the end of December 2024 and began receiving cases on 1 January 2025. It concentrates, among other matters, first and second instance foreign-related commercial cases, applications for review of arbitral awards, applications to confirm foreign-related mediation agreements, and applications to recognise and enforce foreign court judgments. By the end of August 2026 it had received more than 3,700 foreign-related commercial and arbitration-review cases, with parties from 64 countries and regions. In June 2026 it appointed an international commercial expert committee of 20 members from eight countries and regions.

The significance for a governance clause is that the court which supervises the arbitration is identifiable in advance, and it is a specialist court rather than a general civil division. Set-aside applications under Article 71 go to the intermediate court where the institution is located; for a Shanghai-seated arbitration administered by a Shanghai institution, that means the specialist court. Article 72 requires the application within three months of receiving the award — shorter than the six months that applied before the revision — and Article 73 requires the court to rule within two months of accepting the application. A party contemplating a challenge should read those two numbers together: the window to file is short, and the period of uncertainty afterwards is also short.

Article 83 sets out the grounds on which a foreign-related award may be set aside, and they are narrower than the domestic list: no arbitration agreement; the respondent was not notified of the appointment of an arbitrator or of the proceedings, or was unable to present its case for reasons not attributable to it; the composition of the tribunal or the procedure was inconsistent with the rules; or the matters decided fell outside the scope of the agreement or the institution had no power to decide them. Public policy is a separate ground. Notably, the domestic ground of “the evidence on which the award was based was forged” does not appear in the foreign-related list. For a foreign investor, that is a meaningful difference: the review is closer to the model law’s procedural checklist than to a re-examination of the merits.

After the award, and a checklist

Article 70 provides that an award takes effect on the date it is made. Article 75 provides that a party may apply to a people’s court for enforcement, and that the court shall enforce. In practice three steps follow.

The first is enforcement. Where the assets are in China, the application goes to the competent intermediate court. Where the respondent or its property is outside China, Article 85 permits a direct application to a competent foreign court, and Article 88 governs the recognition and enforcement of awards made outside China, on the basis of treaties or reciprocity — which in practice means the New York Convention for the large majority of China’s trading partners.

The second is preservation. Article 39 allows a party to apply for property preservation and for orders requiring or restraining conduct, both during the arbitration and, in urgent cases, before it is commenced. The institution transmits the application to the court. In a share transfer dispute, an order restraining registration of the transfer can preserve the entire remedy; without it, the award becomes a claim for damages against a counterparty that may no longer hold the shares.

The third is the pressure that follows an unpaid award. A party that does not perform is subject to enforcement measures, and to the credit and consumption restrictions that accompany them. For a joint venture partner that intends to keep operating in China, that is often the point at which settlement becomes rational.

A checklist for the clause and for the file:

  • Name the institution exactly, in the language of registration, and add a prevailing-language rule for the bilingual text.
  • State the seat, and state separately where hearings may be held.
  • State the governing law of the contract, distinct from the law of the arbitration.
  • Fix the number of arbitrators and the language.
  • List the instruments covered, and add a residual scope clause.
  • Provide for consolidation and joinder, or explain why not.
  • Specify an address for service and require notice of any change.
  • Record the internal approvals for the clause itself; a clause signed by someone without authority is a separate problem.
  • Diarise the three-month set-aside window from the date the award is received.
  • Identify, before filing, which claims are contractual and which are organisational, and file accordingly.

The direction of travel is toward more, not less, use of arbitration for these disputes. The revised law supports institutions going abroad and foreign institutions coming into free trade zones, and Article 87 expressly encourages foreign parties to choose a Chinese institution and a Chinese seat. For a foreign investor, that encouragement changes the calculation: a Chinese seat no longer implies a less predictable process than it once did, and a Shanghai-seated arbitration supervised by a specialist commercial court is a defensible default for a company whose assets, factories and partners are all in China. The clause should be drafted on that assumption rather than inherited from a template written for a different decade.

Sources

  • 中国人大网·《中华人民共和国仲裁法》(2025-09-12 修订通过,2026-03-01 施行,全文八章九十六条):http://www.npc.gov.cn/npc/c2/c30834/202509/t20250912_447762.html
  • 中国人大网·新修订的仲裁法自 2026 年 3 月 1 日起施行(2025-09-12):http://www.npc.gov.cn/c2/c30834/202509/t20250912_447759.html
  • 中国人大网·全国人民代表大会常务委员会关于修改《中华人民共和国民事诉讼法》的决定(第二百七十九条专属管辖,2023-09-01):http://www.npc.gov.cn/npc/c2/c30834/202309/t20230901_431419.html
  • 上海市人民代表大会常务委员会·《上海市推进国际商事仲裁中心建设条例》(2023-12-01 施行):https://www.shrd.gov.cn/shrd/gbd348hshsrmdbdhcwwyhgb2023ndsh/content/d9779781-5407-4c35-bc7f-8473a40f075b.html
  • 上海国际仲裁中心·《上海国际经济贸易仲裁委员会(上海国际仲裁中心)仲裁规则(2024)》(2024-01-01 施行):https://shiac.org/h5/SHIAC?moduleCode=arbitrate_rules&securityId=AemPnu8_EA7m8anaza734A
  • 上海国际商事法庭·中外专家齐聚上海国际商事法庭(含收案数量与管辖范围,2026-09):https://shicc.a-court.gov.cn/xxfb/no1court_511/docs/202609/d_-10593.html

外资公司治理争议走仲裁还是诉讼:上海的程序选择指引

本文讨论的是哪一类争议

一家外资企业与中方伙伴设立合资公司。几年之后,关系撞上了四堵常见的墙:外方为维持股比需要增资,中方股东不同意;中方拟将股权转让给第三方,外方主张自己的优先购买权被绕过;董事会通过了一项关联交易决议,外方委派的董事认为这次会议根本没有依章程召集;或者双方在任何事项上都无法达成一致,公司不再作出任何决策。

这四件事,每一件都是公司法问题,每一件又都是合同法问题,因为合资合同与股东协议对这四种情形都会有说法。本文要回答的问题很窄、很具体:当事人已经写了仲裁条款,上述争议哪些可以仲裁、哪些必须进中国法院,条款应当怎么写,才不至于让答案由对方的程序异议来决定。

范围是刻意收窄的。本文不讨论修订后仲裁法的整体面貌,不讨论外国仲裁裁决承认与执行的一般问题,也不讨论货物买卖或建设工程争议的仲裁。它只讨论外资企业内部的治理争议——增资、优先购买权、决议、僵局——以及审理这些争议的场所。

在进入细节之前,先立两个框架。

第一,修订后的仲裁法自 2026 年 3 月 1 日起施行。该法于 2025 年 9 月 12 日通过,共八章九十六条,接替了一部施行三十年的法律。对治理条款而言,真正重要的内容多数不是新的:仲裁协议的三项要件、仲裁协议独立性、有效条款排除法院管辖,这些在修订前就已存在。变化发生在边缘,而治理条款恰好长在边缘上。用语从“仲裁委员会”改为“仲裁机构”;仲裁地成为法定概念;临时仲裁在自由贸易试验区和海南自由贸易港登记的企业之间的涉外纠纷中开了一道窄门;申请撤销裁决的期限缩短为三个月。

第二,上海不是中立的背景板。上海有一部建设国际商事仲裁中心的地方性法规,自 2023 年 12 月 1 日起施行;有一家现行仲裁规则自 2024 年 1 月 1 日起施行的机构;还有一家专门的国际商事法庭,2024 年 12 月底设立,自 2025 年 1 月 1 日起受理案件,集中管辖涉外商事审判与仲裁司法审查。对在长三角有实体经营的企业来说,真正的选择很少是抽象的“仲裁还是诉讼”,而是:写哪一家上海机构、按什么条件写、以及败诉方不付款时怎么办。

三项要件,从严把握

修订后的仲裁法第二十七条规定,仲裁协议包括合同中订立的仲裁条款和以其他书面方式在纠纷发生前或者纠纷发生后达成的请求仲裁的协议,并应当具备三项内容:请求仲裁的意思表示、仲裁事项、选定的仲裁机构。第二十九条接着给出让三项要件真正有杀伤力的后果:对仲裁事项或者仲裁机构没有约定或者约定不明确的,当事人可以补充协议;达不成补充协议的,仲裁协议无效。

两条放在一起读,是这个领域里最贵的一堂起草课。写“争议提交上海仲裁解决”的条款,指定的是城市,不是机构。写“争议提交上海的仲裁委员会”的条款,含义不确定,因为上海不止一家。写了一个并不存在的机构名称,或者写了一个只做调解、不做仲裁的机构,第三项要件即告落空。这些瑕疵都不会因为当事人意图明显而被治愈。判断标准不是意图,是机构。

修订后的法律在用语上放宽了,在要求上没有放宽。第八十九条把“仲裁机构”界定为依法设立的仲裁委员会、仲裁院等机构,这终结了一个老争论:自称中心、自称院的机构能不能管理案件。但它救不了一条什么都没写的条款。第二十七条还增加了一条过去没有明确规定的方式:一方在申请仲裁时主张存在仲裁协议,另一方在首次开庭前不予否认的,经仲裁庭提示并记录,视为当事人之间存在仲裁协议。对条款粗糙但对方不愿纠缠的申请人,这是一张安全网;对以为沉默就能保住立场的被申请人,这是一个陷阱。

另外两条补完整图景。第三十条规定仲裁协议独立存在:合同是否成立及其变更、不生效、终止、被撤销或者无效,不影响已经达成的仲裁协议的效力;仲裁庭有权确认合同的效力。这一点在治理争议中比在普通买卖合同中更重要,因为治理请求最常见的抗辩就是:载有该条款的那份文件本身——增资协议、股权转让协议、修订后的合资合同——无效或者从未获得批准。独立性意味着这个抗辩由仲裁庭听,而不是由被请求驳回起诉的法院听。

第三十一条处理对仲裁协议本身的异议:当事人可以请求仲裁机构或者仲裁庭作出决定,也可以请求人民法院作出裁定;一方请求机构或仲裁庭决定、另一方请求法院裁定的,由法院裁定。异议应当在仲裁庭首次开庭前提出。落到实务上,这类案件的管辖之争都很短、很前置、很早出结果。没有一个漫长的预备阶段可以试探水温。

专属管辖管的是法院,不是仲裁

民事诉讼法第二百七十九条列出三类由中国法院专属管辖的案件。第一类是因在中国领域内设立的法人或者其他组织的设立、解散、清算,以及该法人或者其他组织作出的决议的效力等纠纷提起的诉讼。第二类是因与中国领域内审查授予的知识产权的有效性有关的纠纷提起的诉讼。第三类是因在中国领域内履行中外合资经营企业合同、中外合作经营企业合同、中外合作勘探开发自然资源合同发生纠纷提起的诉讼。

第一项直接落在治理争议上,也最常被读成对管辖问题的回答。它不是。专属管辖是一条写给法院的规则:它告诉中国法院必须受理,不得以方便为由让给外国法院。仲裁法第五条回答的是另一个问题:当事人达成仲裁协议,一方向人民法院起诉的,人民法院不予受理,但仲裁协议无效或者法律另有规定的除外。

两条规则在不同层面上运作。专属管辖解决的是审判权在法院之间如何分配,包括外国法院判决能否被承认的问题。仲裁协议解决的是当事人是否已把争议整体移出审判。因此,一条有效的仲裁条款并不冒犯第二百七十九条,它在更早的一步就生效了。第二百七十九条真正的含义在实务中是:剩下的那条诉讼路径,不是当事人可以靠约定转向外国法院的路径;条款一旦失效,案件留在中国。

这个区分对起草有具体后果。约定在新加坡或者香港仲裁的条款,不因为基础合同是合资合同而无效。但兜底很重要。条款日后被认定无效,争议不会去当事人选定的外国法院,而会去有专属管辖权的中国法院。一个为外国仲裁地谈得很辛苦、却接受了模糊机构名称的当事人,实际上谈成的是中国诉讼。

这里有一片真实的灰色地带,应当说清楚而不是抹平。当请求指向股东会决议的效力,即主张决议根本未成立、或者未经法定多数通过,争议标的看起来就落在第二百七十九条第一项,中国法院倾向于把这类组织法上的请求留在法院。当请求主张某股东违反股东协议、以特定方式投票,同一批事实产生的是合同请求,明显可以仲裁。两类请求可以从同一次会议中同时长出。法律没有在文字上划定这条边界,任何条款也无法消除这个风险。条款能做的是明确分配:把可仲裁的请求写清楚,让它以合同请求的形式提出;把组织法上的请求留在它该在的地方。

还有一条规则补完整图景,而它决定的正是“外国法院在这类争议里值多少”。民事诉讼法第二百八十条处理平行诉讼:同一纠纷一方当事人向外国法院起诉、另一方当事人向人民法院起诉的,人民法院依照本法有管辖权的,可以受理;当事人订立排他性管辖协议选择外国法院管辖,且不违反专属管辖规定、不涉及主权、安全或者社会公共利益的,人民法院可以裁定不予受理,已经受理的裁定驳回起诉。第二百八十一条接着规范以外国法院先行受理为由申请中止诉讼的处理,第二百八十二条列出以更便利的外国法院为由拒绝行使管辖权的条件——凡属专属管辖的,这些通道都不开启。

对治理条款而言,后果是:外国法院就中国公司的解散、或者就其决议效力作出的判决,进不了中国。外国法院判决的承认与执行,不像仲裁裁决那样有一部多边公约托底,而是靠条约与互惠。仲裁裁决有这条路,而且是有条约的那条路。这一不对称,而不是对仲裁程序本身的偏好,是把这类争议写进条款的最强实务理由。

四类治理场景

有了上面的框架,四类场景可以分开看。

增资。 增资通常在两份文件上同时体现:一份股东会决议,一份认购协议或者合资合同修订。合同请求——一方承诺认购、承诺投赞成票、承诺不稀释——可以仲裁。增加注册资本的法人行为则是一项登记事项,走市场监督管理部门的程序,任何仲裁庭都无法命令登记机关作为。实务中,这类案件的裁决是判赔,或者判令一方投票并签署文件;登记随后完成,或者没有完成,而第二步的执行正是这类案件变难的地方。起草良好的条款会预见到这一点:把投票义务与签署义务写明确,并在法律允许的范围内约定代签安排——对公司法文件而言,这种许可范围很窄。

股权转让的优先购买权。 合资合同赋予股东优先购买权的,违反该权利向第三方转让会产生形式上属于合同、可以仲裁的请求。救济通常是判令停止转让、赔偿损失,或者两者并用。时间决定一切:受让方已完成股东登记之后才启动仲裁的,仲裁庭面对的是一个它无法处分的第三方,救济就塌缩成损害赔偿。这是“发现违约”与“提交仲裁申请”之间的间隔构成全部案情的场景,也是临时措施——依第三十九条申请,由机构转交法院——比终局裁决更值钱的场景。

撤销董事会或者股东会决议。 如前所述,组织法上的请求留在法院。但同一次会议通常还产生合同请求:一方委派的董事没有按股东协议的要求获得通知;一项保留事项未经协议要求的同意即被决定;一项关联交易绕过了合同约定的审批机制。这些请求可以仲裁,而认定同意机制被违反的裁决,可以在决议仍然留在登记簿上的情况下支持损害赔偿请求。实务中有人把这描述为“仲裁间接做到了它不能直接做到的事”。这个描述是准确的,值得提前设计,而不是事后发现。

僵局与解散。 这一类界限最清楚。请求司法解散公司,是请法院终结一个法人资格;它是持有一定比例表决权的股东享有的法定救济,不是当事人可以提交仲裁庭的请求。仲裁能做的是执行股东协议本应提供的合同出口:卖出选择权、按公式定价的收购、指定第三方评估师的权利、要求对方按专家确定的公允价值受让的权利。把僵局打到判决的公司常常发现,判决解散公司、按清算价值分配财产,这很少是任何一方想要的结果。合同出口通常是更好的商业结果,也正是仲裁条款能够交付的那一种。

四类场景背后的模式是同一个:仲裁触得到钱和承诺,触不到登记簿、法人资格和组织行为。忽略后一半的条款,产出的是正确却难用的裁决。

条款怎么写

条款很短,所以每个字都有分量。六个要素值得写全。

机构,写全称。 使用登记名称,机构公布过等效名称的,一并写上。第八十九条确认仲裁委员会与仲裁院都算,但更稳妥的做法是照抄机构官网上的准确名称,用其登记语言书写,对方是中国企业的,把中文名称一并写入。条款为双语的,加一句“就机构识别而言以中文文本为准”,以免一处翻译差错变成一条无效理由。

仲裁地。 第八十一条允许当事人书面约定仲裁地,并规定:除当事人对仲裁程序的适用法另有约定外,以仲裁地作为仲裁程序的适用法及司法管辖法院的确定依据,仲裁裁决视为在仲裁地作出。写明仲裁地不等于写明开庭地。约定“开庭在上海、仲裁地在新加坡”的条款,选的是新加坡的程序法与新加坡法院的撤销管辖。这是一个正当的选择,但必须是深思熟虑的选择。

合同准据法。 与仲裁地是两件事,且经常被漏掉。合资合同的准据法多数情形下是中国法;不写,就等于邀请一场“仲裁庭先决定适用哪部法、再决定任何实体问题”的争论。

语言与仲裁员人数。 第四十二条允许一名或者三名。文件有两种语言、董事来自三个法域的治理争议,通常选三名;条款应当写明仲裁语言是中文、英文还是两者兼用,以及文件是否可以不经翻译提交。

合并与追加。 治理争议很少只有两方。增资可能涉及公司、两名股东和一名新投资人;股权转让可能涉及受让方;僵局可能牵涉合资公司之上的持股平台。第二十七条允许一份协议覆盖多个事项,但对多起仲裁的合并只字未提。这个空白由机构规则填补:上海这家机构 2024 年规则对合并仲裁与案外人加入都作了规定,条件由机构掌握。不处理合并的条款,等于让当事人去跑几场结论可能互相矛盾的平行仲裁。

提交范围。 除机构名称写错之外,最常见的瑕疵是范围过窄或者过虚。只覆盖“本协议项下争议”的条款,可能够不到合资合同项下的请求,因为那是另一份文件。覆盖“与公司有关的一切争议”的条款,可能被质疑把不可仲裁的事项也包了进来。可用的写法是先列文件——合资合同、股东协议、章程中涉及合同性违约请求的部分、股权转让或者增资协议——再加一条兜底,覆盖与公司设立、运营、终止有关、且当事人依法有权提交仲裁的争议。

两个小点。其一,送达。第四十一条规定仲裁文件按当事人约定的合理方式送达,没有约定或者约定不明确的,按仲裁规则规定的方式送达。治理争议里常有一方不再回应;条款写明送达地址、并约定地址变更须通知的,就把缺席变成一道常规步骤。其二,保密。第五十二条使仲裁以不公开为原则,只有当事人协议公开才可公开,且涉及国家秘密、他人商业秘密或者个人隐私的除外。对涉及定价、客户名单与董事会记录的合资争议,这一默认是优势——但值得确认所选机构规则不会在当事人不希望的情形下允许披露。

上海:机构、仲裁地与司法审查

上海的供给由三层构成,常被混为一谈。

第一层是政策与法规。上海是全国国际商事仲裁中心建设试点的首批地区之一,2023 年 12 月,一部推进该项建设的地方性法规施行。该法规涉及仲裁机构的治理、行业开放、临时仲裁与仲裁地、以及支持政策。它是地方性法规,不是国家法律,但它是若干制度特征先在上海出现的原因。

第二层是机构。上海这家机构的现行规则自 2024 年 1 月 1 日起施行,接替 2014 年的自由贸易试验区仲裁规则。它是一套文件而不是一份文本:一部主规则、航空与数据两部特别规则、一份在线仲裁指引、一份临时仲裁协助服务指引。与治理争议相关的功能包括合并仲裁、案外人加入、临时措施与紧急仲裁员救济、更开放的仲裁员选择方式,以及在香港设立中心。其下设的自由贸易试验区仲裁院、航空仲裁院与各专业平台同属该机构,规则把对平台的约定视为对该机构的约定——这恰好是前面所说“机构名称写错”问题的一剂解药。

第三层是法院。上海国际商事法庭于 2024 年 12 月底设立,自 2025 年 1 月 1 日起受理案件。它集中管辖涉外商事一审、二审案件,仲裁司法审查案件,涉外调解协议确认案件,以及申请承认和执行外国法院民商事判决、裁定的审查案件。截至 2026 年 8 月底,已受理涉外商事与仲裁司法审查类案件三千七百余件,当事人覆盖六十四个国家和地区。2026 年 6 月,它设立了由来自八个国家和地区的二十名专家组成的国际商事专家委员会。

对治理条款而言,意义在于:监督这场仲裁的法院可以事先确定,而且是一家专门法院,不是普通民事审判庭。第七十一条规定的撤销申请向仲裁机构所在地的中级人民法院提出;对由上海机构管理、仲裁地在上海的仲裁,就是这家专门法院。第七十二条要求自收到裁决书之日起三个月内提出,比修订前适用的六个月更短;第七十三条要求法院在受理撤销申请之日起两个月内作出裁定。两个数字要一起读:提起的窗口很短,之后的不确定期也很短。

第八十三条列出涉外仲裁裁决的撤销事由,比国内裁决的清单更窄:没有仲裁协议;被申请人没有得到指定仲裁员或者进行仲裁程序的通知,或者由于其他不属于被申请人负责的原因未能陈述意见;仲裁庭的组成或者仲裁程序与仲裁规则不符;裁决事项不属于仲裁协议范围或者仲裁机构无权仲裁。违背公共利益是单独一项事由。值得注意的是,国内裁决事由中的“裁决所根据的证据是伪造的”并不出现在涉外清单里。对外国投资者而言,这一差别有实际意义:审查更接近示范法式的程序清单,而不是对实体问题的重新审理。

裁决之后,以及一份清单

第七十条规定,裁决书自作出之日起发生法律效力。第七十五条规定,一方不履行的,另一方可向人民法院申请执行,受申请的法院应当执行。实务中随后是三步。

第一步是执行。财产在中国境内的,向有管辖权的中级人民法院申请。被执行人或其财产在中国境外的,第八十五条允许直接向有管辖权的外国法院申请承认和执行;第八十八条规范在中国境外作出的裁决的承认与执行,依据是条约或者互惠——对中国的多数贸易伙伴而言,实际就是《纽约公约》。

第二步是保全。第三十九条允许当事人申请财产保全,以及请求责令一方作出一定行为或者禁止其作出一定行为;既可以在仲裁过程中申请,情况紧急的也可以在申请仲裁前申请,由仲裁机构将申请提交人民法院。在股权转让争议中,一份禁止办理转让登记的裁定可以保住整个救济;没有它,裁决就退化为对一个可能已不持有股权的对手方的损害赔偿请求权。

第三步是裁决未获履行之后的压力。不履行的一方将面对执行措施,以及随之而来的信用与消费限制。对一个打算继续在中国经营的合资伙伴,这往往就是和解变得理性的时点。

条款与卷宗的清单:

  • 准确写明机构名称,使用登记语言,并为双语文本加一条以何种文本为准的规则。
  • 写明仲裁地,另行写明开庭地。
  • 写明合同准据法,与仲裁程序适用的法律分开。
  • 确定仲裁员人数与语言。
  • 列出所覆盖的文件,并加一条兜底范围。
  • 对合并与追加作出安排,或者说明为什么不安排。
  • 指定送达地址,并要求地址变更须通知。
  • 把条款本身的内部授权记录下来;由无权人员签署的条款是另一个问题。
  • 自收到裁决书之日起,为三个月的撤销窗口设置台账提醒。
  • 申请仲裁之前,先分清哪些请求是合同性的、哪些是组织法上的,再据此提出。

方向是这些争议会更多、而不是更少地使用仲裁。修订后的法律支持中国机构走出去、支持境外机构进入自由贸易试验区,第八十七条明确鼓励涉外仲裁当事人选择中国(包括特别行政区)的仲裁机构、约定中国作为仲裁地。对外国投资者而言,这一鼓励改变了算法:中国仲裁地不再意味着比过去更难预测的程序,而对一家资产、工厂与合作伙伴都在中国的公司来说,由专门商事法庭监督的上海仲裁,是一个站得住脚的默认选项。条款应当按这个前提起草,而不是从为另一个十年写的模板里继承下来。

来源

  • 中国人大网·《中华人民共和国仲裁法》(2025-09-12 修订通过,2026-03-01 施行,全文八章九十六条):http://www.npc.gov.cn/npc/c2/c30834/202509/t20250912_447762.html
  • 中国人大网·新修订的仲裁法自 2026 年 3 月 1 日起施行(2025-09-12):http://www.npc.gov.cn/c2/c30834/202509/t20250912_447759.html
  • 中国人大网·全国人民代表大会常务委员会关于修改《中华人民共和国民事诉讼法》的决定(第二百七十九条专属管辖,2023-09-01):http://www.npc.gov.cn/npc/c2/c30834/202309/t20230901_431419.html
  • 上海市人民代表大会常务委员会·《上海市推进国际商事仲裁中心建设条例》(2023-12-01 施行):https://www.shrd.gov.cn/shrd/gbd348hshsrmdbdhcwwyhgb2023ndsh/content/d9779781-5407-4c35-bc7f-8473a40f075b.html
  • 上海国际仲裁中心·《上海国际经济贸易仲裁委员会(上海国际仲裁中心)仲裁规则(2024)》(2024-01-01 施行):https://shiac.org/h5/SHIAC?moduleCode=arbitrate_rules&securityId=AemPnu8_EA7m8anaza734A
  • 上海国际商事法庭·中外专家齐聚上海国际商事法庭(含收案数量与管辖范围,2026-09):https://shicc.a-court.gov.cn/xxfb/no1court_511/docs/202609/d_-10593.html

China’s 2025 Encouraged Catalogue Is in Force: Turning 1,679 Entries into Duty, Land and Filing Outcomes

  • The Encouraged Catalogue for Foreign Investment (2025 Edition) has been in force since 1 February 2026, with 1,679 entries in total, a net increase of 205 over the 2022 edition and 303 entries revised.
    《鼓励外商投资产业目录(2025 年版)》自 2026 年 2 月 1 日起施行,总条目 1679 条,较 2022 年版净增 205 条、修改 303 条。
  • It is built as two annexes rather than one list: a national catalogue of 619 entries, up 100, and a regional catalogue for the central and western regions, the north-east and Hainan of 1,060 entries, up 105.
    它由两个附件而非一份清单构成:全国目录 619 条,增加 100 条;适用于中西部等地区、东北地区与海南省的地区目录 1060 条,增加 105 条。
  • The Encouraged Catalogue is an industrial guide, not a market-access licence: an entry does not admit a project that the Negative List still restricts, and being listed approves nothing by itself.
    鼓励目录是产业指引,不是准入许可:列入条目并不能让《外商投资准入负面清单》仍然限制的项目获得准入,被列入本身也不批准任何事情。
  • Import Duty Exemption is available for self-use equipment imported within the total investment amount, together with the technology, parts and components that arrive with it under contract, subject to two exclusion lists, with import VAT still collected in full.
    进口自用设备免征关税,限于投资总额内进口的自用设备以及按合同随设备进口的技术、配套件与备件,须排除两份不予免税目录所列商品,进口环节增值税仍照章征收。
  • The confirmation letter that unlocks the exemption carries a code, and the code is what the customs system reads: AC plus three digits for a national entry, HB plus four for a regional one.
    用于享受免税的《项目确认书》上载有产业政策条目代码,海关系统读的就是这个代码:全国目录条目由 AC 加三位数字组成,地区目录条目由 HB 加四位数字组成。
  • Industrial Land Pricing for an intensive-use encouraged industrial project may be set at no less than 70 per cent of the national minimum standard for the corresponding land grade, which is a floor written into a 2017 State Council notice, not a discount to be assumed in a business plan.
    集约用地的鼓励类外商投资工业项目,出让底价可按不低于所在地土地等别相对应全国工业用地出让最低价标准的 70% 执行;这条底线写在 2017 年的一份国务院通知里,不是商业计划里可以当然假设的折扣。
  • The Central And Western Regions annex is province-specific, and the 2025 revision added one named entry for each of the six central provinces, from carbon-based new materials in Shanxi to smart agricultural equipment in Hunan.
    地区目录按省列条,2025 年修订为中部六省各增一条具名条目,从山西的碳基新材料到湖南的智能化现代农业装备。

China’s 2025 Encouraged Catalogue Is in Force: Turning 1,679 Entries into Duty, Land and Filing Outcomes | 2025 年版鼓励目录已施行:1679 条如何换成关税、用地与备案上的结果

What the catalogue is, and what it is not

The Encouraged Catalogue for Foreign Investment is one of the oldest instruments in China’s foreign investment toolkit and, for most companies, the least read. It is not a permission list, it is not a prohibition list, and it does not appear in the standard sequence of approvals that a new project passes through. It is a statement of where the state would like foreign capital to go, and it is attached to a small number of concrete consequences.

The 2025 edition was issued as Order No. 37 of the National Development and Reform Commission and the Ministry of Commerce, dated 15 December 2025, published on 24 December 2025, and effective from 1 February 2026. The 2022 edition was repealed on the same date. The legal basis is the Foreign Investment Law and its implementing regulations, which require a catalogue setting out the specific industries, fields and regions in which foreign investors are encouraged and guided to invest.

The document’s own opening paragraph contains the sentence that a compliance team should read first, because it disposes of most of the confusion around what an entry means. It says that, subject to compliance with market access policy, where a listed industry is identified by the Industrial Structure Adjustment Guidance Catalogue or another national industrial catalogue as a type of industry to be eliminated or prohibited, its encouraged status falls away automatically.

Two things follow from that sentence. The first is that the catalogue is subordinate to market access policy: it operates within the space that the Negative List leaves, and it does not enlarge that space. The second is that the catalogue is internally conditional: an entry can be encouraged on paper and lose that character because another catalogue has since classified the activity differently. A company that checks the entry once and files the printout is checking half the question.

The structure is two annexes. The national catalogue applies nationwide. The regional catalogue applies to the central and western regions, the north-east and Hainan, and it is organised province by province. Both are part of the same order and both are revised together, but they are not interchangeable, and the difference between them is not cosmetic: it determines the code on the confirmation letter, the description of the project’s nature, and in some cases whether a project qualifies at all.

It is worth being clear about the function of the catalogue in the life of a project. A foreign-invested project goes through establishment or change filing under the Foreign Investment Law, sector-specific licensing where the activity requires it, environmental, energy and planning approvals where they apply, and market access screening against the Negative List. The encouraged catalogue does not replace any of that. What it does is attach benefits: duty treatment on imported equipment, land pricing treatment, and eligibility for certain regional tax policies where a separate instrument provides for them. The catalogue is the key, not the door.

There is one further structural feature that explains why companies find the document harder to use than its length suggests. The two annexes are drafted in different registers. The national catalogue reads as an industrial policy document: it describes technologies, processes, product families and service types in the language of a five-year plan. The regional catalogue reads as an investment promotion document: it describes what a particular province would like to attract, in the language of that province’s existing industrial base. A reader who has learned to navigate one annex will not automatically be able to navigate the other, and the entries that generate the most value for a manufacturing investor are usually in the second.

The numbers, and where the growth sits

The headline figures are these. Total entries: 1,679. Net increase over the 2022 edition: 205. Entries revised: 303.

The split matters more than the total. The national catalogue has 619 entries, an increase of 100 and a revision of 131. The regional catalogue has 1,060 entries, an increase of 105 and a revision of 172. Roughly two-thirds of the catalogue, in other words, is regional, and the regional part is where a project’s location becomes a legal variable rather than a commercial preference.

The relationship between the net increase and the number revised is worth pausing on, because it tells a company something about how to read the document. A net increase of 205 alongside 303 revisions means the edition did not simply add pages to the previous one. Something close to a fifth of the catalogue was rewritten, and a revision can be as consequential as an addition: wording that narrows a description can remove a project from scope just as effectively as deleting the entry, and wording that broadens one can bring in a project that previously did not fit. Companies that compare the two editions entry by entry, rather than checking only that their entry still exists, are doing the more useful comparison.

The official account of the revision identifies three directions of change. The first is advanced manufacturing: the national catalogue adds or extends entries in terminal products, components and raw materials. The second is modern services: additions and extensions in business services, technical services, scientific research and service consumption. The third is regional balance: the regional catalogue is expanded in line with what each area is actually good at, covering the central and western regions, the north-east and Hainan.

The third direction produced the most concrete news of the revision, and it is the part that a company with a plant in central China should read closely. At a State Council Information Office press conference on 16 November 2026, the Ministry of Commerce identified six entries added to the regional catalogue for the six central provinces. Shanxi received development, production and processing of carbon-based new materials. Anhui received specialised services for the commercialisation of new energy vehicle technological achievements. Jiangxi received stamping-die forming and manufacturing of electronic and electrical components. Henan received research, development and production of high-end intelligent hoisting and engineering machinery. Hubei received research, development, manufacturing and production of new-type energy storage equipment. Hunan received production of intelligent modern agricultural equipment.

Read as a set, the six entries describe a theory of the central provinces that is neither the coastal export-processing model nor a generic “high technology” label. Each is anchored to an existing industrial base: coal and coke chemistry in Shanxi, an automotive cluster in Anhui, electronics assembly in Jiangxi, construction and material-handling machinery in Henan, battery and storage supply chains in Hubei, agricultural machinery in Hunan. The entries are narrow by design, which is what makes them usable: a project either matches the description or it does not, and the confirmation letter will say which.

There is a practical asymmetry in that narrowness that deserves emphasis. A narrow entry is easy to match and easy to miss. It is easy to match because the words are specific. It is easy to miss because a company that reads the catalogue looking for its industry may not recognise its own project in a description written from the perspective of a value chain rather than a corporate org chart. A supplier of forming dies to an electronics assembler in Jiangxi is not, in its own internal language, in the business of “stamping-die forming and manufacturing of electronic and electrical components”. It is, in the language of the catalogue.

The import duty exemption, and the code the customs system reads

The benefit that most companies actually claim is the duty treatment on imported equipment, and it is administered by customs rather than by the development or commerce authorities.

The current instrument is Customs Announcement No. 13 of 2026, issued on 27 January 2026 and effective from 1 February 2026, on the implementation of the 2025 catalogue. Its first article states the rule: from 1 February 2026, for foreign-invested projects within the scope of the 2025 catalogue, including capital increase projects, self-use equipment imported within the total investment amount, together with the technology and the parts and components imported with that equipment under contract, are exempt from customs duty and subject to import VAT in accordance with the rules, except for goods listed in the Catalogue of Import Commodities Not Exempt from Duty for Foreign-Invested Projects and the Catalogue of Major Technical Equipment and Products Not Exempt from Import Duty.

Four elements of that sentence carry the operational weight.

The first is “within the total investment amount”. The exemption is not a general relief on imported equipment; it is bounded by the project’s registered total investment. A project that imports equipment beyond that amount has, by definition, equipment outside the scope, and the file has to show which is which.

The second is “self-use”. Equipment imported for use by the project is treated differently from equipment imported for sale, lease or use by another entity. The boundary is not always obvious in a group with shared service companies, and it is the point at which a plausible claim becomes a documentation problem.

The third is the two exclusion lists. Both are negative lists sitting inside a positive benefit: the goods named in them do not get the exemption even where the project qualifies. Checking the entry is therefore necessary but not sufficient; the equipment also has to be checked.

The fourth is the treatment of import VAT. The announcement exempts duty and collects VAT in accordance with the rules. The two are not the same, and a financial model that assumes both are waived will overstate the benefit by an amount that depends on the equipment mix.

The procedural part is where the codes matter. When the competent authorities issue the Confirmation of Domestic and Foreign-Funded Projects under State Encouragement, or the receipt for the initial report or the change report of a foreign-invested company or partnership, they record the applicable industrial policy entry code and the nature of the project. Under the 2026 announcement, entries in the national catalogue carry a code composed of AC and three digits — the first entry, for example, is AC001. Entries in the regional catalogue carry a code composed of HB and four digits — the first entry for Shanxi, for example, is HB1401. Where the national catalogue applies, the nature of the project is recorded as foreign investment, or as domestic goods within a foreign-funded project. Where the regional catalogue applies, it is recorded as a central and western advantage industry for foreign investment, or again as domestic goods within a foreign-funded project.

Those codes are not decoration. They are the field the customs system reads when a declaration is made, and they are the reason a project’s classification has to be settled before the first shipment rather than after it. A declaration made under the wrong code is not corrected by explaining the business in a covering letter.

The announcement also handles the transition, and the transition rules are the part most often overlooked. Projects approved, verified or filed before 1 February 2026 that fall within the 2025 catalogue may obtain a confirmation under the new catalogue and proceed. Projects that fall within the 2022 catalogue but not the 2025 catalogue may still proceed if the authorities issued the relevant confirmation or receipt under the 2022 catalogue before 1 February 2027 — a deadline, not an open-ended grandfathering. And projects under construction that were outside the 2022 catalogue but inside the 2025 catalogue may enjoy the treatment, except that duty already paid on equipment imported before that point is not refunded.

The last of those three is the one that catches companies by surprise, because it inverts the usual assumption that a new and more generous catalogue makes a company whole. It does not. It makes the company eligible going forward, and leaves the past where it is.

One further obligation belongs in the same paragraph. Equipment admitted under the exemption remains goods under customs supervision. During the supervision period it may not be sold, transferred, exchanged, leased, mortgaged, pledged, put to another use, used in another location or shipped out without customs approval and the relevant formalities, and the holder must report the use of the goods to the competent customs office by 30 June each year during the supervision period. The exemption is a status attached to specific goods, not a one-off tax saving.

Land: the 70 per cent floor and its conditions

The second benefit is land, and it is the one most often described inaccurately in company presentations.

The rule is in the State Council notice on measures to expand opening up and actively utilise foreign investment, issued in January 2017. Article 17 of that notice states three things in sequence. Foreign-invested enterprises and domestic enterprises apply the same land policies. Land continues to be supplied on a priority basis to encouraged foreign-invested industrial projects that use land intensively. And in determining the reserve price for the grant of land use rights, the price may be set at no less than 70 per cent of the national minimum price standard for industrial land corresponding to the grade of the locality.

Three conditions are compressed into that sentence, and each of them does work.

The project must be encouraged, which means it must match an entry in the catalogue in force at the time. The project must be industrial, which excludes a good deal of what a services company might otherwise expect. And the project must use land intensively, which is an assessment of floor area ratio, investment intensity and output per unit of land rather than a self-description.

What the rule produces is a floor, not a discount. The 70 per cent figure is the lowest proportion of the national minimum standard that the reserve price may be set at; it is not a rate that applies automatically, and it is not a national price. The actual reserve price is set in the grant scheme for the specific parcel, by the local government, having regard to the grade of the land, the industrial policy, and the local plan. A business plan that books land at 70 per cent of the national minimum, before the parcel has been identified, is booking an assumption.

The distinction between a floor and a discount matters for a second reason. Because the rule is framed as a permission for local authorities rather than an entitlement for investors, the practical question is not whether the rule exists but whether the local grant scheme applies it. That question is answered in the negotiation over the investment agreement, at the point when the land parcel, the investment intensity and the construction timetable are being settled together. It is not answered by citing the notice in a letter after the scheme has been published.

There is also a sequencing point that follows from the rest of this article. The land treatment and the duty treatment both depend on the same underlying fact — that the project matches an encouraged entry — but they are administered by different authorities at different moments. The land question is settled with the local government before the parcel is granted. The duty question is settled with customs before the first shipment arrives. A project that establishes its encouraged status once, thoroughly, and documents it, uses the same file for both. A project that treats them as separate negotiations pays for the analysis twice.

Three things the catalogue does not give you

The most useful service a catalogue article can perform is to state the boundaries plainly, because the three most common errors in this area all involve assuming that the catalogue gives more than it does.

The first boundary is market access. The catalogue does not admit anything. Market access for a foreign investor is governed by the Special Administrative Measures for Foreign Investment Access, the Negative List, currently the 2024 edition, issued as Order No. 23 and in force since 1 November 2024, with 29 restrictive measures nationwide and the manufacturing sector’s restrictions eliminated. Where the Negative List restricts an activity, an entry in the encouraged catalogue does not lift the restriction; where the Negative List is silent, the activity is managed on the same basis as for domestic investors, and the encouraged catalogue is not needed to permit it. The two documents answer different questions and are frequently confused because both are lists.

The second boundary is corporate income tax. The encouraged catalogue does not itself carry a corporate income tax rate. The 15 per cent rate that companies often associate with it comes from a different instrument with a different scope: the 2020 announcement on continuing the western development corporate income tax policy, issued jointly by the Ministry of Finance, the State Taxation Administration and the National Development and Reform Commission. That announcement applies a 15 per cent rate from 1 January 2021 to 31 December 2030 to encouraged industrial enterprises located in the western region, and it defines such an enterprise by reference to a different catalogue — the Catalogue of Encouraged Industries in the Western Region — with the requirement that the business whose projects are listed in that catalogue constitutes the main business and generates more than 60 per cent of total revenue. It also defines the western region by an enumerated list of provinces, autonomous regions and municipalities, plus named prefectures elsewhere that may apply the policy by analogy.

Two catalogues, two scopes, one rate. A company in Hubei with an entry in the regional annex of the encouraged catalogue does not thereby get 15 per cent; Hubei is not in the enumerated western region, and the relevant catalogue is a different document. A company in Sichuan with an entry in the western region catalogue may get it, subject to the 60 per cent test, whether or not its project appears in the foreign investment catalogue. The two can overlap; they do not substitute for each other.

The third boundary is the duty-free qualification for a foreign-invested research and development centre. That qualification exists, it is valuable, and it is a separate determination made under a separate instrument, with its own thresholds, its own application file and its own approval process conducted at provincial level. A company that has an encouraged entry and imports equipment for a research centre should not assume that the equipment exemption under the catalogue and the research centre qualification are the same benefit claimed twice, or that one application covers both. They are separate gates with separate evidence.

Five steps, from entry to file

Set out as a sequence, the work is unglamorous and finite.

The first step is to find the entry. That means reading the catalogue in force, in both annexes, against the actual description of what the project will do — not against the company’s internal business unit name. The search should be recorded: which entries were considered, which was selected, which were rejected and why. A file that contains only the selected entry is a file that cannot answer a question about the alternatives.

The second step is to match the project. The entry describes an activity; the project has a scope, a location, an investment total and an output. The match has to be made at the level of the activity, and it has to survive the location test, because the regional annex applies only where the project is actually located in the province whose entry is being relied on. A project in one province cannot rely on another province’s entry, and a project in a province covered by the regional annex can still rely on a national entry if one fits better.

The third step is to assemble the file. For the duty benefit, that means the application form for duty exemption on imported and exported goods; the confirmation issued by the competent authority, or the initial or change report receipt, together with the project filing certificate, the approval document or the feasibility study approval; the import and export contracts, invoices and product information; and anything else customs requires. The file has to be internally consistent: the equipment in the contracts has to be the equipment in the project description, the amount has to sit inside the total investment, and the goods have to survive the two exclusion lists.

The fourth step is to obtain the confirmation. This is where the entry code is fixed, and where the description of the project’s nature is fixed with it. Two practical points arise. The first is that the code follows the catalogue version, so a project that obtained a confirmation under the 2022 catalogue and is still importing equipment should check whether the confirmation it holds is still the right instrument, and by when. The second is that a change of plan after the confirmation is not a formality: if the project’s scope, location or investment changes materially, the confirmation and the code may need to be revisited before the next shipment.

The fifth step is to keep the record. The exemption creates a supervision period, an annual reporting obligation by 30 June, and restrictions on disposal, lease, mortgage and relocation without approval. Those are not archive items; they are live obligations that attach to specific machines, and they travel with the equipment if the business is reorganised. A company that treats the exemption as a tax win at the import stage and stops thinking about it will discover the obligation at the worst possible moment, which is when it wants to move a line to another site.

One test applies across all five steps and is worth stating on its own. The file should be intelligible to a reader who knows nothing about the business. That reader — a customs officer, an auditor, a provincial reviewer, a successor in the compliance function three years from now — will be trying to answer a single question: does the equipment in the contracts correspond to the activity in the entry, within the amount in the project. If answering that question requires a briefing on how the company is organised, the file is not finished. The entries are written in the language of activities for exactly this reason, and a project description that mirrors that language is doing most of the work.

Two traps, and what to watch

The first trap is the mismatch between the annex relied on and the nature recorded. The customs announcement distinguishes the nature of a project that relies on the national catalogue from one that relies on the regional catalogue, and the distinction is not merely descriptive: it determines which code applies, and the code is what the system reads. A project in a central province that relies on a national entry is a different thing, in the file, from a project in the same province that relies on the provincial entry. Both may be legitimate. Only one of them matches the paperwork if the paperwork was drafted on the assumption that location alone determines the answer.

The second trap is the treatment of capital increase projects. The duty rule expressly includes them, and this is easy to overlook because the mental model most teams carry is of a greenfield investment with a single total investment figure. An existing foreign-invested company that increases its total investment and imports additional equipment for the expanded scope is within the rule, but the file has to show the increase, the revised total, and the link between the new equipment and the expanded activity. A company that assumes expansion is outside the regime leaves a benefit on the table; a company that assumes it is inside without documenting the increase creates a problem at audit.

Beyond the traps, two things are worth watching.

The first is the interaction between the catalogue and sector-specific opening measures. The 2025 revision expands services entries, and services are the area where market access restrictions and encouraged status most often sit in the same sentence. A services project that reads its entry as an access permission will be disappointed; a services project that reads its entry as a benefit attached to an activity it is already permitted to conduct will find the file straightforward. The distinction is the same one described above, and it will keep recurring as the services list grows.

The second is timing. The catalogue runs from 1 February 2026, and the transitional window for confirmations issued under the 2022 catalogue closes on 1 February 2027. Companies with equipment programmes that straddle that date — orders placed in 2025, deliveries in 2027, phased commissioning across both — are the ones for whom the transition rules are not academic. The work that matters for them is a shipment-by-shipment mapping against the two catalogues, done now, while there is still time to obtain the right confirmation for the right shipment.

The catalogue itself is a document of intent, and its value to a company is entirely a function of how precisely the company can describe its own project in the document’s language. That is a drafting exercise, not a lobbying one, and it is best done before the first contract is signed rather than after the first container arrives.

Sources

  • 中国政府网·《鼓励外商投资产业目录(2025 年版)》(国家发展改革委、商务部令第 37 号,2025-12-15,2026-02-01 施行):https://www.gov.cn/zhengce/zhengceku/202512/content_7052666.htm
  • 国家发展改革委·《鼓励外商投资产业目录(2025 年版)》2025 年第 37 号令(含目录全文附件):https://www.ndrc.gov.cn/xxgk/zcfb/fzggwl/202512/t20251224_1402564.html
  • 国家发展改革委·优化外资结构布局 更大力度吸引外商投资(2025-12-29,答记者问,含 1679 条与两附件条目数):https://www.ndrc.gov.cn/wsdwhfz/202512/t20251229_1402711.html
  • 海关总署·公告 2026 年第 13 号(关于执行《鼓励外商投资产业目录(2025 年版)》有关事项的公告,2026-01-27):https://www.customs.gov.cn/customs/2026-01/28/article_2026012810023759168.html
  • 中国政府网·国务院关于扩大对外开放积极利用外资若干措施的通知(国发〔2017〕5 号,第十七条用地价格):https://www.gov.cn/zhengce/content/2017-01/17/content_5160624.htm
  • 国家税务总局·关于延续西部大开发企业所得税政策的公告(财政部 税务总局 国家发展改革委公告 2020 年第 23 号):https://www.chinatax.gov.cn/chinatax/n810341/n810765/c101653/202004/c5155091/content.html
  • 中国政府网·2024 年版全国外商投资准入特别管理措施(负面清单)发布(29 条,2024-11-01 施行):https://www.gov.cn/lianbo/bumen/202409/content_6973049.htm
  • 法治网·商务部多措并举支持中部地区崛起(2026-11-16 国新办发布会,中部六省新增条目):https://www.legaldaily.com.cn/index/content/2026-11/16/content_9491852.html
  • 宁波海关·鼓励类投资项目进口减免税享惠指南(海关答疑,办理流程与后续监管):https://ningbo.customs.gov.cn/ningbo_customs/2026-04/20/article_2026063011585436091.html

2025 年版鼓励目录已施行:1679 条如何换成关税、用地与备案上的结果

这份目录是什么,不是什么

《鼓励外商投资产业目录》是中国外资政策工具箱里资格最老的文件之一,对大多数企业来说也是被读得最少的一份。它不是许可清单,不是禁止清单,也不出现在一个新项目要走的常规审批序列里。它说的是国家希望外资往哪里去,并且挂着少数几项具体后果。

2025 年版以国家发展改革委、商务部第 37 号令发布,成文日期 2025 年 12 月 15 日,12 月 24 日公开,自 2026 年 2 月 1 日起施行,2022 年版同日废止。上位依据是《外商投资法》及其实施条例——后者要求制定鼓励外商投资产业目录,列明鼓励和引导外国投资者投资的特定行业、领域、地区。

文件开头那一段里有一句话,合规团队应当最先读,因为它把围绕”列入条目意味着什么”的大部分混乱一次解决掉了。原文的意思是:在符合市场准入政策的前提下,如所列产业被《产业结构调整指导目录》等国家相关产业目录明确为淘汰、禁止等类型产业,其鼓励类属性自然免除。

这句话推出两件事。第一,目录从属于市场准入政策:它在负面清单留出的空间里运作,并不扩大这个空间。第二,目录自身是附条件的:一条条目可以在纸面上是鼓励类,却因为另一份目录后来对该活动作了不同分类而丧失这个属性。只查一次条目、把打印件归档的公司,只查了问题的一半。

结构上是两个附件。全国目录适用于全国。地区目录适用于中西部等地区、东北地区以及海南省,按省排列。两者同属一份令、同时修订,但不可互换,差别也不是装饰性的:它决定《项目确认书》上的代码、决定项目性质的记载,在某些情况下还决定项目是否够格。

目录在一个项目的生命周期里起什么作用,值得说清。一个外商投资项目要按《外商投资法》走设立或者变更报告,要按活动性质取得相应的行业许可,在适用时还要过环保、能耗、规划等审批,并且要对照负面清单做准入判断。鼓励目录不替代其中任何一项。它做的是挂上好处:进口设备的关税待遇、用地价格待遇,以及在另有专门文件规定时可适用的某些区域税收政策资格。目录是钥匙,不是门。

还有一处结构特征,解释了为什么这份文件用起来比它的厚度所暗示的更难。两个附件的写作体例不同。全国目录读起来像一份产业政策文件:它用五年规划的语言描述技术、工艺、产品族与服务类型。地区目录读起来像一份招商文件:它用该省既有产业基础的语言,描述这个省想吸引什么。学会了在一个附件里找条目的读者,不会自动在另一个附件里也找得到;而制造类投资者最常出效益的条目,往往在第二个里。

数字,以及增量落在哪里

总量是这样:总条目 1679 条,较 2022 年版净增 205 条,修改 303 条。

拆分比总量更要紧。全国目录 619 条,增加 100 条,修改 131 条。地区目录 1060 条,增加 105 条,修改 172 条。也就是说,这本目录大约三分之二是区域性的,而区域部分正是”项目落在哪里”从商业偏好变成法律变量的地方。

净增与修改之间的关系值得停一下,因为它告诉企业该怎么读这份文件。净增 205 条、同时修改 303 条,意味着这一版不是在上版之上加页。接近五分之一的条目被重写过,而重写的分量不亚于新增:措辞收紧,可以把一个项目挡在范围之外,效果和删掉条目一样;措辞放宽,可以把原本对不上的项目纳进来。逐条比对两版、而不是只确认自己那条还在的公司,做的是更有用的比对。

官方对本次修订的说明给出三个方向。一是鼓励引导外商投资先进制造业,全国目录在终端产品、零部件、原材料等领域新增或扩展条目。二是鼓励引导外商投资现代服务业,在商务服务、技术服务、科学研究、服务消费等领域新增或扩展条目。三是优化区域布局,结合各地资源禀赋、特色优势和产业发展实际,扩大中西部等地区、东北地区与海南省的鼓励范围。

第三个方向产出了本次修订中最具体的新闻,也是在中部设厂的企业应当细读的部分。2026 年 11 月 16 日国务院新闻办公室发布会上,商务部点名了为中部六省各新增的一条地区目录条目:山西为碳基新材料开发生产和加工;安徽为新能源汽车科技成果转化专业服务;江西为电子电气零部件冲压模具成形制造;河南为高端化智能化起重工程机械研发生产;湖北为新型储能装备研发制造和生产;湖南为智能化现代农业装备生产。

把这六条当成一组来读,它们描述的是一套关于中部省份的判断,既不是沿海出口加工模式,也不是笼统的”高新技术”标签。每一条都锚在既有的产业基础上:山西的煤化工与碳素,安徽的汽车集群,江西的电子组装,河南的工程与物料搬运机械,湖北的电池与储能链条,湖南的农机。条目窄是有意为之,而窄正是它可用的原因:项目要么符合这段描述,要么不符合,《项目确认书》会写明是哪一种。

这种窄带来一处值得强调的不对称:窄条目容易对上,也容易漏掉。容易对上,是因为用词具体;容易漏掉,是因为照着行业去翻目录的公司,可能认不出自己的项目出现在一段从产业链而不是从企业组织架构出发写成的描述里。一家为江西电子组装厂供货的冲压模具企业,在自己的内部语言里并不做”电子电气零部件冲压模具成形制造”这门生意。但在目录的语言里,它做。

进口设备免税,以及海关系统读的那个代码

企业实际最常申领的好处是进口设备的关税待遇,而它由海关执行,不由发展改革或商务部门执行。

现行依据是海关总署 2026 年第 13 号公告,2026 年 1 月 27 日发布、2 月 1 日起执行,内容是关于执行 2025 年版目录的有关事项。第一条把规则讲清:自 2026 年 2 月 1 日起,对属于 2025 年版目录范围的外商投资项目(包括增资项目),在投资总额内进口的自用设备和按照合同随上述设备进口的技术及配套件、备件,除《外商投资项目不予免税的进口商品目录》和《进口不予免税的重大技术装备和产品目录》所列商品外,按照国发〔1997〕37 号文和海关总署公告 2008 年第 103 号及其他相关规定免征关税,照章征收进口环节增值税。

这句话里有四个要素承担了操作重量。

第一个是”在投资总额内”。免税不是对进口设备的普遍减免,它被项目的投资总额框住。进口设备超出这个总额的项目,按定义就有落在范围之外的部分,而材料必须说清哪一部分是哪一部分。

第二个是”自用”。项目自用的设备,与为销售、租赁或者供其他主体使用而进口的设备,待遇不同。在拥有共享服务公司的集团里,这条界线并不总是清楚,而它正是”看似站得住的主张”变成”举证问题”的那一点。

第三个是两份不予免税目录。两者都是嵌在优惠里的负面清单:即便项目够格,列在其中的商品也拿不到免税。因此查条目是必要的,但不充分;设备本身也要查。

第四个是进口环节增值税的处理。公告免征关税、照章征收进口环节增值税。两者不是一回事,把两项都当成免征的财务模型,会高估收益,高估多少取决于设备构成。

程序部分就是代码起作用的地方。主管部门按照目录出具《国家鼓励发展的内外资项目确认书》,或者《外商投资(公司/合伙企业)初始报告回执》《变更报告回执》时,会记载适用的产业政策条目代码与项目性质种类。按 2026 年这份公告,全国目录条目由 AC 加三位数字组成,例如第 1 条对应 AC001;地区目录条目由 HB 加四位数字组成,例如山西省第 1 条对应 HB1401。适用全国目录的,项目性质记为”外商投资”或”外资项目内资商品”;适用地区目录的,记为”外资中西部优势产业”或”外资项目内资商品”。

这些代码不是装饰。它们就是报关时海关系统读的那个字段,也是项目的分类必须在第一批货到港之前定下来、而不是之后补的原因。用错了代码的报关,不会因为附一封解释业务的说明函而被纠正。

公告还处理了衔接,而衔接规则是最容易被跳过的部分。2026 年 2 月 1 日以前审批、核准或者备案的项目,属于 2025 年版目录范围的,可以取得按新目录出具的确认书并照此办理。属于 2022 年版目录但不属于 2025 年版目录的,只要主管部门在 2027 年 2 月 1 日以前按 2022 年版目录出具了相关确认书或者回执,仍可办理——这是一个截止日,不是无限期的老项目保留。不属于 2022 年版目录但属于 2025 年版目录的在建项目,可以享受该待遇,但此前已征税进口的设备,所征税款不予退还。

第三条最容易让企业意外,因为它颠倒了”新目录更宽、公司自然被补齐”的直觉。它并不补齐。它让公司从今往后够格,把过去留在原地。

还有一项义务应当写进同一段。依免税进口的设备属于海关监管货物。在监管期限内,未经海关批准并办理相关手续,不得擅自销售、转让、调换、出租、抵押、质押、移作他用、异地使用或者退运出境;在监管年限内,减免税申请人应当于每年 6 月 30 日前向主管海关报告减免税货物使用状况。免税是附着在具体货物上的一种状态,不是一次性的省税。

用地:70% 这条底线,以及它的条件

第二项好处是用地,也是企业材料里被描述得最不准确的一项。

规则在 2017 年 1 月国务院那份关于扩大对外开放积极利用外资若干措施的通知里。该通知第十七条按顺序讲了三件事:外商投资企业与内资企业同等适用相关用地政策;继续对集约用地的鼓励类外商投资工业项目优先供应土地;在确定土地出让底价时,可按不低于所在地土地等别相对应全国工业用地出让最低价标准的 70% 执行。

这句话里压着三个条件,每一个都在起作用。

项目必须是鼓励类,也就是必须对上当时有效的那版目录里的条目。项目必须是工业项目,这就排除了服务业企业本来可能期待的一大块。项目必须是集约用地,而这是一个关于容积率、投资强度、单位土地产出的评估,不是自我描述。

这条规则产出的是底线,不是折扣。70% 是出让底价可以定到的最低比例,不是自动适用的费率,也不是一个全国统一的价格。实际的出让底价在具体地块的出让方案里由地方政府确定,要综合土地等别、产业政策与当地规划。在还没确定地块之前就按全国最低价标准的 70% 把土地写进商业计划的,写进去的是一个假设。

底线与折扣的区分还有第二层意义。因为规则被写成对地方政府的授权,而不是对投资者的权利,实务问题就不是”这条规则在不在”,而是”本地出让方案用不用它”。这个问题在投资协议的谈判里得到回答,时点是地块、投资强度与建设时序一起谈定的时候;而不是在方案已经公布之后,发一封援引该通知的函。

还有一处时序问题,从前面的内容可以推出来。用地待遇与关税待遇依赖同一个基础事实——项目对上一条鼓励类条目——但由不同机关在不同时点执行。用地问题在土地出让之前与地方政府谈定;关税问题在第一批货到港之前与海关谈定。一次把鼓励类身份确认扎实、并且留下记录的项目,同一套材料两处都用得上;把两件事当成两场独立谈判的项目,同一份分析要付两次钱。

目录不给你三样东西

一篇讲目录的文章能做的最有用的事,是把边界讲明白,因为这个领域最常见的三个错误,都来自以为目录给得比实际更多。

第一条边界是市场准入。目录不批准任何东西。外资的市场准入由《外商投资准入特别管理措施(负面清单)》决定,现行是 2024 年版,以第 23 号令发布,自 2024 年 11 月 1 日起施行,全国范围限制措施 29 条,制造业领域限制措施已经清零。负面清单限制的活动,鼓励目录里有一条也不会解除限制;负面清单没有提的,按内外资一致原则管理,本来也不需要鼓励目录来许可。两份文件回答不同的问题,之所以经常被混为一谈,是因为两者都是清单。

第二条边界是企业所得税。鼓励目录本身不带所得税税率。企业常与之挂钩的 15%,来自另一份文件、另一个范围:财政部、税务总局、国家发展改革委 2020 年关于延续西部大开发企业所得税政策的公告。该公告自 2021 年 1 月 1 日至 2030 年 12 月 31 日,对设在西部地区的鼓励类产业企业减按 15% 征收企业所得税,并把”鼓励类产业企业”定义为以《西部地区鼓励类产业目录》中规定的产业项目为主营业务、且主营业务收入占收入总额 60% 以上的企业;它还用一份列举的省区市名单界定西部地区,另列若干州、市可以比照执行。

两本目录,两个范围,一个税率。一家在湖北、对上鼓励目录地区附件某条目的公司,并不因此拿到 15%——湖北不在那份列举的西部地区名单里,而且适用的是另一份目录。一家在四川、对上《西部地区鼓励类产业目录》某条目的公司可能拿到,但要过 60% 那道测试,与它的项目是否出现在外资鼓励目录里无关。两者可以重叠,但不能互相替代。

第三条边界是外资研发中心的免税资格。这个资格存在,也有价值,但它是在另一份文件下、按另一套门槛、另一套申请材料、另一套由省级部门实施的审核程序单独认定的。已经对上鼓励类条目、又要为研发中心进口设备的公司,不应当假定目录项下的设备免税与研发中心资格是同一项好处领两次,也不应当假定一份申请覆盖两者。它们是两道门,各有各的证据。

五步:从查条目到留档

按顺序排开,这件工作并不光鲜,但是有限的。

第一步是查条目。这意味着拿现行有效的那版目录、两个附件都翻,对着项目实际要做什么去比对——不是对着公司内部的事业部名称。查的过程要留痕:考虑过哪些条目、选了哪一条、排除了哪些、为什么。只留一条选中条目的材料,是回答不了”你还考虑过什么”这个问题的材料。

第二步是对项目。条目描述的是一项活动;项目有范围、有地址、有投资总额、有产出。比对要在活动这一层做,而且要过得了地址测试——地区附件只在项目确实位于其所援引条目所属省份时才适用。一个省的项目不能援引另一个省的条目;位于地区附件覆盖省份的项目,如果全国目录某条更贴切,也仍然可以援引全国目录。

第三步是备材料。就关税优惠而言,包括《进出口货物征免税申请表》;主管部门出具的《项目确认书》,或者初始报告回执、变更报告回执,连同项目备案证明、核准批复文件或者可行性研究批复;进出口合同、发票以及相关货物的产品情况资料;以及海关认为需要提供的其他材料。材料要内部自洽:合同里的设备就是项目描述里的设备,金额落在投资总额之内,货物过得了两份不予免税目录。

第四步是走核定。条目代码在这里定下来,项目性质的记载也随之定下来。由此有两个实务点。其一,代码随目录版本走,所以一家按 2022 年版目录取得确认书、仍在进口设备的项目,应当核对手上的确认书是否仍是正确的文书,以及期限到哪一天。其二,确认之后的方案变更不是形式问题:如果项目范围、地址或者投资发生实质变化,确认书与代码可能需要在下一批货之前重新处理。

第五步是留档。免税会带来监管期限、每年 6 月 30 日前的报告义务,以及未经批准不得处置、出租、抵押、移址的限制。这些不是档案事项,而是附着在具体机器上的持续义务,并且会随着业务重组跟着设备走。把免税当成进口环节的一次性税收收益、此后不再想的公司,会在最不合适的时刻发现这项义务——也就是它想把一条产线搬到另一个厂区的时候。

五步之上有一个检验,值得单独说。材料应当让一个对这门生意一无所知的读者也能看懂。这个读者——海关关员、审计师、省级审核人员、三年后接手合规职能的继任者——要回答的只是一个问题:合同里的设备,是否对应条目里的活动,并且落在项目的金额之内。如果回答这个问题需要先听一遍公司怎么组织起来的说明,材料就没做完。条目正是用活动的语言写的,而一份照着这种语言写出来的项目描述,已经把大部分活干完了。

两个坑,以及该盯什么

第一个坑是所援引附件与所记载性质之间的错配。海关这份公告把援引全国目录的项目性质与援引地区目录的项目性质分开,而这个区分不只是描述性的:它决定适用哪个代码,而代码是系统读的东西。同一个中部省份里,援引全国目录条目的项目,与援引本省条目的项目,在案卷上是两件不同的事。两者都可能正当,但如果材料是按”地址决定答案”的假设起草的,只有一种会对得上。

第二个坑是增资项目的处理。关税规则明确把增资项目包含在内,这一点容易被忽略,因为多数团队脑子里的模型是单笔投资总额的一次性绿地投资。已设立的外商投资企业增加投资总额、为扩大后的范围进口设备,落在规则之内,但材料必须显示这次增加、修订后的总额,以及新设备与扩大后活动之间的对应关系。认定扩产在制度之外的公司在白丢一项好处;认定在制度之内却不把增资记录下来的公司,在稽查时会遇到麻烦。

坑之外,有两件事值得盯。

第一件是目录与行业开放措施的交互。2025 年修订扩大了服务业条目,而服务业恰恰是准入限制与鼓励身份最常出现在同一句话里的领域。把自己的条目读成准入许可的服务业项目会失望;把条目读成”附在已经获准从事的活动上的好处”的项目,会发现材料很顺。这个区分就是上面讲的那个区分,而且随着服务业清单变长,它会反复出现。

第二件是时点。目录自 2026 年 2 月 1 日起跑,按 2022 年版目录出具的确认书的过渡窗口在 2027 年 2 月 1 日关闭。设备计划跨过这个日期的公司——2025 年下单、2027 年到货、调试期横跨两头——正是过渡规则对他们不是纸上谈兵的那一批。对他们真正要紧的工作,是把每一批出货对着两版目录逐笔映射一遍,现在做,因为现在还有时间为对的货取得对的确认书。

目录本身是一份意图文件,而它对一家公司的价值,完全取决于这家公司能不能用这份文件的语言准确描述自己的项目。那是一件起草工作,不是一件游说工作,而且最好在签第一份合同之前做完,而不是在第一个集装箱到港之后。

来源

  • 中国政府网·《鼓励外商投资产业目录(2025 年版)》(国家发展改革委、商务部令第 37 号,2025-12-15,2026-02-01 施行):https://www.gov.cn/zhengce/zhengceku/202512/content_7052666.htm
  • 国家发展改革委·《鼓励外商投资产业目录(2025 年版)》2025 年第 37 号令(含目录全文附件):https://www.ndrc.gov.cn/xxgk/zcfb/fzggwl/202512/t20251224_1402564.html
  • 国家发展改革委·优化外资结构布局 更大力度吸引外商投资(2025-12-29,答记者问,含 1679 条与两附件条目数):https://www.ndrc.gov.cn/wsdwhfz/202512/t20251229_1402711.html
  • 海关总署·公告 2026 年第 13 号(关于执行《鼓励外商投资产业目录(2025 年版)》有关事项的公告,2026-01-27):https://www.customs.gov.cn/customs/2026-01/28/article_2026012810023759168.html
  • 中国政府网·国务院关于扩大对外开放积极利用外资若干措施的通知(国发〔2017〕5 号,第十七条用地价格):https://www.gov.cn/zhengce/content/2017-01/17/content_5160624.htm
  • 国家税务总局·关于延续西部大开发企业所得税政策的公告(财政部 税务总局 国家发展改革委公告 2020 年第 23 号):https://www.chinatax.gov.cn/chinatax/n810341/n810765/c101653/202004/c5155091/content.html
  • 中国政府网·2024 年版全国外商投资准入特别管理措施(负面清单)发布(29 条,2024-11-01 施行):https://www.gov.cn/lianbo/bumen/202409/content_6973049.htm
  • 法治网·商务部多措并举支持中部地区崛起(2026-11-16 国新办发布会,中部六省新增条目):https://www.legaldaily.com.cn/index/content/2026-11/16/content_9491852.html
  • 宁波海关·鼓励类投资项目进口减免税享惠指南(海关答疑,办理流程与后续监管):https://ningbo.customs.gov.cn/ningbo_customs/2026-04/20/article_2026063011585436091.html

A Deal Between Two Foreign Groups, Stopped by China’s Security Review: What the Control Test Turns On

  • On 4 November 2026 SoftBank received notice that the Office of the Working Mechanism for Foreign Investment Security Review in China had prohibited its acquisition of ABB’s robotics business in China, and both parties disclosed the decision without stating a reason.
    2026 年 11 月 4 日,软银收到通知,中国外商投资安全审查工作机制办公室已禁止其收购 ABB 在华机器人业务;交易双方对外披露了这一决定,但均未说明理由。
  • The reviewed act was the acquisition of a China-situated business, so a Foreign-to-Foreign Deal between a European seller and a Japanese buyer still sat inside the review, and the nationality of the seller changed nothing.
    被审查的行为是取得一家位于中国的企业,因此这笔由欧洲卖方与日本买方构成的、发生在两家外资之间的交易,仍然落在审查范围内,卖方的国籍并不改变这一点。
  • The Control Test in Article 4 of the 2020 Measures is disjunctive: a stake above 50 per cent, or voting rights that materially affect board or shareholder resolutions, or any other arrangement that gives material influence over operations, personnel, finance or technology.
    2020 年《外商投资安全审查办法》第四条的控制权测试是并列的三项:持股超过 50%,或者表决权足以对董事会、股东会决议产生重大影响,或者任何其他能对经营决策、人事、财务、技术产生重大影响的安排。
  • Antitrust clearance and Security Review Clearance are decisions under different statutes, taken by different bodies against different tests, and public reporting that the transaction had passed merger control did not move the security review.
    反垄断审查通过与否,与安全审查的通过与否,是两套法律、两个机关、两套标准下的两个决定;公开报道显示该交易此前已通过经营者集中审查,但这并没有影响安全审查的结果。
  • A prohibition is not a fine: Article 12 requires the parties to dispose of the equity or assets and restore the position before the investment, which is why the Foreign-to-Foreign Deal structure is expensive to unwind once closed.
    禁止投资不是罚款:第十二条要求当事人限期处分股权或者资产,恢复到投资实施前的状态;这正是这类外资之间交易一旦交割便极难回退的原因。
  • The sector list matters more than the price: Robotics Assets fall within major equipment manufacturing and key technology, two of the categories named in Article 4, so a mid-sized enterprise value does not lower the filing obligation.
    行业清单比交易金额更要紧:机器人资产落在重大装备制造与关键技术两类之中,两者都是第四条点名的范围,中等规模的企业估值并不会降低申报义务。
  • The statute offers a conditional-pass route and a pre-filing consultation channel, and both are more useful before signature than after a prohibition.
    办法本身设了附条件通过与申报前咨询两条通道,而这两条通道在签约之前使用,比在收到禁止决定之后使用有价值得多。

A Deal Between Two Foreign Groups, Stopped by China’s Security Review: What the Control Test Turns On | 两家外资之间的交易被中国安全审查拦下:控制权判断究竟看什么

What the two announcements actually said

The facts are short and unusually clean, because both sides published them within thirty-six hours of each other.

On 5 November 2026 ABB issued a statement headed “Update on divestment of ABB Robotics”. It recorded that in October 2025 ABB had signed an agreement to divest its Robotics division to SoftBank Group Corp. for an enterprise value of 5.375 billion US dollars, that the transaction remained subject to regulatory approvals and customary closing conditions, that the process of obtaining those approvals was managed by SoftBank, and that ABB had been informed by SoftBank that the Office of the Working Mechanism for Foreign Investment Security Review in China had prohibited the acquisition of the ABB Robotics business in China. The statement added that the two companies were in close dialogue and working constructively on the transaction.

On 6 November 2026 SoftBank published its own update. It confirmed that on 4 November 2026 it had received notification from the same office that the acquisition of ABB’s robotics business in China had been prohibited, and that it remained in close dialogue with ABB.

That is the whole of the official record. Neither statement gives a reason. Neither names a legal provision. Neither describes the scope of what was prohibited beyond the phrase “the ABB Robotics business in China”. No decision text has been published, and the review office itself has issued no announcement. A spokesperson for the foreign ministry, asked about the matter at a regular press conference on 5 November, referred the question to the competent authorities.

The deal itself is well documented from the announcement in October 2025. The robotics division had around 7,000 employees and revenue of 2.3 billion US dollars in 2024, about 7 per cent of ABB group revenue, at an operational EBITA margin of 12.1 per cent. ABB described the business as having limited commercial and technical synergy with the rest of the group. The strategic logic on the buyer’s side was stated plainly at signature: robotics as the physical layer of artificial intelligence.

The wording of both statements is worth a second look, because it is more careful than it appears. Neither company writes that the decision is final, that an appeal is available, or that the transaction is terminated. Both use the statutory verb — prohibited — and both add a sentence about continuing dialogue. That combination is consistent with the architecture of the Measures: a prohibition is one of the outcomes available at the end of a special review, and it is addressed to the parties rather than to the transaction in the abstract. The statements also observe the division of labour between the two companies: ABB’s filing says the approval process is managed by SoftBank, and SoftBank’s filing says it is managing the process. Neither company speaks for the regulator, and neither purports to summarise the reasoning.

There is also a small timing detail with a practical moral. The notification reached SoftBank on 4 November. ABB published on 5 November. SoftBank published on 6 November. Two listed groups, in two jurisdictions, disclosed within forty-eight hours of a notice that carried no reasons and no remedy. Whatever else is uncertain, the disclosure obligation is not: a prohibition is a disclosable event for both parties, and a deal team should assume that the counterparty’s announcement will appear before any agreed communications plan has been executed.

The China footprint is the part that matters for the review, and it is substantial. The robotics business has its global headquarters in Shanghai. Its Shanghai mega factory, opened in 2022, is one of three robotics manufacturing bases worldwide. More than ninety per cent of the robots the business delivers to Chinese customers are made in China. The business maintains a research and development team in China of roughly three hundred people, and China is its largest delivery market.

One boundary has to be drawn firmly at this point, because it is where most commentary has gone wrong. Reports in the financial press have attributed the decision to concerns about data from physical artificial intelligence operations being used to train models outside China, and to a wish that production capacity remain in the country. Those attributions are to people described as familiar with the matter. They are not the regulator’s stated position, they are not in either company’s statement, and no official document confirms them. They may be right. They cannot be presented as the reason. What can be said with confidence is narrower and still useful: the transaction was prohibited, the reason was not disclosed, and the parties say they are still talking.

Why a foreign-to-foreign deal still needs Chinese clearance

The first question a deal team asks about a case like this is why China is involved at all. The seller is European. The buyer is Japanese. The consideration moves from Tokyo to Zurich. No Chinese entity is buying or selling anything.

Article 2 of the Measures answers the question in a single clause. Foreign investment means a foreign investor directly or indirectly investing inside China, and it includes, at item two, a foreign investor acquiring the equity or assets of a domestic enterprise by way of merger or acquisition. The object of the review is the China-situated business, not the flow of consideration. If the perimeter of a global carve-out includes a Chinese company that owns factories, employs engineers, holds customer contracts and generates revenue in China, then the acquisition of control over that company is an investment in China, whoever signs the cheque.

The word “indirectly” in the same article closes the obvious workaround. An acquisition structured through a Singapore or Netherlands holding company is still an acquisition by a foreign investor, indirectly, of the Chinese business underneath. The review looks through the chain to the operating entity.

This is why the nationality of the incumbent owner is irrelevant, and why the phrase “the business is already foreign-owned” is not an argument. The test is not whether foreign capital is present before and after. The test is whether control over a China-situated business changes hands, and whether that business sits in a listed sector. A robotics company owned by a European industrial group and a robotics company owned by a Japanese technology group are both foreign-owned; the difference between them is who holds the control, and control is the thing the statute regulates.

There is a second structural point that distinguishes this regime from merger control and that deal teams routinely miss. The security review framework contains no turnover threshold, no market share threshold and no de minimis exemption. Scope is determined by two variables only: whether the target operates in a listed sector, and whether the acquisition confers actual control. A transaction that falls below every merger control threshold can still be a mandatory security review filing. A transaction that clears merger control comfortably can still be prohibited. The two regimes do not share a gate.

The perimeter question then becomes the most consequential drafting decision in the whole transaction, and it is usually taken early and reviewed late. Two points deserve attention at that stage.

The first is the difference between a share deal and an asset deal. Article 2 covers the acquisition of the equity or assets of a domestic enterprise, so a structure that transfers a business by moving assets rather than shares is not outside the review; it is simply described in different words. In a carve-out, the two are often combined: shares of the Chinese operating company are transferred, while intellectual property, manufacturing equipment or distribution rights move under separate agreements. All of it belongs to the same analysis.

The second is the identity of the invested enterprise. The review attaches to the entity that holds the business, not to the brand under which the business trades. Where a global division is organised as a head office, several manufacturing companies, a research company and a sales company, each of those is a candidate invested enterprise, and the filing has to describe what each one does, what it owns and what changes after closing. A team that has mapped the group structure for tax purposes has done most of this work already; a team that has not will find that the first draft of the filing is really a first draft of the structure chart.

The control test, read against the text

Article 4 is the operative provision, and its drafting repays close reading because it is the part of the framework that a deal team can actually work with before signing.

The article first lists the sectors. National defence, including military industry, military supporting industry, and investment in the vicinity of military and defence facilities, is the first category and carries no control requirement at all: any investment there is reportable. The second category is longer and does carry a control requirement. It covers important agriculture, important energy and resources, major equipment manufacturing, important infrastructure, important transport services, important cultural products and services, important information technology and internet products and services, important financial services, key technology, and other important fields, in each case where the investor acquires actual control of the invested enterprise.

Actual control is then defined in three limbs. The first is holding more than 50 per cent of the equity. The second is holding less than 50 per cent but enjoying voting rights that can materially affect resolutions of the board, the shareholders’ meeting or the shareholders’ assembly. The third is any other circumstance that enables the foreign investor to exert material influence over the enterprise’s operating decisions, personnel, finance or technology.

Three features of that definition deserve emphasis.

First, the limbs are disjunctive. The article says that actual control “includes the following circumstances”, and each limb stands alone. A 51 per cent stake is control under the first limb whatever the shareholders’ agreement says about board seats. A 30 per cent stake with a veto over the technology roadmap can be control under the third limb. There is no weighting and no safe percentage below which a deal team can stop thinking.

Second, the second and third limbs are qualitative and fact-specific. “Materially affect resolutions” and “material influence over operating decisions” are standards, not rules, and they are applied to the actual governance architecture: board composition, quorum requirements, reserved matters, the identity of the general manager, control of the budget, control of the research agenda. A structure that gives the acquirer a majority of board seats and the right to appoint the chief technology officer is not made benign by a 40 per cent economic interest.

Third, technology is named expressly in the third limb, alongside operations, personnel and finance. For an industrial robotics business, that is not a marginal addition. The value of the business sits in controllers, servo systems, machine vision, process software and the data generated on production lines. An acquirer that controls the technology roadmap has material influence over technology even if its equity is modest, and the article says so in terms.

Two further consequences follow from reading the limbs together rather than separately.

The first is that voting rights are assessed as they actually operate, not as they are nominally allocated. Entitlements that shift votes — a voting trust, an irrevocable proxy, a shareholders’ agreement under which one party instructs another how to vote, a concert party arrangement — are part of the picture, because the second limb asks what the voting rights can do, not who holds the share certificate. A structure in which the acquirer holds 35 per cent but controls the votes of a further 20 per cent is, on the face of the text, a structure in which the acquirer’s voting rights may materially affect resolutions.

The second is that actual control in this framework is not the same concept as control for accounting purposes, and the two should not be conflated in a board paper. Consolidation turns on power over the relevant activities and exposure to variable returns; the security review turns on the three limbs in Article 4, one of which is purely numerical and two of which are qualitative. A conclusion that an entity need not be consolidated is not a conclusion that no actual control has been acquired. The tests are built for different purposes and will sometimes diverge.

The counterpart of the control test is the remedy menu. Article 9 provides that where a transaction does affect national security, the office shall prohibit it, but may instead approve it with conditions where conditions can eliminate the effect on national security and the parties accept them in writing, with the conditions set out in the decision. Article 13 provides that the office, with relevant departments and local governments, supervises implementation, and may require supporting documents and conduct on-site inspection to verify compliance with the conditions.

The existence of that route is the single most practical fact in the framework. A conditional approval is a negotiated outcome, and negotiation is possible only while the file is open. A party that discovers the remedy question after a prohibition has been issued has nothing left to negotiate with.

Two gates, two standards

The most common analytical error in commentary on this case has been to treat the antitrust outcome and the security review outcome as two stages of one process, with the second a formality once the first is passed.

They are unrelated. Merger control sits under the Anti-Monopoly Law and is administered by the State Administration for Market Regulation. Its question is whether a concentration would eliminate or restrict competition: market definition, shares, concentration levels, barriers to entry, countervailing buyer power, efficiencies. Its outcomes are clearance, clearance with restrictive conditions, or prohibition, and its procedure distinguishes simple cases from ordinary ones.

Security review sits under the Foreign Investment Law and the National Security Law, and is administered by the working mechanism office established in the National Development and Reform Commission and led jointly with the Ministry of Commerce. Its question is whether an investment affects or may affect national security. Its outcomes are a decision that no review is needed, a decision that the investment does not affect national security, a conditional approval, or a prohibition. Its sector list and its control test have no counterpart in antitrust analysis.

The timetables are also separate and differently shaped. Under Article 7, the office has fifteen working days from receipt of a compliant filing to decide whether review is needed, and the parties may not implement the investment before that decision. Under Article 8, a general review must be completed within thirty working days of the decision to review, and the parties again may not implement during it. Under Article 9, a special review must be completed within sixty working days of its launch, extendable in special circumstances with written notice. Under Article 10, time taken by the parties to supply supplementary materials does not count towards any of those periods.

Because the two questions are different, they are answered from different evidence. An antitrust file is built from market data: shares, capacity, bidding records, customer switching, prices. A security review file is built from the description of the business and of what changes after closing: who the investor is, what the target does, which technologies it holds, where its data sits, who will sit on the board, and what the investor intends to do with the assets over time. A company that hands the security review to the antitrust team because the same transaction is involved will usually produce a file that is technically complete and substantively beside the point.

Public reporting indicates that the transaction had passed Chinese merger control earlier in 2026 and had also been cleared in Europe. On the record available, that is consistent with everything above: a clean antitrust file says nothing about a security review file, and the two decisions were taken by different bodies against different statutory questions.

For a deal team the operational conclusion is blunt. Build two timetables, not one. The merger control timetable will usually be the shorter and the more predictable. The security review timetable is the one that governs the long-stop date, and it contains an item — the suspension of the clock while supplementary materials are prepared — that has no equivalent in the antitrust process and that the parties control only indirectly.

What a prohibition does, and what it does not

The consequence of a prohibition is set out in Article 12 and is more severe than the word suggests.

Where the office decides to prohibit, the parties may not implement the investment; and if the investment has already been implemented, they must, within a time limit, dispose of the equity or assets and take other necessary measures to restore the position that existed before the investment was implemented and to eliminate the effect on national security.

That is a power to unscramble a completed transaction. It is not a penalty and it is not a fine; it is an order to reverse. For a carve-out of an operating division with factories, employees, customer contracts and integrated supply chains, reversing is not a matter of transferring shares back. It requires separating systems, re-papering contracts, re-establishing a standalone management structure and, in practice, absorbing the transaction costs twice.

The framework contains three further provisions that a deal team should read together with Article 12, because they define the cost of getting scope wrong.

Article 16 addresses failure to file. For an investment within the reporting scope, a party that implements the investment without filing is ordered to file within a time limit; if it refuses, it is ordered to dispose of the equity or assets and restore the pre-investment position. The exposure therefore does not depend on the review reaching a negative conclusion. It attaches to the act of closing without filing.

Article 14 addresses change of plan. Where the office has decided that no review is needed or has approved an investment, and the parties then change the investment plan in a way that affects or may affect national security, they must file again. Approvals are attached to a described transaction, not to a relationship.

Article 19 addresses the record. A party in any of those situations has the matter entered as a bad credit record in the national credit information system and is subject to joint sanctions under relevant rules.

One further feature of that design deserves separate mention, because it is where the practical damage sits. The order runs against the parties to the investment. The consequences do not stop there. A carve-out of an operating division is usually financed, and the financing documents will contain representations about regulatory approvals; a restoration order turns those representations into breaches. The division’s customers will have been notified of the change of control and will have consented, or will have terminated and re-signed; a restoration order requires the exercise to be run again, in the opposite direction, with a counterparty that now knows the transaction failed once. Employees will have been transferred or re-papered. None of those parties is a party to the review, and none of them is protected by it.

That is the reason the analysis has to be done before signature rather than after. By the time an order is issued, the cost is not the legal cost of compliance. It is the cost of undoing arrangements with banks, customers and employees who were told, in good faith, that the transaction had cleared every gate that applied to it.

Set against that structure, the calculus for a mid-market cross-border carve-out is straightforward. The cost of filing is measured in months and in professional fees. The cost of not filing, if the transaction turns out to be in scope, is measured in unwinding a closed deal. The asymmetry is large enough that the only defensible default is to file where the sector and control tests are arguably met.

Three things to do before signature

The useful output of this case for a deal team is not an opinion about the outcome. It is a set of steps that change the shape of the risk before the agreement is signed.

The first is a security review feasibility assessment and a filing-path design, completed before signature rather than after. That means identifying, entity by entity, what sits inside the transaction perimeter and where it is organised; testing each China entity against the sector list; and testing the post-closing governance architecture against the three control limbs, including reserved matters, board composition and technology decision rights. It also means using the consultation channel in Article 5, which allows a party to consult the office on relevant questions before filing. That channel is in the statute, it is expressly available pre-filing, and it is under-used relative to its value, because a consultation converts an open question about scope into a documented position that the board can rely on.

The second is to make security review a named closing condition, separate from merger control and separate from any other regulatory approval. The drafting points that matter are unglamorous and decisive. The condition should name the review expressly rather than sitting inside a general “regulatory approvals” definition, so that the allocation of risk is unambiguous. The long-stop date should be built from the statutory clock — fifteen plus thirty plus sixty working days, plus an allowance for extension, plus an allowance for the period during which the clock is suspended while supplementary materials are prepared — rather than from a commercial estimate. And the agreement should say in advance who bears what if the outcome is a prohibition: who pays the costs, whether a reverse termination fee applies, and how a remedy negotiation is funded and controlled.

The third is to assess the indivisibility of the China assets before deciding the perimeter. This is the question that decides whether the transaction is a China control transaction in substance.

If the China manufacturing base, the China research team and the China customer relationships can be separated from the rest of the global business without destroying its value, the parties have a genuine choice: exclude China from the perimeter and proceed with the remainder, or include it and accept the review risk. If they cannot be separated — because the product is designed in one place and made in another, because the supply chain is integrated, because the customers are served from a single global platform — then the transaction is, in economic substance, an acquisition of control over the China business, and the review outcome is binary rather than incremental.

Where the answer is indivisibility, the preparation that pays is a remedy package designed in advance: commitments on production capacity, on supply to the domestic market, on continuity of research and development, on data handling, and on the standing of the local management team. Article 9 allows a conditional approval where conditions can eliminate the effect on national security and the parties accept them in writing. A package that has been thought through, costed and cleared internally before filing is a different instrument from a package improvised in the final weeks of a special review.

Underneath those three steps sits a document that costs little and is worth a great deal: a written scope memorandum, dated before signature, recording the sector analysis, the control analysis and the perimeter decision, with the reasons for each. Its purpose is not to persuade a regulator. Its purpose is to be produced, years later, to a board, an auditor, a lender or a court, in answer to the question of whether the company considered the filing obligation at the time. Companies that file have the memorandum and do not need it. Companies that decide not to file need it and often do not have it.

One caution applies to all three steps. They are preparatory, and preparation is not a substitute for the local conversation. The office operates through a process that involves relevant departments and, where appropriate, local government, and the practical questions a deal team will have — whether a particular description of the business is adequate, whether a particular change of governance is material, what a remedy package would need to contain — are answered in that process rather than in the text. The three steps are what allows a company to have that conversation usefully instead of arriving at it with a structure that cannot be changed.

The open questions

Three things remain unknown, and it is worth being explicit about them rather than filling the gaps with inference.

The reason for the prohibition is not public. The Measures do not require the office to publish decisions or reasons, and no decision text has appeared. Any account of the reasoning is therefore an account of someone’s understanding, not of the record. That is unsatisfying for advisers and it is a feature of the regime rather than an oversight: a review system built around national security will not, in any jurisdiction, publish the analysis that supports a prohibition.

Whether the parties can revive the transaction is also unknown. Article 11 allows a party to modify the investment plan during review, with the clock restarting from receipt of the modified plan, and to withdraw, in which case the review terminates. Both mechanisms operate inside an open file. After a prohibition, the route back would be a new filing on a materially different transaction, and neither company has said it intends to take it. Their statements say only that dialogue continues.

What a conditional approval would look like is the third gap, and it is the one that matters most for planning. The Measures say only that conditions must be capable of eliminating the effect on national security, must be accepted in writing by the parties, and must be listed in the decision; Article 13 adds that compliance is supervised and can be verified by requiring documents and conducting on-site inspection. Nothing in the text limits the subject matter of a condition. That breadth is deliberate — a condition has to fit the transaction — and it means the working assumption for a deal team should be that anything capable of addressing the specific concern is on the table, and that the parties are expected to propose it rather than wait to be told.

And the significance of the case is not yet established. This is reported to be the second prohibition under the current framework in 2026; the earlier matter, in April, is reported to have involved a requirement to restore the pre-closing position after completion. Two data points do not make a pattern, and neither decision has been explained. What can be said is that the framework has now been used at the prohibition end of its range twice in a single year, which is a different fact from the framework merely existing.

For a company planning a cross-border carve-out with Chinese operations, the practical reading is narrower than the headlines and more useful. The review is a scope question and a schedule question. Both are answerable before signature, both are answerable with the statute in hand rather than with market rumour, and the answers determine whether the transaction should be signed at all, on what conditions, and with what walk-away rights.

Sources

  • ABB·Update on divestment of ABB Robotics(2026-11-05 声明原文):https://resources.news.e.abb.com/attachments/published/140154/en-US/0A15DBC92198/20261105_Update_on_divestment_of_ABB_Robotics_EN.pdf
  • 软银集团·Update on Acquisition of ABB Ltd’s Robotics Business(2026-11-06):https://group.softbank/en/news/press/20261106
  • ABB·ABB 与软银签约拟出售机器人业务(2025-10-08,交易企业估值 53.75 亿美元):https://new.abb.com/news/zh-CHS/detail/129790/abb-to-divest-robotics-division-to-softbank-group
  • 中国政府网·《外商投资安全审查办法》(发展改革委、商务部令第 37 号,2020-12-19 公布,2021-01-18 施行):https://www.gov.cn/gongbao/content/2021/content_5582626.htm
  • 商务部·国家发展改革委、商务部发布《外商投资安全审查办法》(2020-12-19):https://www.mofcom.gov.cn/xwfb/rcxwfb/art/2020/art_593675063f2d45af8c8ab673f2836ef2.html
  • 澎湃新闻·日本软银以 54 亿美元收购 ABB 机器人的计划受挫(2026-11-05,媒体转述,非监管机关披露):https://www.thepaper.cn/newsDetail_forward_34413380

两家外资之间的交易被中国安全审查拦下:控制权判断究竟看什么

两份声明究竟说了什么

事实很短,而且罕见地干净,因为交易双方在三十六小时之内各自公开了同一件事。

2026 年 11 月 5 日,ABB 发布题为《ABB 机器人业务出售进展更新》的声明。声明记载:2025 年 10 月,ABB 已签署协议,将其机器人业务出售给软银集团,企业估值 53.75 亿美元;该交易尚需获得监管机构批准并满足惯例成交条件;监管审批流程由软银负责推进;ABB 已接到软银通报,中国外商投资安全审查工作机制办公室已禁止软银收购 ABB 机器人在华业务。声明最后一句是,双方正保持密切沟通,并就该交易开展建设性磋商。

2026 年 11 月 6 日,软银发布自己的更新公告,确认已于 2026 年 11 月 4 日收到同一办公室的通知,其收购 ABB 机器人业务在华部分已被禁止,并表示与 ABB 保持密切沟通。

官方的全部记录就是这些。两份声明都没有给出理由,都没有援引具体条款,对”被禁止的是什么”的描述也只到”ABB 机器人在华业务”这一短语为止。没有任何决定文本公开,审查机关本身也没有发布消息。11 月 5 日外交部例行记者会上,发言人被问及此事,答复是建议向中方主管部门询问。

交易本身在 2025 年 10 月的公告里记录得很完整。机器人业务约有七千名员工,2024 年销售收入 23 亿美元,约占 ABB 集团总收入的 7%,运营息税摊销前利润率 12.1%。ABB 对该业务的描述是,与集团其他业务单元之间的商业与技术协同有限。买方在签约时把战略逻辑讲得很直白:机器人是人工智能的物理层。

两份声明的措辞值得再看一眼,因为它们比表面上更谨慎。两家公司都没有写这个决定是终局的,没有写存在何种救济途径,也没有写交易已经终止。两家都用了条文里的那个动词——禁止——并且都补了一句沟通仍在继续。这个组合与办法的架构是吻合的:禁止是特别审查走到终点时可以作出的一种结果,而它是向当事人作出的,不是向抽象的交易作出的。两份声明还恰好印证了两家公司的分工:ABB 的声明写监管审批流程由软银负责推进,软银的公告写自己正在管理这一流程。两家都没有代表监管机关说话,也都没有试图概括理由。

还有一处时间细节,带出的教训很实用。通知在 11 月 4 日到达软银,ABB 在 11 月 5 日发布,软银在 11 月 6 日发布。两家在两个法域上市的公司,在一份既无理由也无救济的通知到达后四十八小时内完成了披露。无论其他事情多么不确定,披露义务本身是确定的:禁止对双方都是应当披露的事件,交易团队应当假定对方的公告会早于任何商定的沟通方案落地。

对审查而言真正要紧的是中国部分的体量,而它相当可观。该业务的全球总部设在上海;2022 年投产的上海超级工厂,是集团全球三大机器人制造基地之一;交付中国客户的机器人,九成以上在中国本地生产;在华研发团队近三百人;中国是其最大的交付市场。

这里必须划一条硬边界,因为绝大多数评论都在这一点上出了偏差。财经媒体把这一决定归因于对物理人工智能业务产生的数据被用于在境外训练模型的顾虑,以及希望产能留在中国境内。这些归因都来自”知情人士”。它们不是监管机关的表态,不在任何一家公司的声明里,也没有任何官方文件予以确认。它们可能是对的,但不能被当作理由来陈述。可以确定地说的要窄得多,但仍然有用:交易被禁止,理由未披露,双方称仍在沟通。

为什么两家外资之间的交易仍要中国批准

面对这样的案子,交易团队的第一个问题总是:中国为什么会介入。卖方是欧洲企业,买方是日本企业,对价从东京流向苏黎世,没有任何中国主体在买或在卖。

第二条一个条款就把这个问题回答了。该条把外商投资定义为外国投资者直接或者间接在境内进行的投资活动,并在第二项明确列出:外国投资者通过并购方式取得境内企业的股权或者资产。审查的对象是位于中国的企业,不是对价的流向。如果一项全球剥离的交易边界内包含一家中国公司,而这家公司拥有工厂、雇有工程师、持有客户合同、在中国产生收入,那么取得对这家公司的控制权,就是对中国的投资——不论支票由谁签。

同一条里的”间接”二字,堵住了最明显的绕道设计。通过新加坡或者荷兰控股公司实施的收购,仍然是外国投资者间接取得下层中国业务的收购。审查会穿透股权链条,看向实际经营的实体。

这也解释了为什么原持有人的国籍无关紧要,以及为什么”这块业务本来就是外资的”不构成理由。判断的不是交易前后是否存在外资,而是对一家位于中国的企业的控制权是否易手,以及这家企业是否落在清单所列行业之内。一家由欧洲工业集团持有的机器人公司,和一家由日本科技集团持有的机器人公司,都是外资;两者的区别在于谁持有控制权,而控制权正是这套办法规范的对象。

还有一处结构性差别,是交易团队经常忽略的:安全审查框架里没有营业额门槛、没有市场份额门槛、也没有微量豁免。范围只由两个变量决定——标的是否经营清单所列行业,以及收购是否取得实际控制权。一笔远低于任何经营者集中申报标准的交易,仍可能是强制性的安全审查申报;一笔顺利通过经营者集中审查的交易,仍可能被禁止。两套制度不共用一个闸门。

于是,交易边界的划定就成了整笔交易里最有分量的起草决定,而它通常在早期被草率定下、在后期才被重新审视。这个阶段有两点值得盯住。

第一点是股权交易与资产交易的区别。第二条覆盖的是取得境内企业的股权或者资产,因此以转移资产而非转移股份的方式剥离一项业务,并不在审查之外,只是被描述成了另一套词。在业务剥离中,两者常常并用:中国运营公司的股份被转让,而知识产权、生产设备或者经销权则依据另外的协议转移。它们属于同一个分析。

第二点是被投资企业的识别。审查附着于持有业务的那个实体,而不附着于业务所使用的品牌。当一项全球业务被组织成总部、若干制造公司、一家研发公司和一家销售公司时,每一个都是潜在的被投资企业,申报材料必须逐一说明它做什么、拥有什么、交割后发生什么变化。为税务目的梳理过集团架构的团队,这项工作已经完成大半;没有梳理过的团队会发现,申报材料的第一稿,其实就是架构图的第一稿。

控制权测试:回到条文本身

第四条是操作条款,它的起草方式值得逐字读,因为这是框架中交易团队在签约前真正能用的部分。

该条先列行业。第一类是国防,包括军工、军工配套,以及在军事设施和军工设施周边地域投资;这一类不附加控制权要求,任何投资都需申报。第二类更长,且附有控制权要求,涵盖关系国家安全的重要农产品、重要能源和资源、重大装备制造、重要基础设施、重要运输服务、重要文化产品与服务、重要信息技术和互联网产品与服务、重要金融服务、关键技术以及其他重要领域,且以取得所投资企业的实际控制权为条件。

实际控制权随后由三项情形界定。第一项是持有企业 50% 以上股权。第二项是持股不足 50%,但享有的表决权能够对董事会、股东会或者股东大会的决议产生重大影响。第三项是其他导致外国投资者能够对企业的经营决策、人事、财务、技术产生重大影响的情形。

这个定义有三处需要强调。

第一,三项是并列的。条文说实际控制权”包括下列情形”,每一项各自独立成立。持股 51% 就是第一项意义上的控制,无论股东协议如何约定董事席位;持股 30% 但对技术路线拥有否决权,可能构成第三项意义上的控制。这里没有权重,也没有一个可以让人停止思考的安全比例。

第二,第二项与第三项是定性判断,取决于事实。”产生重大影响”与”能够产生重大影响”是标准,不是规则,适用的对象是实际的治理架构:董事会构成、表决机制与出席比例、保留事项、总经理人选、预算控制权、研发议程控制权。一个让收购方取得多数董事席位并有权任命首席技术官的结构,不会因为经济权益只有四成而变得无害。

第三,技术被明确写进第三项,与经营决策、人事、财务并列。对一家工业机器人企业来说,这不是可有可无的补充。这项业务的价值在于控制器、伺服系统、机器视觉、工艺软件,以及产线上生成的数据。控制了技术路线图的收购方,即便股权比例不高,也对技术有重大影响——条文就是这么写的。

把三项放在一起读,而不分开读,还能得出两个推论。

第一个推论是,表决权按它实际如何运作来评价,而不是按名义上如何分配。能够移动投票权的安排——表决权信托、不可撤销的委托、约定一方按另一方指示投票的股东协议、一致行动安排——都属于判断范围,因为第二项问的是表决权能起什么作用,而不是股票凭证在谁手上。收购方持股三成五、同时控制另外两成的投票权,从条文表面看,就是一个其表决权可能对决议产生重大影响的结构。

第二个推论是,这套框架里的实际控制权,与会计意义上的控制不是同一个概念,两者不应在董事会材料里混用。合并报表看的是对被投资方相关活动的权力与可变回报的风险敞口;安全审查看的是第四条的三项情形,其中一项纯属数字,另两项是定性判断。得出”不必合并”的结论,并不等于得出”没有取得实际控制权”的结论。两套测试服务于不同目的,有时会分道扬镳。

与控制权测试相对应的,是救济菜单。第九条规定,申报的交易影响国家安全的,应当作出禁止投资的决定;但通过附加条件能够消除对国家安全的影响,且当事人书面承诺接受附加条件的,可以作出附条件通过的决定,并在决定中列明附加条件。第十三条规定,审查决定由工作机制办公室会同有关部门、地方人民政府监督实施;对附条件通过的,可以要求提供证明材料、现场检查等方式核实附加条件的实施情况。

这条通道的存在,是整套框架里最实用的一个事实。附条件通过是一个谈出来的结果,而谈判只在案卷还开着的时候可能。等禁止决定已经作出,才发现救济方案还没准备的人,手上已经没有可交换的东西。

两个闸门,两套标准

围绕本案最常见的分析错误,是把反垄断结果与安全审查结果当成同一流程的两个阶段,并认为前者通过之后,后者只是形式。

两者并不相关。经营者集中审查依据反垄断法,由国家市场监督管理总局主管,它问的是这项集中是否会排除、限制竞争:相关市场界定、份额、集中度、进入壁垒、买方力量、效率抗辩。它的结果是通过、附加限制性条件通过或者禁止,程序上区分简易案件与普通案件。

安全审查依据外商投资法与国家安全法,由设在国家发展改革委、由国家发展改革委与商务部牵头的工作机制办公室主管。它问的是这项投资是否影响或者可能影响国家安全。它的结果是不需要审查、通过、附条件通过或者禁止。它的行业清单与控制权测试,在反垄断分析里没有对应物。

时间表也是分开的,形状还不一样。按第七条,办公室自收到符合规定的材料之日起十五个工作日内,作出是否需要进行安全审查的决定,作出决定前当事人不得实施投资。按第八条,决定审查的,应当自决定之日起三十个工作日内完成一般审查,审查期间同样不得实施投资。按第九条,启动特别审查的,应当自启动之日起六十个工作日内完成,特殊情况下可以延长,延长须书面通知。按第十条,当事人补充提供材料的时间,不计入上述审查期限。

因为两个问题不同,支撑它们的证据也不同。反垄断案卷由市场数据搭建:份额、产能、投标记录、客户转换、价格。安全审查案卷由业务描述与交割后变化的描述搭建:投资者是谁,标的是做什么的,持有哪些技术,数据放在哪里,谁将进入董事会,以及投资者打算在更长的时间里如何处置这些资产。一家公司如果因为”是同一笔交易”就把安全审查交给反垄断团队去办,通常交出的会是一份技术上完整、实质上跑偏的材料。

公开报道显示,该交易已于 2026 年早些时候通过中国经营者集中审查,同年亦获欧盟放行。就已公开的记录看,这与上述分析完全一致:一份干净的反垄断案卷,对安全审查案卷什么也说明不了;两个决定由不同机关、针对不同法律问题分别作出。

对交易团队来说,操作层面的结论很直接:排两张时间表,而不是一张。经营者集中那张通常更短、更可预测;安全审查那张才是决定最后交割日的,而且里面有一项在反垄断流程中没有对应物的安排——准备补充材料期间时钟暂停,而暂停多久,当事人只能间接控制。

禁止决定做了什么,没做什么

禁止的后果规定在第十二条,比这个词听起来更重。

办公室作出禁止投资决定的,当事人不得实施投资;已经实施的,应当限期处分股权或者资产以及采取其他必要措施,恢复到投资实施前的状态,消除对国家安全的影响。

这是一项把已完成的交易拆回去的权力。它不是处罚,不是罚款,而是恢复原状的命令。对一项包含工厂、员工、客户合同与一体化供应链的业务剥离来说,恢复原状不是把股份转回去那么简单。它要求拆分系统、重签合同、重建独立的管理架构,并且在实践中把交易成本付两遍。

框架里还有三条,应当与第十二条放在一起读,因为它们界定了”范围判断错了”的代价。

第十六条管的是未申报。属于申报范围的投资,当事人未申报即实施的,责令限期申报;拒不申报的,责令限期处分股权或者资产,恢复到投资实施前的状态。也就是说,风险并不取决于审查最终得出否定结论,而附着于”没申报就交割”这个动作本身。

第十四条管的是方案变更。办公室作出不需要审查或者通过审查的决定后,当事人变更投资方案,影响或者可能影响国家安全的,应当重新申报。批准附于一项被描述过的交易,而不附于一段关系。

第十九条管的是记录。出现第十六条、第十七条、第十八条情形的,应当作为不良信用记录纳入国家有关信用信息系统,并按规定实施联合惩戒。

这套设计里还有一处值得单独提,因为实际损害恰恰落在那里。命令是针对投资当事人的,后果却不止于当事人。一项经营性业务的剥离通常伴随融资,融资文件里会有关于监管批准的陈述与保证;恢复原状的命令会把那些陈述变成违约。被剥离业务的客户此前已被告知控制权变更,并且已经同意,或者已经解约重签;恢复原状意味着整个流程要反向再做一遍,而对手方此时已经知道这笔交易失败过一次。员工也已经被转移或者重签合同。这些主体都不是审查的当事人,也都不在审查的保护范围内。

这就是分析必须在签约前完成、而不是签约后补做的原因。等到命令作出,代价已经不是履行命令的法律成本,而是向银行、客户与员工——他们曾被告知这笔交易已通过所有适用于它的闸门——收回安排的成本。

把这些放在一起,一笔中型跨境剥离的账其实很好算。申报的成本以月和专业费用计;该申报而未申报、且事后被认定属于范围的成本,以拆解一笔已交割交易计。这个不对称足够大,以至于唯一站得住脚的默认做法是:只要行业与控制权两项测试存在可争辩的成立空间,就申报。

签约之前该做的三件事

这个案子对交易团队真正有用的产出,不是对结果的评论,而是一组能在协议签署前改变风险形状的动作。

第一件,是在签约前完成安全审查可行性评估与申报路径设计,而不是在签约后。这意味着逐主体梳理交易边界内有什么、注册在哪里;把每一家中国主体对照行业清单过一遍;把交割后的治理架构对照三项控制权情形过一遍,包括保留事项、董事会构成与技术决策权。这也意味着用上第五条给出的咨询通道——当事人向工作机制办公室申报前,可以就有关问题向其咨询。这条通道写在办法里,明确可用于申报前,而相对于它的价值,用的人偏少。一次咨询,能把一个关于范围的开放问题,转成董事会可以依赖的有记录立场。

第二件,是把安全审查写成一项具名的交割先决条件,与经营者集中、与其他任何监管批准分开列明。真正起作用的起草细节都不起眼,却具有决定性。条件应当直接点名安全审查,而不是塞进一个笼统的”监管批准”定义里,这样风险分配才不含糊。最后交割日应当按法定时钟搭建——十五加三十加六十个工作日,加上延长的余量,加上准备补充材料期间时钟暂停的余量——而不是按商业估算。协议还应当事先写清:如果结果是禁止,谁承担什么,费用由谁支付,是否适用反向终止费,救济方案的谈判由谁主导、如何出资。

第三件,是在确定交易边界之前,先评估中国资产的不可分割性。这个问题决定这笔交易在实质上是不是一笔中国控制权交易。

如果中国的制造基地、中国的研发团队与中国的客户关系,可以从全球业务中分离出去而不摧毁其价值,双方就有真正的选择:把中国排除在边界之外、推进其余部分,或者把中国包进来、承担审查风险。如果分不开——因为产品在一地设计、在另一地制造,因为供应链是一体的,因为客户由同一个全球平台服务——那么这笔交易在经济实质上就是取得对中国业务的控制权,审查结果是非此即彼,而不是可以逐步退让的。

答案若是”分不开”,真正划算的准备是提前设计一套救济方案:关于产能的承诺、关于境内市场供应的承诺、关于研发连续性的承诺、关于数据处理的承诺、关于本地管理团队地位的承诺。第九条允许在附加条件能够消除对国家安全影响、且当事人书面承诺接受的情况下作出附条件通过。一套在申报前已经想清楚、算过成本、内部通过的方案,与一套在特别审查最后几周临时拼出来的方案,是完全不同的两件工具。

这三件事之下,还有一份成本很低、价值很高的文件:一份落款在签约之前的范围备忘,记录行业分析、控制权分析与边界决定,并写明各自的理由。它的用途不是说服监管机关,而是在若干年之后,被交给董事会、审计师、贷款人或者法院,用来回答”公司在当时是否考虑过申报义务”这个问题。申报的公司有这份备忘,但用不上;决定不申报的公司用得上,却往往没有。

三件事都有一条共同的注意点。它们是准备工作,而准备不能替代当面的沟通。办公室运行的流程会涉及有关部门,并在适当情形下涉及地方政府;交易团队真正会有的那些问题——对业务的某一种描述是否足够、治理架构的某一处变化是否重大、一套救济方案需要包含什么——是在这个流程里得到回答的,而不是在条文里。这三件事的意义,是让公司能够有效地进行那场对话,而不是带着一个已经改不动的结构走进对话。

尚未回答的问题

有三件事仍属未知,把它们说清楚,比用推断去填补空白更有价值。

禁止的理由不公开。办法并未要求办公室公布决定或理由,也没有任何决定文本出现。因此,任何关于推理过程的叙述,都是某个人理解的叙述,而不是记录的叙述。这对顾问来说不舒服,但它是这套制度的一个特征,而不是疏漏:任何法域里,围绕国家安全搭建的审查体系都不会公布支撑禁止决定的分析。

双方能否让这笔交易复活,同样未知。第十一条允许当事人在审查期间修改投资方案,审查期限自收到修改后方案之日起重新计算;也允许撤销投资,撤销则审查终止。两种机制都发生在案卷还开着的时候。禁止决定之后,回来的路只能是以一项实质不同的交易重新申报,而两家公司都没有表示打算这么做。它们的声明只说沟通在继续。

第三处空白,是附条件通过会长成什么样,而它对规划最要紧。办法只说了三件事:附加条件要能够消除对国家安全的影响,要经当事人书面承诺接受,要在决定中列明;第十三条又补充,附加条件由办公室会同有关部门、地方人民政府监督实施,可以通过要求提供证明材料、现场检查的方式核实。条文没有对附加条件的题材作任何限制。这种宽是有意的——条件必须贴合具体的交易——它意味着交易团队的作业假设应当是:凡是能够回应那项具体关切的手段,都摆在桌上,而且当事人被期待主动提出,而不是等着被告知。

这个案子的意义也还没有定型。据报道,这是 2026 年现行框架下的第二起禁止;此前 4 月那一起,据报道涉及在交割后被要求恢复原状。两个数据点不构成规律,两个决定都没有被解释。可以说的只是:这套框架在一年之内,两次被用到了它区间的禁止一端——这与”框架存在”本身,是两个不同的事实。

对正在筹划带中国业务的跨境剥离的企业来说,务实的读法比标题窄,也比标题有用。审查是一个范围问题加一个时间表问题。两者都能在签约前回答,都能拿着条文而不是拿着市场传闻回答;而答案决定这笔交易该不该签、在什么条件下签、以及退出权怎么写。

来源

  • ABB·Update on divestment of ABB Robotics(2026-11-05 声明原文):https://resources.news.e.abb.com/attachments/published/140154/en-US/0A15DBC92198/20261105_Update_on_divestment_of_ABB_Robotics_EN.pdf
  • 软银集团·Update on Acquisition of ABB Ltd’s Robotics Business(2026-11-06):https://group.softbank/en/news/press/20261106
  • ABB·ABB 与软银签约拟出售机器人业务(2025-10-08,交易企业估值 53.75 亿美元):https://new.abb.com/news/zh-CHS/detail/129790/abb-to-divest-robotics-division-to-softbank-group
  • 中国政府网·《外商投资安全审查办法》(发展改革委、商务部令第 37 号,2020-12-19 公布,2021-01-18 施行):https://www.gov.cn/gongbao/content/2021/content_5582626.htm
  • 商务部·国家发展改革委、商务部发布《外商投资安全审查办法》(2020-12-19):https://www.mofcom.gov.cn/xwfb/rcxwfb/art/2020/art_593675063f2d45af8c8ab673f2836ef2.html
  • 澎湃新闻·日本软银以 54 亿美元收购 ABB 机器人的计划受挫(2026-11-05,媒体转述,非监管机关披露):https://www.thepaper.cn/newsDetail_forward_34413380

The APEC Card in China After the Ninety-Day Stay: What It Replaces and What It Cannot

  • China raised the maximum stay for holders of the APEC Card from 60 to 90 days with effect from 1 November 2026, a change of one parameter that this article takes as given rather than re-explains.
    中国自 2026 年 11 月 1 日起将亚太经合组织商务旅行卡持卡人在华最长停留期由 60 天延长至 90 天;这是一个单参数的变化,本文把它当作既定前提,不再重复解释。
  • The Ninety-Day Stay ceiling applies to a single entry and the announcement says nothing about an annual cumulative limit, so the operational question is how a stay is counted rather than how many trips are allowed.
    90 天是单次入境的上限,公告对年度累计停留未作任何规定;因此操作上的问题是怎么计算一次停留,而不是一年能来几次。
  • Whose card it is matters more than the ceiling: Chinese nationals apply through local foreign affairs offices under five eligibility categories, while foreign executives hold cards issued by their own economies.
    卡是谁的,比上限更重要:中国公民按五类申请对象通过地方外事部门申办,而外国高管持有的是其所属经济体签发的卡。
  • The card is a Business Visit facilitation and not a work authorisation, and the Work Permit Boundary is tested by the substance of the relationship rather than by the length of the stay.
    这张卡是商务访问便利,不是工作许可;工作许可的边界看的是关系的实质,而不是停留的长短。
  • A card carries its own validity and is used together with the passport presented at application, so a renewed passport creates an administrative task that a longer stay allowance does not solve.
    卡本身有独立有效期,并与申办时所持护照配套使用;护照换发会产生一项行政事务,而停留期延长解决不了这件事。
  • Unilateral visa-free entry, 240-hour visa-free transit and the card look interchangeable but differ in permitted activity, maximum stay and whether an application must be made in advance.
    单方面免签、240 小时过境免签与这张卡看上去可以互换,但在许可活动、最长停留和是否须提前申请上各不相同。
  • As of November 2026 China applied unilateral visa-free entry to 50 countries, full mutual visa exemption to 29 and 240-hour visa-free transit to 57, according to figures released by exit-entry authorities.
    据边检机关公开发布,截至 2026 年 11 月,中国对 50 个国家实施单方面免签,与 29 个国家全面互免签证,240 小时过境免签适用国家为 57 个。
  • The costliest mistake is to treat the extra thirty days as a probation period for a posting, because a business visit that becomes employment in substance is an immigration problem and a payroll problem at the same time.
    代价最高的错误,是把多出的 30 天当成某个岗位的试用期;一旦商务访问在实质上变成雇佣,它同时是出入境问题和薪酬合规问题。

The APEC Card in China After the Ninety-Day Stay: What It Replaces and What It Cannot | 亚太经合组织商务旅行卡在华停留延至 90 天后:能替代什么、替代不了什么

One parameter, taken as given

The policy fact is short enough to state in a sentence and this article will not spend more space on it. From 1 November 2026, a holder of the APEC Business Travel Card may stay in China for up to 90 days per entry, up from 60. The measure was announced by the Chinese side in late October 2026 as a way of keeping business travel across the Asia-Pacific region moving, and it applies to holders of cards issued by any participating economy.

Everything else about the card is unchanged. The eligibility rules, the issuing route, the economies that participate, the dedicated channels at ports, and the fact that the card is a travel facilitation rather than a work authorisation all sit exactly where they were on 31 October 2026. The policy itself has been described elsewhere; what has not been described is what a holder or an employer does differently on 2 November, and that is the whole subject here.

That framing matters because a change of one number invites two opposite errors. The first is to treat it as nothing: thirty days is a rounding error against a long assignment, so the calendar stays as it was and the benefit is never collected. The second is to treat it as everything: the ceiling has moved, therefore the card now covers what it previously did not, therefore the work permit conversation can be deferred. Both errors are common, and they fail in different ways. The first wastes an allowance that was granted. The second converts a facilitation into an exposure.

The right reading sits between them. The card has become more useful for a defined class of trips and no more useful for any other class. Finding out which class a given trip belongs to is a matter of arithmetic, documentation and one judgement about the nature of the activity — and it is worth doing before the next travel calendar is fixed, because calendars are usually fixed once a year and then defended.

There is a reason the number 90 has more than arithmetic significance. Sixty days sits awkwardly against a business quarter: it covers two months of a three-month cycle, so a stay that follows a quarterly rhythm either ends early or overruns. Ninety days covers a quarter with room at both ends. For project work organised around quarterly milestones — a commissioning phase, a budget cycle, an audit season, a product launch window — the new ceiling is the first one that fits the shape of the work rather than the shape of the immigration rule. That is the substantive gain, and it is not visible in the announcement, which simply states the new maximum.

Counting a stay: what the ceiling caps and what it does not

The first thing to get right is the unit. The 90 days is a ceiling on a single stay, counted from the day of entry. It is not an annual quota, it is not a monthly allowance and it is not a running balance that a holder can top up by leaving and returning. A holder who enters on 1 March and leaves on 29 May has used 90 days in one stay; a holder who enters on 1 March, leaves on 31 March and re-enters on 1 April has used two stays of 31 and, on the same arithmetic, up to 90 days each.

The second thing is what the announcement does not say. It says nothing about an annual cumulative limit, and it says nothing about a cooling-off period between stays. That silence is not an invitation. The ceiling on a single stay is one control; the assessment made at the border is another; and the purpose of the visit is a third. A pattern of back-to-back stays that each sit just inside the ceiling will be read as what it is, and the reading will not be flattering.

The third thing is that a ceiling is not an entitlement. A 90-day maximum means that the maximum permissible stay is 90 days where the entry is granted on that basis; it does not mean that every entry produces 90 days. The length actually granted is decided at the port, and it depends on the purpose stated, the supporting documents produced and the officer’s assessment. A holder whose invitation letter covers two weeks and whose return ticket is dated in eleven weeks is asking to be asked a question.

For an employer, the arithmetic has a practical shape that is easy to formalise. Track days per entry, not days per year. Record the entry date and the exit date for each trip, because the exit date is what closes the count. And keep the count somewhere that a travel manager, an HR function and a finance function can all see, because the three of them will otherwise each keep a different one.

There is a fourth point, less obvious and more consequential. Because the ceiling applies per entry, a trip that would previously have needed a mid-assignment exit can now be run as one stay — but only if the itinerary is designed that way from the start. An itinerary built around a 60-day ceiling contains a border run at day 55 that nobody questions, because it was always there. Removing it requires someone to decide that it is no longer needed, and that decision is not made by the policy; it is made by whoever owns the calendar.

A worked example makes the difference visible. Take a supplier qualification programme with four stages: a document review at the supplier’s office, a two-week line observation, a corrective-action discussion, and a follow-up verification six weeks later. Under a 60-day ceiling, the natural plan is one trip covering the first three stages and a second trip for the follow-up, with a gap in the middle that the supplier fills as it chooses. Under a 90-day ceiling, the same four stages can sit in a single stay, with the follow-up done by the same person who raised the findings. The saving is not the airfare. It is the loss of context that happens when a finding is handed over to someone who was not in the room, and that loss is usually invisible in a budget.

The same arithmetic works in reverse for trips that should not be extended. A negotiation round that genuinely needs five days does not become more productive by being attached to a ninety-day stay, and a traveller who stays on after the work is done creates a question about purpose that a shorter trip would never have raised. The ceiling is a permission, not a plan.

Whose card, whose application

The most frequently skipped question in conversations about the card is whose card is being discussed, and the answer changes the entire administrative path.

For a Chinese national, the card is applied for through the foreign affairs office of the locality where the employer is registered, and eligibility is defined by five categories: government officials handling APEC affairs or card business, and Chinese nationals serving in APEC bodies; managerial, core business and technical staff of large centrally managed state enterprises; staff of local state enterprises; staff of private enterprises; and Chinese nationals of the mainland employed in Sino-foreign equity joint ventures, wholly foreign-owned enterprises, and enterprises funded from Hong Kong, Macao or Taiwan. Applicants must hold a valid Chinese passport, need to travel frequently to APEC economies for work, and must have no criminal record and no record of refusal by a participating economy.

That fifth category is the one that matters most to a foreign-invested company, and it is frequently misread in both directions. It does not give a foreign executive a card. What it does is give the Chinese senior staff of a foreign-invested entity a route to one, which is a different and often more valuable proposition: the people who need to move repeatedly between the Chinese operation and the group’s other Asian sites are usually Chinese nationals, and their travel is usually the travel that a visa queue slows down.

For a foreign national, the card is issued by their own economy, under that economy’s rules, and China’s role is to recognise it at the border. This has two consequences. The first is that the application timetable is set elsewhere: a European or Australian executive applies at home, through the channel their economy operates, and the Chinese entity cannot accelerate it. The second is that the card’s coverage is set by the issuing economy’s bilateral arrangements, so two executives of the same group holding cards from different economies may not have identical access.

The Chinese-side application chain is worth understanding even by companies that will never use it, because it explains why a card is a planning instrument rather than a quick fix. As local foreign affairs offices describe it, the sequence runs from the employer’s online submission, through acceptance and preliminary review at a designated service point, to review by the municipal foreign affairs office, approval by the Ministry of Foreign Affairs, endorsement by each participating economy, and finally centralised production and issuance of the card. Each stage has its own queue, and the stage that takes longest is the one outside Chinese control: the endorsements. Supporting materials typically include the employer’s business licence, social insurance records, a criminal record certificate, the passport data page, a specimen signature and a photograph.

Two implications follow. The first is timing: a card obtained in anticipation of a need is an asset, and a card applied for because a trip is imminent is a delay. The second is that the employer, not the traveller, is the applicant in practice, which means the nomination decision is a management decision about which roles need mobility, not an administrative favour granted to whoever asks.

Three operational details sit on top of this and are worth knowing before a first application. The card is used together with the passport presented when it was issued, so a renewed passport creates a mismatch that has to be dealt with administratively. A holder may only enter the economies that have approved the card, as shown by the abbreviations on its reverse. And the dedicated channels at ports are a convenience, not a right: using one without an approved economy or without the required visa can end in refusal, which is a worse outcome than joining the ordinary queue.

One further detail is specific to the region and is easy to overlook. Travel between the mainland, Hong Kong and Chinese Taipei continues to be handled under the pre-existing arrangements, and the card does not apply to those movements. A cardholder flying from Shanghai to Hong Kong is not using the card, whatever the itinerary looks like.

Where the card stops

Four boundaries in the rules define the card, and the extension moved none of them.

The first is employment. The card permits business visits — meetings, negotiations, inspections, audits, negotiations with suppliers, attendance at conferences, supervision of a project. It does not permit a holder to take up a position inside a Chinese entity: a local employment contract, local payroll, a reporting line into the local organisation. That activity requires a work permit and the corresponding residence permit, and the fact that 90 days is now available does not change the requirement. What it changes is the temptation.

The second is activity unrelated to the business visit. The card is issued for a purpose, and the purpose is commercial. A holder who spends a long stay on activities that have nothing to do with the stated business purpose has created a discrepancy between the basis of entry and the conduct in the country, and the discrepancy is the problem rather than the activity itself.

The third is time. A stay that runs past the permitted period is an overstay, with consequences that attach to the individual and, through the record, to future applications. The fact that the ceiling is now higher makes overstay less likely by accident and more likely by design, because a longer permitted stay is easier to lose track of. The remedy is a reminder set at the point of entry, not at the point of departure.

The fourth is the card’s own lifecycle. The card has a validity period, it is used with the passport it was issued against, and it can be affected by other immigration events. In three participating economies — Australia, New Zealand and Papua New Guinea — only one valid entry permission is allowed at a time, so a visa granted during the card’s validity can override it and shorten the access the holder thought they had. That is a rule of those economies rather than of China, and it is the kind of detail that surfaces at an airport.

There is a fifth boundary that is not in the rules but in the record, and it becomes more important as stays get longer. Every entry produces a record of the purpose stated, and every subsequent entry is read against it. A holder who enters four times in a year, each time describing the visit as supplier audit, and who on the fourth occasion is in fact running a joint venture’s operations, has created a set of records that contradict each other. The point is not that the officer will reconstruct the year. It is that when something else goes wrong — an overstay, a dispute, an inspection — the file is read as a whole, and a file that is internally consistent is a different kind of file from one that is not.

The practical discipline is therefore cumulative rather than per-trip. Keep the invitation letter, the schedule and the return ticket for each stay, and keep them together. Ask, once a quarter, whether the description of the activity still matches what the person is actually doing. That question is cheap to answer and expensive to have never asked, because the answer determines whether the next trip is a business visit or a work permit application, and by the time the question is forced the person is usually already in the country.

Taken together, the boundaries describe a simple test that a company can apply without legal advice. Ask who bears the cost of the person’s time, who directs the work, and who pays. A business visitor is paid by the home entity and directed by the home entity while on site for a defined purpose. An employee is paid locally and directed locally. The length of the stay is evidence, not the test.

Choosing between the card and the alternatives

Three instruments let a person enter China without a conventional visa application, and a fourth lets them work. They overlap in appearance and diverge in every operational dimension that matters.

InstrumentMaximum stay per entryPermitted activityApplication in advanceMultiple entry
APEC Business Travel CardUp to 90 daysBusiness visits, meetings, negotiations, short-term commercial activityYes, through the issuing economyYes, within validity
Unilateral visa-free entryUp to 30 daysBusiness, tourism, visiting relatives, exchange visits, transitNoYes
240-hour visa-free transitUp to 240 hoursTransit to a third country, within a defined areaNo, but an onward ticket is requiredPer movement
Work permit and residence permitAs permittedEmploymentYes, and the permit comes firstAs permitted

The table is a decision aid, not a rulebook. Each instrument has its own conditions, and the detail of the other three sits outside this article. What the comparison does show is where the card is genuinely irreplaceable and where it is merely convenient.

The card’s irreplaceable advantage is repetition combined with length. A single 90-day stay can be replicated across a validity period, which no visa-free arrangement offers at that length. For an executive who visits four times a year for three weeks each, the card saves the application cycle rather than the days. For an executive who needs one 80-day stay, the card is the only instrument on the list that provides it without a visa application.

The card’s weakness is that it is not a substitute for the others in either direction. A national of a visa-free country who visits for a week does not need a card, and obtaining one would be an unnecessary administrative project. A holder whose trip is a two-day stopover en route to a third country does not need the card either, and using it for that purpose is a category error rather than an efficiency. The instruments are not ranked; they are matched to trip shapes.

There is a practical hierarchy worth writing down for a travel desk. If the trip is under 30 days and the traveller’s nationality is on the unilateral visa-free list, no application is needed at all. If the trip is a genuine transit to a third country and fits within 240 hours and the permitted area, transit is available without an application. If the trip is longer than 30 days, or repeats often, or the traveller’s nationality is not on the visa-free list, the card becomes the instrument of choice. And if the activity is employment, none of the first three apply and the work permit route is the only route.

The hierarchy describes single trips, and most real itineraries are not single trips. A traveller may fly into Shanghai for two days of meetings, continue to Seoul, and return three weeks later for a plant visit. In that pattern the first leg may sit comfortably within visa-free entry and the second may need the card, and the correct handling is two entries on two different bases rather than one entry stretched to cover both. Mixing bases within a single entry is where itineraries become indefensible: a traveller admitted for transit who stays on for meetings has not combined two instruments, they have used one instrument for a purpose it does not cover.

Two specific temptations are worth naming. The first is the refundable onward ticket bought to satisfy a transit requirement, with no intention of taking the flight. That is not a documentation gap; it is a false statement about the journey, and it converts a legitimate facilitation into a misrepresentation. The second is the habit of treating the visa-free allowance as a buffer that can absorb the first days of a longer assignment while a card application is pending. It cannot, because the two bases have different permitted activities, and the buffer period is exactly when the activity is most likely to drift.

The figures give a sense of how crowded that landscape now is. According to numbers released by exit-entry authorities in November 2026, China applies unilateral visa-free entry to 50 countries, holds full mutual visa exemption with 29, and applies 240-hour visa-free transit to 57. In that environment, the card’s value is not access; it is duration and repetition for people whose access is already settled.

The company-side routine that makes the extra days usable

A policy change produces no benefit until someone changes a process. Six routines convert the extra thirty days into something a business can use.

Decide who holds a card, and why. The population is narrower than most groups assume: Chinese senior staff of the China entity, plus foreign executives whose home economy issues cards. A short register — name, nationality, card number, issuing economy, validity, passport number — is the foundation of everything else, and most companies do not have one.

Separate the two application paths in the travel policy. Chinese nationals go through the local foreign affairs office; foreign nationals go through their own economy. A single internal process that treats them as one will mislead at least half the applicants, and it usually misleads the ones with the least time to spare.

Record entry and exit dates per trip, and compute days per stay rather than days per year. This is the only way to use the ceiling deliberately and the only way to notice an overstay before the border does.

Fix the documentation habit before the first long stay. An invitation letter that describes the purpose, a meeting or audit schedule, accommodation details and a return ticket answer the question the officer is actually asking. A traveller who can produce them in thirty seconds has a different experience from one who cannot.

Review the posting, not just the trip, whenever a stay approaches the ceiling. If the activity has drifted from supervision into management, or from advice into decision-making, the drift is the risk. Payroll and social insurance treatment are the tell: a person on the home payroll who directs local staff for three months is a business visitor with an unusual amount of influence; a person moved onto local payroll is an employee, and the paperwork should say so.

Set the card’s own dates. Validity, passport alignment and any visa granted for a single-entry-permission economy all have dates, and each of them can invalidate an assumption made at the planning stage.

One further routine deserves separate mention because it is usually missing: deciding who owns the register. In most groups the answer is nobody. Travel is booked by assistants, employment sits with human resources, immigration sits with legal or with a vendor, and the card sits with whoever happened to apply for it. The register is the only artefact that spans all four, and it survives staff turnover only if someone is named as its owner. The same applies to people who are not employees in the ordinary sense — secondees, contractors, advisers seconded into a project, staff of a joint venture partner. They travel on the same routes and their status is usually documented worse than anyone else’s.

None of this is expensive. It is the sort of work that takes an afternoon to design and ten minutes a quarter to maintain, and its value is that it makes the difference between a policy that exists and a policy that is used.

Two failures worth naming, and an outlook

Two failures recur, and they are worth naming because both look like efficiency at the moment they are chosen.

The first is the probation-period posting. A company wants to test a candidate for a China-based role, so it sends them on a card for 85 days, keeps them on the home payroll, and calls the arrangement a long business visit. The arrangement is defensible only if the person is genuinely not working in a local position — advising, auditing, negotiating, supervising a project. If they are running a department, the description is wrong, and the exposure is not limited to immigration: it also touches tax residence, social insurance and the enforceability of the employment relationship. The extra 30 days did not create this problem. It made it easier to postpone solving it.

The second is the substitution error. A company discovers that a traveller’s nationality is on the visa-free list and cancels the card application, or discovers the card and cancels the visa application, without checking which instrument covers the actual activity. The instruments are not fungible. Visa-free entry carries a 30-day ceiling and a defined list of purposes; the card carries 90 days and a business purpose; neither carries employment. A travel desk that treats them as interchangeable will eventually book someone on the wrong basis, and the cost of that error is borne by the traveller at a border rather than by the desk in a spreadsheet.

The outlook is that the direction of travel is toward more duration and more instruments, not fewer. The card extension sits in the same period as the widening of unilateral visa-free entry and the expansion of transit arrangements, and the pattern is consistent: China is making short-term business mobility easier for people whose access is already lawful, while keeping the boundary between visiting and working where it has always been.

For a company, the sensible response is not to chase each announcement but to build the register and the calendar that make any of them usable. The organisations that benefit from mobility policies are rarely the ones that read them first. They are the ones that already know who travels, on what document, for what purpose, and for how long — because when a ceiling moves, the only thing that has to change is a number in a file that already exists.

Sources

  • 中国领事服务网·APEC 商务旅行卡简介(计划沿革、21 个经济体、过渡成员、中方五类申请对象与申办条件):https://cs.mfa.gov.cn/zggmcgx/cgqzapecswlxk/cgqzjj_149486/apecswlxkjj/202607/t20260713_11981082.html
  • 央广网·中方决定将 APEC 商务旅行卡持卡人在华最长停留期延长至 90 天(自 2026-11-01 起实施):https://www.cnr.cn/newscenter/native/gd/kx/20261030/t20261030_527857478.shtml
  • 中国日报网·成都航空口岸出入境人员突破 600 万人次(四川边检总站 2026-11-11 披露,含单方面免签 50 国、全面互免 29 国、240 小时过境免签 57 国口径):https://ex.chinadaily.com.cn/exchange/partners/82/rss/channel/cn/columns/j3u3t6/stories/WS6af525d9e4b0e14252fca9fa.html
  • 外交部·亚太经合组织基本情况(成员与观察员、合作方式):https://www.fmprc.gov.cn/gjhdq_676201/gjhdqzz_681964/lhg_682278/jbqk_682280/

亚太经合组织商务旅行卡在华停留延至 90 天后:能替代什么、替代不了什么

一个参数,当作既定前提

政策事实短到一句话就说完,本文不再在它上面多花篇幅。自 2026 年 11 月 1 日起,亚太经合组织商务旅行卡持卡人在华最长停留期由每次 60 天延长至 90 天。中方在 2026 年 10 月下旬对外公布这一决定,定位是持续便利亚太区域商务人员往来;适用于各参与经济体签发的旅行卡持卡人。

卡的其余部分没有变。申请资格、签发路径、参与经济体范围、口岸专用通道,以及这张卡是通行便利而不是工作许可这一条,都停在 2026 年 10 月 31 日的位置上。政策本身已有专文写过;没被写过的,是持卡人或其雇主在 11 月 2 日会做什么不同的事——而这正是本文的全部题目。

这样界定范围,是因为一个数字的变化会引出两种相反的错误。第一种是当成什么都没发生:30 天放在一段长驻里像个零头,于是日历照旧,便利从未被取用。第二种是当成什么都变了:上限抬了,所以这张卡现在覆盖了它以前不覆盖的事,所以工作许可那场对话可以往后放。两种错误都很常见,失败方式却不同。第一种浪费了一项已经给出的额度;第二种把一项便利转换成一处理险敞口。

正确的读法在两者之间。这张卡对某一类明确的行程变得更有用,对其他任何一类都没有变得更有用。判断某一次行程属于哪一类,需要一点算术、一点单证,以及一次关于活动性质的判断——而这件事值得在下一版出差日历定稿之前做完,因为日历通常一年只定一次,然后一路守到底。

90 这个数字之所以不只是算术,还有一个理由。60 天与一个商业季度并不同构:它只覆盖三个月周期里的两个月,因此按季度节奏安排的停留,要么提前收尾,要么越界。90 天覆盖一整个季度,两端还留有余地。对按季度节点组织的项目工作——一段调试期、一个预算周期、一个审核季、一个产品上市窗口——新的上限是第一个贴合工作形状、而不是贴合出入境规则形状的上限。这是实质性的收益,而它在公告里看不见,公告只是把新的最大值写了出来。

算一次停留:上限管什么,不管什么

首先要弄清计量单位。90 天是单次停留的上限,自入境之日起算。它不是年度配额,不是月度额度,也不是一个可以靠出境再入境来充值的滚动余额。3 月 1 日入境、5 月 29 日离境,这一次停留用掉了 90 天;3 月 1 日入境、3 月 31 日离境、4 月 1 日再入境,则是两次停留,按同一套算术,各自最长可达 90 天。

其次要看清公告没说什么。它没有规定年度累计停留上限,也没有规定两次停留之间必须间隔多久。这种沉默不是邀请。单次停留的上限是一道控制;口岸的审查是第二道;来华事由是第三道。一连串首尾相接、每一次都刚好停在上限之内的行程,会被读成它本来的样子,而这个读法不会好看。

再次,上限不是权利。90 天意味着在按此给予入境的前提下,可停留的最长时间是 90 天;它不意味着每一次入境都会给出 90 天。实际给出的天数由口岸决定,取决于申报的事由、出示的材料和现场的判断。邀请函覆盖两周、回程票却订在第十一周的持卡人,是在主动请人提问。

对雇主来说,这套算术有一个容易制度化的形状。按每次入境天数计数,不按年度累计。逐次记录入境日与离境日,因为离境日才是把这一次计数关上的那一天。并且把这份计数放在出差管理、人力资源和财务都能看到的地方,否则这三方各自会维护一份不同的版本。

还有第四点,不那么显眼,后果却更实。因为上限是按次计的,过去需要中途出境一次的行程,现在可以做成一次连续停留——但前提是行程从一开始就这么设计。围绕 60 天上限排出来的行程里,第 55 天有一次没人质疑的「出境跑一趟」,因为它一直在那里。要把它删掉,需要有人判定它不再必要;这个判定不是政策做出的,是掌握日历的那个人做出的。

举一个例子,差别就看得见。假设一个供应商审核项目分四段:在供应商办公室做文件复核、两周产线观察、整改讨论、以及六周后的复查。在 60 天上限之下,自然的排法是一趟走完前三段,第二趟做复查,中间留出一段由供应商自己决定的空档。在 90 天上限之下,同样的四段可以放在一次停留里,复查由提出问题的同一个人完成。省下的不是机票,而是一处问题被交给一个当时不在场的人时丢掉的那段背景——而这种损失通常不会出现在预算表里。

同一套算术,对不该延长的行程反向成立。一轮真正只需要五天的谈判,不会因为挂在一段 90 天的停留里就变得更有效率;而一位在工作结束之后继续留下的人,会造出一处关于事由的疑问,那是更短的行程本来不会引出的。上限是许可,不是计划。

谁的卡,谁来申办

关于这张卡的讨论里,最常被跳过的问题是「这是谁的卡」,而这个答案会改变整条行政路径。

对中国公民而言,卡由用人单位注册地的地方外事部门受理,申请资格按五类对象界定:从事亚太经合组织事务的政府官员或负责旅行卡业务的政府官员,以及在亚太经合组织任职的中国籍官员;中央管理的大型国有企业中的管理人员、主要业务人员和技术人员;地方国有企业人员;民营企业人员;中外合资、外商独资和台港澳资企业中的中方(大陆)人员。申请人须持有效中国护照,因工作需要经常前往亚太经合组织经济体,且无刑事犯罪记录、无被实施旅行卡计划的经济体拒签的记录。

第五类对外资企业最要紧,也最常被两头误读。它不给外国高管发卡;它给的是外资企业中的中方骨干一条办卡路径,而这往往是更有价值的一件事:需要在中国实体与集团其他亚洲站点之间反复往返的人,通常是中方人员,而被签证排队拖慢的,通常也是他们的行程。

对外国公民而言,卡由其所属经济体按本国规则签发,中国的作用是在口岸予以承认。这有两个后果。第一,申请时间表定在别处:一位欧洲或澳大利亚高管在本国、按本国运营的渠道申办,中国实体无法加速。第二,卡覆盖哪些经济体,取决于签发经济体之间的双边安排,因此同一集团里持不同经济体签发的两张卡,通行范围可能并不一致。

中方这一侧的申办链条,即便永远用不上的企业也值得弄懂,因为它解释了为什么这张卡是规划工具,而不是应急手段。按地方外事部门公布的流程,链条从企业在系统内提交申请开始,经受理点受理与初审、市级外事办公室审核、外交部审批、亚太经合组织各经济体核发签证,最后由外交部统一制证并颁发。每一段都有自己的排队,而最慢的一段不在中方手里:各经济体的核准。材料通常包括企业营业执照副本、社保记录凭证、无犯罪记录证明、有效护照首页、个人签名与白底彩色照片。

由此有两个含义。第一是时点:在需求出现之前办下来的卡是资产,因为行程迫近才去申请的卡是延误。第二是实务上申请人是用人单位而不是出差人,这意味着「提名谁办卡」是一项关于哪些岗位需要往来便利的管理决策,而不是给谁的一个行政人情。

在此之上还有三个操作细节,第一次申办之前值得知道。卡与申办时所持护照配套使用,护照换发之后两者对不上,必须按签发机关要求另行处理。持卡人只能出入已批准其卡片的经济体,以卡片背面的经济体英文缩写为准。口岸的专用通道是便利,不是权利:在未获批的经济体、或者缺少所需签证的情况下走专用通道,可能以被拒入境收场,那比老老实实排队更糟。

还有一个细节是区域性的,很容易漏。内地、中国香港与中国台北之间的人员往来仍按原办法办理,旅行卡对这些往来不适用。一位持卡人从上海飞香港,这一次没有在用这张卡,无论行程看上去多像。

这张卡停在哪里

规则里有四条边界界定了这张卡,而这次延长一条都没有动。

第一条是雇佣。卡许可的是商务访问——会议、洽谈、考察、审核、与供应商谈判、参加会议、监督项目。它不许可持卡人在中国境内实体任职:本地劳动合同、本地发薪、向本地组织汇报。那类活动需要工作许可及相应居留证件,而 90 天现在可用这一事实并不改变这项要求。它改变的是诱惑。

第二条是与商务访问无关的活动。卡因某一事由而签发,那个事由是商业性的。一次长停留如果主要花在与申报事由无关的事情上,就在入境依据与境内行为之间造出了一处不一致,而问题出在这处不一致,不在那些活动本身。

第三条是时间。停留超出许可期限就是逾期居留,后果挂在个人身上,并通过记录影响后续申请。上限抬高之后,因疏忽而逾期的概率下降,因设计而逾期的概率上升,因为一段更长的许可停留更容易被忘掉。办法是在入境那一刻设提醒,而不是在离境那一刻。

第四条是卡自己的生命周期。卡有有效期,与签发时所依据的护照配套使用,也会受其他出入境事件影响。在澳大利亚、新西兰、巴布亚新几内亚三个经济体,同一时间只允许持一种有效入境许可,因此在卡有效期内获批的签证可能覆盖掉这张卡,把持卡人以为自己拥有的通行范围缩短。这是那三个经济体的规则,不是中国的规则,也正是那种在机场才会浮出来的细节。

还有第五条边界,它不在规则里,而在记录里,并且随着停留变长而更要紧。每一次入境都会留下一条关于申报事由的记录,而每一次后续入境都会被拿去和它对照。一位持卡人一年入境四次,每次都把访问描述为供应商审核,到第四次其实是在管一家合资企业的运营——他造出的是一组互相矛盾的记录。要点不在于查验人员会把这一年重建出来,而在于当别的地方出问题——一次逾期、一场纠纷、一次检查——档案是被整份读的;一份内部自洽的档案,和一份不是的,是两种不同性质的档案。

因此,纪律是累积的,不是逐次的。把每一次停留的邀请函、日程与回程票留下来,并且放在一起。每季度问一次:对活动的描述,是否仍然匹配这个人实际在做的事。这个问题回答起来很便宜,从没问过却很贵,因为答案决定下一趟行程是商务访问还是工作许可申请;而等到这个问题被强制提出时,人通常已经在境内了。

五条合起来,构成一个企业不用请律师就能用的检验。问三件事:这个人的时间成本由谁承担,工作由谁指挥,薪酬由谁支付。商务访问者由派出实体付薪、在现场为某一明确目的受派出实体指挥;雇员由本地付薪、受本地指挥。停留长度是证据,不是标准。

在卡与替代方案之间做选择

有三种工具可以让人不经常规签证申请入境中国,第四种可以让人工作。它们外观上重叠,在每一个真正要紧的操作维度上都不相同。

工具单次最长停留许可活动是否须提前申请是否可多次入境
亚太经合组织商务旅行卡最长 90 天商务访问、会议、洽谈、短期商务活动须,向签发经济体申办可,在有效期内
单方面免签不超过 30 天经商、旅游观光、探亲访友、交流访问、过境不须可
240 小时过境免签不超过 240 小时前往第三国的中转,限规定区域不须,但须持联程客票按次
工作许可与居留证件按许可期限受雇工作须,且须先取得工作许可按许可

这张表是决策辅助,不是规则手册。每种工具各有条件,另外三种的细节不在本文范围。它真正显示的,是这张卡在哪里不可替代、在哪里只是方便。

卡不可替代之处,是「可重复」加「够长」的组合。一段 90 天的停留可以在有效期内重复,这是任何免签安排都给不了的长度。对一位一年去四次、每次三周的高管,卡省下的是申请周期而不是天数;对一位需要一次 80 天停留的高管,卡是清单上唯一能不经签证申请就给到这个长度的工具。

卡的短处,是它在任何一个方向上都不是其他工具的替代品。免签国家的公民去一周,不需要卡,办卡只是给自己加一个行政项目。持卡人两天后转机去第三国,也不需要卡,用卡做这件事是归类错误,不是效率。这些工具不是排位的,是被匹配到不同形状的行程上的。

对出差管理岗来说,有一条值得写下来的次序。行程在 30 天以内、且旅客国籍在单方面免签名单上,什么都不用申请。行程是真正的前往第三国中转、且在 240 小时与规定区域内,可以走免签过境。行程超过 30 天、或者反复往返、或者旅客国籍不在免签名单上,卡才是首选。而如果活动本身是受雇工作,前三种都不适用,工作许可是唯一的路。

这条次序描述的是单一行程,而真实行程大多不是单一行程。一位旅客可能先飞上海开两天会,接着去首尔,三周后再回来验厂。在这种排法里,前一段可能稳稳落在免签范围内,后一段可能需要用卡,正确的处理是两次入境、两个不同依据,而不是把一次入境拉长去覆盖两段。在单次入境里混用依据,正是行程变得说不清的地方:一位以中转事由获准入境、却留下来开会的人,并没有把两种工具组合起来,他只是用了一种工具去做它不覆盖的事。

有两种具体的诱惑值得点名。第一种是为了满足中转要求而购买可退的续程机票,本无乘机意图。那不是单证瑕疵,那是关于行程的不实陈述,它把一项正当的便利变成了虚假陈述。第二种是习惯性地把免签额度当成一段缓冲,用来吸收较长派驻的头几天,同时等卡办下来。它不能当缓冲,因为两种依据许可的活动不同,而缓冲期恰恰是活动最容易发生漂移的那段时间。

几个数字可以看出这片版图如今有多密。据边检机关 2026 年 11 月公开发布,中国对 50 个国家实施单方面免签,与 29 个国家全面互免签证,240 小时过境免签适用国家为 57 个。在这样的环境里,这张卡的价值不在准入,而在为那些准入早已解决的人提供时长与重复。

让多出的天数真能用起来的企业日常

政策变化不会自己产生收益,除非有人改了流程。六件日常动作,可以把多出的 30 天变成业务能用的东西。

第一,定清谁持卡、为什么持卡。这个人群比多数集团以为的窄:中国实体的中方骨干,加上由本国经济体签发卡的外国高管。一份简短的台账——姓名、国籍、卡号、签发经济体、有效期、护照号——是其余一切的基础,而多数企业没有这份台账。

第二,在出差制度里把两条申请路径分开。中国公民走地方外事部门,外国公民走所属经济体。把两者当成一件事的内部流程,至少会误导一半申请人,而且往往误导的是最没时间折腾的那一半。

第三,逐次记录出入境日期,按每次停留算天数,而不是按年度算。这是有意识地使用上限的唯一办法,也是在口岸之前发现逾期的唯一办法。

第四,在第一次长停留之前把材料习惯固定下来。写明事由的邀请函、会议或审核日程、住宿信息、回程票,回答的正是查验人员真正在问的问题。三十秒内拿得出来的人,和拿不出来的人,体验完全不同。

第五,每当停留接近上限,复核的是「派驻」而不只是「行程」。如果活动从监督漂成了管理,从建议漂成了决策,漂移本身就是风险。薪酬与社保口径是提示器:一位由境外发薪、连续三个月指挥本地员工的人,是一位影响力异常大的商务访问者;一位被转到本地发薪的人,是雇员,单证就应当这么写。

第六,把卡自己的几个日期管住。有效期、与护照的对应关系、以及为「单一入境许可」经济体获批的任何签证,都带日期,每一个都可能推翻规划阶段的某个假设。

还有一件日常动作值得单独说,因为它通常缺席:定清这份台账归谁。多数集团里答案是没人。机票由助理订,用工归人力资源,出入境事务归法务或外部服务商,卡归当初恰好去办的那个人。台账是唯一横跨这四者的物件,而它只有在有人被指定为责任人的时候,才能熬过人员流动。这条同样适用于那些在通常意义上不算雇员的人——派驻人员、承包商、借调进项目的顾问、合资伙伴的员工。他们走同样的路线,而他们的状态通常比任何人都记录得更差。

这些都不贵。设计起来一个下午,每季度维护十分钟,价值在于它决定了政策是「存在」还是「被使用」。

两种值得点名的失败,以及一点前瞻

有两种失败反复出现,值得点名,因为它们在被人选中的那一刻都像是效率。

第一种是「试用期式派驻」。企业想考察一位拟任中国岗位的人选,于是用卡把人派来 85 天,薪酬留在境外,把安排称作一次较长的商务访问。只有在这位人士确实没有在本地岗位工作——只是提供建议、审核、谈判、监督项目——这个说法才站得住。如果他在管一个部门,描述就是错的,而敞口不只限于出入境:它还牵动税收居民身份、社会保险,以及这段用工关系本身的可执行性。多出的 30 天没有制造这个问题,它只是让推迟解决变得更方便。

第二种是「替代错误」。企业发现某位旅客的国籍在免签名单上,就取消了办卡;或者发现了这张卡,就取消了签证申请;两者都没有核对哪种工具覆盖真实的活动。这些工具不能互换。免签有 30 天上限和一份明确的事由清单;卡有 90 天和商务事由;两者都不含雇佣。把三者当成可互换的出差管理岗,迟早会把某个人按错误的依据送出境,而这项错误的成本由旅客在口岸承担,不由那个岗位在表格里承担。

前瞻的方向是时长更长、工具更多,而不是更少。卡的上限调整,与单方面免签范围的扩大、过境安排的扩容处在同一段时间里,脉络是一致的:中国在让那些准入本来就合法的人更容易短期往来,同时把「访问」与「工作」之间那条线留在原处。

对企业的合理回应,不是追每一条公告,而是把台账和日历建起来,让任何一条公告都用得上。从往来便利政策里受益的组织,很少是第一批读到它的人,而是那些本来就知道谁在出差、拿什么证件、为什么事由、去多久的组织——因为当上限移动时,唯一需要改的,是一份本来就存在的档案里的一个数字。

来源

  • 中国领事服务网·APEC 商务旅行卡简介(计划沿革、21 个经济体、过渡成员、中方五类申请对象与申办条件):https://cs.mfa.gov.cn/zggmcgx/cgqzapecswlxk/cgqzjj_149486/apecswlxkjj/202607/t20260713_11981082.html
  • 央广网·中方决定将 APEC 商务旅行卡持卡人在华最长停留期延长至 90 天(自 2026-11-01 起实施):https://www.cnr.cn/newscenter/native/gd/kx/20261030/t20261030_527857478.shtml
  • 中国日报网·成都航空口岸出入境人员突破 600 万人次(四川边检总站 2026-11-11 披露,含单方面免签 50 国、全面互免 29 国、240 小时过境免签 57 国口径):https://ex.chinadaily.com.cn/exchange/partners/82/rss/channel/cn/columns/j3u3t6/stories/WS6af525d9e4b0e14252fca9fa.html
  • 外交部·亚太经合组织基本情况(成员与观察员、合作方式):https://www.fmprc.gov.cn/gjhdq_676201/gjhdqzz_681964/lhg_682278/jbqk_682280/

Registering a Foreign Enterprise Permanent Representative Office: What the Documents Have to Prove

  • A Representative Office has no legal personality and no registered capital, so the contribution and subscription rules that occupy so much attention in company formation never reach it.
    代表机构不具有法人资格,也不设注册资本,因此公司设立中最受关注的那套出资与认缴规则,根本落不到它身上。
  • Four differences separate it from a foreign-invested enterprise: no registered capital, no profit-making activity, a defined Non-Profit Liaison scope, and liability resting with the foreign enterprise rather than with the office.
    它与外商投资企业有四点本质差异:不设注册资本、不得从事营利性活动、活动范围限于法定的非营利性联络、责任由外国企业而非代表机构承担。
  • The document set has three layers: the parent’s standing, the authorisation and the chief representative, and the premises.
    材料清单分三层:外国企业的主体资格、授权与代表人、驻在场所。
  • From 7 November 2023 an Apostille issued by the contracting state replaces consular legalisation for parties in states party to the Convention, while non-contracting states still require notarisation plus consular authentication.
    自 2023 年 11 月 7 日起,缔约国主体可用该国签发的附加证明书替代领事认证;非缔约国仍须公证加使领馆认证。
  • The Annual Report Obligation runs from 1 March to 30 June each year and covers the parent’s continued existence, the office’s activity, and its income and expenditure as audited by an accounting firm.
    年度报告义务为每年 3 月 1 日至 6 月 30 日,内容包括外国企业合法存续情况、代表机构业务活动开展情况,以及经会计师事务所审计的费用收支情况。
  • Change registration is due within 60 days of the change, and within 30 days of approval where the change required approval first; a change of name or scope is not a matter for the office’s own discretion.
    变更登记应自变更之日起 60 日内申请;依法须先经批准的,自批准之日起 30 日内申请。名称或者业务范围的变动不由代表机构自行决定。
  • Acting outside the permitted scope attracts a fine of RMB 10,000 to 100,000 and, in serious cases, revocation of the registration certificate.
    从事法定活动范围以外的活动,可处 1 万元以上 10 万元以下罚款;情节严重的,吊销登记证。
  • Deregistration is due within 60 days of any of four triggering events, and the file must include the tax deregistration certificate and a customs clearance certificate.
    出现四种情形之一,应在 60 日内申请注销登记,材料须包括税务登记注销证明与海关出具的相关证明。

Registering a Foreign Enterprise Permanent Representative Office: What the Documents Have to Prove | 外国企业常驻代表机构登记:材料背后的三层逻辑

A form that is not a company, and four differences that follow

The permanent representative office is the oldest foreign presence in China and the one most often misunderstood. It is old enough that practitioners still call it by the name of a regulation repealed in 2011, and common enough that almost every international group has had one at some point. It is also the form that generates the largest number of false expectations, because it looks like a company and is not one.

The governing instrument is the Regulations on the Registration and Administration of Permanent Representative Offices of Foreign Enterprises, promulgated by State Council Order No. 584 on 19 November 2010 and in force from 1 March 2011. The Regulations have been amended since, but the architecture has not moved: registration is constitutive, the registration authority sits at provincial level, and the office exists only within the scope the Regulations allow.

Article 2 supplies the definition that does most of the work. A representative office is an office established in China by a foreign enterprise to conduct non-profit-making activities related to that enterprise’s business. It does not have legal personality. Every practical consequence follows from those two sentences. No legal personality means no separate legal person, no shareholders, no equity, and therefore no registered capital. Non-profit-making activity means the office is not a vehicle for earning revenue in China, and the Regulations treat any departure from that as a violation rather than as a business decision.

The word “permanent” in the English rendering is slightly misleading. It does not mean indefinite. Article 16 provides that the term of residence may not exceed the remaining existence of the foreign enterprise, which means the office’s authorised life is bounded by the parent’s. A parent with a fixed corporate term confers a fixed term on its office, and a parent that is itself approaching dissolution cannot confer a long one.

Registration is not a formality that follows the decision to open an office; it is the act that brings the office into existence. Article 24 provides that the date of issue of the registration certificate is the date of establishment. Before that date there is no office, and Article 35 makes unregistered establishment an offence punishable by an order to cease activity and a fine of RMB 50,000 to 200,000.

The registration authority is the market supervision and administration department of the province, autonomous region or municipality directly under the central government. That is a level higher than most company registration, and it is deliberate: the office represents a foreign enterprise rather than a domestic entity, and the authority is placed where the file can be reviewed with the necessary context.

Two features of the regime are worth noting because they shape the documents. The first is that the foreign enterprise is responsible for the truth of the application documents, which Article 4 states directly. The second is that the office must maintain accounting records showing the funds the foreign enterprise remits and the office’s own expenditure, and may not use the account of another enterprise, organisation or individual. Both provisions push the evidentiary burden onto the parent, which is why the document set is heavier than the office’s own scale would suggest.

It is easier to understand the office by contrast than by description. Four differences from a foreign-invested enterprise do most of the explanatory work.

The first is registered capital. A foreign-invested company has it, subscribes to it, and must contribute within the period the law allows. A representative office does not have it at all. This is not an exemption and not a deferral; there is nothing to subscribe to, because there is no equity and no legal person to hold it. The contribution rules that apply to companies — the five-year subscription period, the acceleration of benefit obligations, the disclosure of capital contributions — have no application here, not because the office is treated leniently but because the concepts do not attach. A group that plans an office as a “company without capital” has the wrong model and will build the wrong file.

The second is profit. Article 13 provides that a representative office may not engage in profit-making activity, subject to the exception for offices permitted to do so by an international treaty or agreement to which China is a party, other than provisions to which China has declared a reservation. The exception is narrow and specific. It does not create a general permission for an office that happens to have a commercially useful function.

The third is scope. Article 14 permits two categories of activity connected with the foreign enterprise’s business: market research, display and promotional activities relating to the enterprise’s products or services; and liaison activities relating to the sale of those products, the provision of those services, domestic procurement and domestic investment. Where an approval is required by law, administrative regulation or State Council decision for the office to carry on those activities, the approval must be obtained. The list is exhaustive in form and narrower than most groups assume.

The fourth is liability. Because the office has no legal personality, it has no independent patrimony. Its obligations are the foreign enterprise’s obligations, and a judgment against the office is in substance a judgment against the parent. That has an upside and a downside. The upside is that the office does not need to be capitalised to be credible. The downside is that the parent is exposed without the screen a subsidiary would provide, and that exposure is not reduced by the office’s modest size.

There is a fifth difference that is procedural rather than substantive, and it is the one that generates the most correspondence. The office’s registration matters are filed at provincial level, and the office must announce its establishment and any change in the media designated by the registration authority. A company is used to a national registration system with local windows. An office is used to a provincial registrar with a publication requirement attached.

The document set has three layers

The Regulations list seven items in Article 23. The market supervision administration’s material specifications expand these into a longer list and, more usefully, group them. Read as three layers, the list becomes intelligible.

The first layer is the parent’s standing: proof of the foreign enterprise’s domicile, and proof of lawful operation showing that it has existed for two years or more. The material specification is precise about what this means — a document issued by the relevant authority in the enterprise’s home country or region evidencing the enterprise’s existence for at least two years, or other relevant proof of operation. The two-year requirement is the substantive filter in the whole list. It excludes newly incorporated parents, and it excludes parents whose home jurisdiction does not issue the kind of document that proves continuous existence. Groups that set up a fresh holding company for China purposes and then apply for an office discover this at the window rather than in the plan.

The same layer includes the parent’s articles of association or organisational agreement, and a credit reference issued by a financial institution that has business dealings with the foreign enterprise. The material specification adds a practical relief: where that financial institution is a domestic institution in mainland China, no notarisation or authentication is required. For a group whose China entity already banks with a local bank, that single sentence removes an entire authentication chain.

The second layer is authority and personnel: the foreign enterprise’s authorisation or certification of its authorised signatory, the appointment of the chief representative and representatives, and their curricula vitae and identity documents. Article 11 requires the foreign enterprise to appoint one chief representative, who may sign the application documents within the scope of a written authorisation, and permits the appointment of one to three representatives. The authorisation document is therefore not a formality; it is the instrument that makes the signature on the application valid.

Article 12 disqualifies certain persons from acting as chief representative or representative: those sentenced for harming China’s national security or public interest, those who were chief representative or representative of an office whose registration was revoked or which was ordered closed for such activity and fewer than five years have passed since, and other cases prescribed by the market supervision administration of the State Council.

The third layer is the premises: the lawful right to use the office’s location. Article 15 lets the foreign enterprise choose the location, and allows the relevant authorities to require a change of location on national security or public interest grounds, with notice to the registration authority. Article 18 requires the registration certificate to be displayed conspicuously at the premises. The premises documents therefore do two jobs: they prove a right to occupy, and they fix the address that the certificate will state.

Three practical requirements sit across all three layers. Documents in a foreign language must be accompanied by a Chinese translation, with the translator’s or translation unit’s details. Where an approval is required before establishment, the approval document must be filed and the application made within 90 days of the approval. And the establishment must be announced by the foreign enterprise in the media designated by the registration authority.

One item deserves separate mention because it is frequently omitted. Where the foreign enterprise relies on an international treaty or agreement to establish an office permitted to carry on profit-making activity, the corresponding documents must be submitted under that treaty or agreement. The exception in Article 13 is real, but it is proved by documents, not asserted.

The name, the certificates and the register

Three provisions sit between the file and the office’s daily life, and they are usually read once and then forgotten.

The first is the name. Article 10 prescribes its composition and its order: the nationality of the foreign enterprise, the enterprise’s Chinese name, the name of the city where the office is located, and the words “representative office”. The order is not a style preference. It means the name states, in sequence, where the parent comes from, who it is, where the office sits and what kind of thing the office is. Three categories of content are prohibited: anything harmful to China’s national security or public interest, the name of an international organisation, and anything prohibited by law, administrative regulation or State Council decision.

The office must conduct its business under the registered name, and Article 38 makes a departure from that an offence attracting an order to correct and a fine of RMB 10,000 to 30,000, with revocation if uncorrected. In practice the rule bites in an unexpected place: a group that markets its China presence under a global brand name, and lets the office use that name in signage or on documents without the registered Chinese suffix, has departed from the registered name without intending to.

The second is the certificates. Two documents are issued: the registration certificate for the office, and representative certificates for the chief representative and the representatives. Article 18 requires the registration certificate to be displayed conspicuously at the premises. Article 19 prohibits forging, altering, renting out, lending or transferring either certificate, and requires an office that loses or damages one to declare it void in the designated media and apply for a replacement. The same article provides that where the registration authority approves a change, approves a deregistration, revokes a change or revokes the certificate, the original certificates automatically cease to be valid — which is a reminder that the certificate is evidence of a current state, not a permanent licence.

The third is the register. Article 17 requires the registration authority to record the office’s registered matters in the register of representative offices and to make it available to the public for inspection and copying. That has a practical consequence that is easy to overlook: the register, not the certificate on the wall, is what a counterparty, a bank or a court will check. An office whose registered matters have not been updated is misdescribed in the register even if its certificate looks correct, because the certificate follows the register rather than the other way round.

Two publication duties attach to these documents and they are not symmetrical. Article 20 requires the foreign enterprise to announce the office’s establishment and any change in the media designated by the registration authority. Where the office is deregistered, or its establishment is revoked or its certificate revoked, the announcement is made by the registration authority rather than by the enterprise. The asymmetry is sensible: an enterprise has an interest in publishing its arrival and may have less interest in publishing its departure.

The certificates also function as the key to everything downstream. Article 25 provides that the office, the chief representative and the representatives use the registration certificate and the representative certificates to apply for residence, employment, taxation and foreign exchange registration. That single sentence explains why the certificates matter beyond display. Without them, the chief representative cannot regularise residence, the office cannot complete its tax registration, and the foreign exchange registration that lets the parent fund the office cannot be opened. The registration is not one compliance step among several; it is the step that makes the others available.

Three authentication routes

The authentication of the parent’s documents is where files most often stall, and the position changed materially in 2023. There are now three routes, and the correct one depends on where the parent is established.

The first route applies to parties in states party to the Convention Abolishing the Requirement of Legalisation for Foreign Public Documents. China acceded to the Convention on 8 March 2023, and it entered into force for China on 7 November 2023. From that date, a public document issued in a contracting state and intended for use in China needs an Apostille issued by the competent authority of that state, and no longer needs authentication by the Chinese embassy or consulate. The Ministry of Foreign Affairs is the competent authority for Apostilles issued in China, and it has entrusted the foreign affairs offices of provinces, autonomous regions, municipalities and some municipalities to issue Apostilles for documents emanating from their areas. Chinese Apostilles take the form of a sticker with a silver national emblem seal, and they can be verified online.

The second route applies to parties in states that are not contracting states, or that are contracting states which China does not recognise as possessing sovereign statehood or with which the Convention does not apply. For those, the parent’s documents must be notarised by a notary in the home country and then authenticated by the Chinese embassy or consulate there. Where the home state has no diplomatic relations with China, the chain lengthens: authentication by the embassy or consulate of a third state that does have relations with China, then authentication by the Chinese embassy or consulate in that third state. Where the document emanates from a dependent territory, it must first be notarised in the territory, then authenticated by the home state’s foreign affairs authority, then authenticated by the Chinese embassy or consulate there.

The third route applies to investors from Hong Kong, Macao and Taiwan. Their subject-qualification documents are notarised by local notaries under the special provisions or arrangements. For Hong Kong and Macao, the notarial document must be transmitted by the designated body of the Ministry of Justice; where it has been transmitted electronically by that body, the hard copy need not be submitted. Article 44 of the Regulations provides that offices established by enterprises from Hong Kong, Macao and Taiwan are registered by reference to the Regulations.

Two points of practice follow from the three routes.

The first is that the route is determined by the parent’s jurisdiction, not by the applicant’s convenience. A group with a Singapore parent and a Cayman intermediate holding company has to decide which entity is the foreign enterprise for registration purposes, and that decision determines which route applies to which document. Substituting a notarised document from one jurisdiction for an Apostilled document from another is the single most common authentication error.

The second is that the 2023 change did not shorten every chain. It shortened the chain for contracting states by removing the consular step. For non-contracting states the chain is as long as it was, and the third-state authentication route still applies where diplomatic relations are absent. Groups that read “Apostille now applies in China” as a general simplification will be disappointed in exactly those jurisdictions where the file was already slow.

What the office may and may not do

The most frequently asked question about a representative office is whether it can sign contracts and charge for services. The answer is in Articles 13, 14 and 37, and it is narrower than the question implies.

An office may conduct market research, display and promotional activities relating to the foreign enterprise’s products or services. It may conduct liaison activities relating to the sale of those products, the provision of those services, domestic procurement and domestic investment. That is the permitted universe. Marketing in the sense of promotion is inside it. Selling is not. Providing services for a fee is not. Receiving revenue in its own name is not.

The consequences of stepping outside are graded. Article 35 provides that an office engaging in profit-making activity is ordered to correct, has its illegal gains confiscated, has tools, equipment, raw materials and products used for the activity confiscated, and is fined RMB 50,000 to 500,000; in serious cases the registration certificate is revoked. Article 37 provides that an office conducting activity outside the scope of Article 14 is ordered to correct within a time limit and, if it fails to do so, is fined RMB 10,000 to 100,000; in serious cases the certificate is revoked.

The distinction between the two provisions matters. Profit-making activity is treated more severely than activity outside the permitted scope, which tells you that the prohibition on earning is regarded as the core of the form rather than as a technical limit.

A related question is whether the office can employ staff. It can, but not as an employer in the ordinary sense, because it has no legal personality and cannot be a party to a labour contract. The Regulations require the chief representative, the representatives and the staff to comply with the laws and administrative regulations on entry and exit, residence, employment, taxation and foreign exchange registration, and Article 25 provides that the office, the chief representative and the representatives use the registration certificate and representative certificates to apply for residence, employment, taxation and foreign exchange registration. The employment structure is therefore built on service arrangements rather than on direct hire, which is a separate body of practice.

A third question is whether the office can hold assets. It can hold the funds the parent remits and the equipment it needs, and Article 7 requires accounting records that show both. What it cannot do is use another entity’s account, which is a rule aimed at the practice of routing an office’s costs through a related company’s books. For a group that runs several offices in the same city, the accounting separation has to be real, not notional.

Clocks: establishment, change, annual report, exit

The regime is full of deadlines, and they are the part most easily missed because they are spread across four chapters.

On establishment, the registration authority must decide within 15 days of accepting the application, and may consult relevant departments before deciding. If it approves, it must issue the registration certificate and representative certificates within five days of the decision. If it refuses, it must issue a refusal notice within five days, stating the reasons. Where an approval is required before establishment, the application must be made within 90 days of that approval.

On change, Article 26 requires an application whenever a registered matter changes, and Article 27 sets the clock: within 60 days of the change, or within 30 days of approval where the change required approval before registration. Article 28 deals with the case where the term of residence expires and the office wishes to continue: the application must be made within 60 days before expiry. The registration authority must decide a change application within 10 days, and issue the replacement certificates within five days of approval.

Article 31 creates a separate filing obligation, distinct from change registration, for five categories of change at the level of the foreign enterprise: a change of authorised signatory, of the enterprise’s form of liability, of its capital or assets, of its business scope, and a change of representative. Those must be filed within 60 days of the change. The distinction between registration and filing is worth internalising, because the two have different triggers and different consequences.

On the annual report, Article 6 requires the office to submit a report to the registration authority between 1 March and 30 June each year. The content is prescribed: the foreign enterprise’s lawful continued existence, the office’s business activity, and its income and expenditure as audited by an accounting firm. The audit requirement is the operative one. An office that has not maintained separate books cannot produce an audited statement of its own income and expenditure, and the annual report becomes a compliance problem rather than an administrative task.

On exit, Article 32 lists four triggering events, each with a 60-day clock: the foreign enterprise withdraws the office; the term of residence expires and the office does not continue; the foreign enterprise terminates; or the office’s approval is revoked or it is ordered closed. Article 33 requires the application to be accompanied by the tax deregistration certificate, a certificate from customs that relevant matters have been settled or that the office has not gone through the relevant procedures, and other documents prescribed by the State Council’s market supervision administration. The registration authority must decide within 10 days.

Two consequences of missing these clocks are worth stating. Article 38 provides that failure to submit the annual report, failure to use the registered name, failure to adjust the premises on request, failure to publish, and failure to complete change, deregistration or filing procedures each attract an order to correct and a fine of RMB 10,000 to 30,000, with revocation if uncorrected. Article 39 provides that an office engaged in serious illegal activity harming national security or public interest has its certificate revoked, and that the foreign enterprise may not establish another office in China for five years from the revocation.

A pre-flight checklist

For a group opening its first office, the sequence that avoids most of the avoidable delay is short.

Confirm the two-year existence of the entity that will be the foreign enterprise, and confirm that its home jurisdiction can issue a document proving it. Decide which entity that is before assembling anything, because the choice determines the authentication route and, in a group with intermediate holdings, the choice is not obvious.

Map the authentication route for each document. Contracting state: notarisation where required by the home jurisdiction plus an Apostille. Non-contracting state: notarisation plus consular authentication, with the third-state route where diplomatic relations are absent. Hong Kong, Macao or Taiwan: local notarisation with transmission by the designated Ministry of Justice body for Hong Kong and Macao.

Prepare the Chinese translations and the translator’s details at the same time as the originals. A file that arrives with untranslated documents is not a file that is nearly complete; it is a file that has not started.

Decide the chief representative and the representatives, and make sure the written authorisation covers the signature on the application. The number of representatives is capped at three, and the appointments have to be documented in the form the specification requires.

Secure the premises and the documents proving the right to use them. Confirm that the address stated will be the address on the certificate, and that the certificate can be displayed there.

Diarise the four clocks: the 90-day window after any pre-establishment approval, the 60-day and 30-day windows for changes, the 1 March to 30 June annual report window, and the 60-day deregistration windows.

Two closing points about the form itself.

The first is that the office is not a stepping stone that converts into a company. There is no conversion procedure, because there is nothing to convert: no legal person, no capital, no equity. A group that wants a company later incorporates one and, if it wishes, closes the office. The two run in parallel or in sequence, not by transformation.

The second is that the office’s documentary burden is front-loaded and its ongoing burden is light by comparison. Almost all the difficulty sits in the first filing and in the annual audit. A group that invests in getting the first file right, and in keeping the office’s books separate from the first month, will find the following years uneventful. A group that treats the first filing as an exercise in producing whatever the window will accept will spend the following years explaining the gaps.

Sources

  • 中国政府网·《外国企业常驻代表机构登记管理条例》(国务院令第 584 号,2010-11-19 公布,2011-03-01 施行):https://www.gov.cn/zhengce/2010-11/25/content_2602568.htm
  • 中国政府网·国务院公报版《外国企业常驻代表机构登记管理条例》(含 2013 年、2018 年修订表述):https://www.gov.cn/gongbao/content/2019/content_5468900.htm
  • 中华人民共和国外交部·《取消外国公文书认证要求的公约》将于 2023 年 11 月 7 日在中国生效实施:https://www.fmprc.gov.cn/wjbxw_new/202310/t20231023_11165858.shtml
  • 中国领事服务网·领事认证(附加证明书)简介:https://cs.mfa.gov.cn/zggmcgx/lsrz/jj/
  • 中国领事服务网·《取消外国公文书认证要求的公约》缔约国名单:https://cs.mfa.gov.cn/dygmd/202607/t20260724_11991225.html
  • 青岛市企业登记服务网·外国(地区)企业常驻代表机构设立登记办事指南(依据国市监注发〔2022〕24 号附件 3):https://qydjfw.qingdao.gov.cn/files/2026/08/20260821165015218.pdf

外国企业常驻代表机构登记:材料背后的三层逻辑

一种不是公司的形态,以及由此而来的四点差异

常驻代表机构是外资在华最古老的形态,也是最常被误解的一种。它古老到从业者至今仍用一部 2011 年已废止的规定的名字称呼它,也普遍到几乎每一家国际集团都曾经设过一家。它同时也是最容易催生错误预期的形态,因为它看起来像公司,而它不是。

现行依据是《外国企业常驻代表机构登记管理条例》,2010 年 11 月 19 日以国务院令第 584 号公布,自 2011 年 3 月 1 日起施行,此后经修订,但骨架没有移动:登记具有设权效力,登记机关在省级,代表机构只在条例允许的范围内存在。

第二条给了定义,而这个定义承担了大部分解释工作。代表机构是外国企业依照条例规定,在中国境内设立的从事与该外国企业业务有关的非营利性活动的办事机构,不具有法人资格。两句话,把实务后果基本说完了。不具有法人资格,意味着没有独立的法律人格、没有股东、没有股权,因而也没有注册资本。非营利性活动,意味着它不是在中国境内取得营业收入的载体,而条例把任何偏离都当作违法处理,不当作经营决策。

“常驻”这个词容易误导。它不表示无限期。第十六条规定,代表机构的驻在期限不得超过外国企业的存续期限——办事机构的法定寿命被母公司的寿命框住。一家有固定存续期的母公司,给到办事处的也是固定期限;一家自己已经接近解散的母公司,给不出长期限。

登记不是”决定设立”之后的手续,而是让代表机构得以存在的那件事。第二十四条规定,登记证签发日期为代表机构成立日期。在此之前没有代表机构;第三十五条把未经登记擅自设立或者从事代表机构业务活动定为违法,责令停止活动,处 5 万元以上 20 万元以下罚款。

登记机关是省、自治区、直辖市人民政府市场监督管理部门。这一层级比多数公司登记要高,而且是有意的:代表机构代表的是外国企业,不是一个境内主体,把权限放在能带着必要背景读这份卷宗的层级上。

这套制度有两处特征,直接决定了材料的样子。其一,外国企业要对申请文件、材料的真实性负责,第四条写得很直接。其二,代表机构应当依法设置会计账簿,真实记载外国企业经费拨付和自身费用收支情况,并置于驻在场所,且不得使用其他企业、组织或者个人的账户。两条都把举证责任推给母公司——这也是为什么,材料的分量比代表机构自身的体量要大得多。

理解代表机构,用对比比用描述省力。与外商投资企业的四点差异,能解释大部分问题。

第一点是注册资本。外商投资公司有注册资本,要认缴,要在法律允许的期限内实缴。代表机构根本没有这一项。这不是豁免,也不是缓缴:没有股权,没有法律人格去持有出资,也就没有可认缴的对象。适用于公司的那套出资规则——五年认缴期、出资义务加速到期、出资情况披露——在这里不适用,不是因为对它宽容,而是因为概念挂不上去。把代表机构当成”不设资本的公司”来规划的集团,模型错了,卷宗也会搭错。

第二点是营利。第十三条规定,代表机构不得从事营利性活动;中国缔结或者参加的国际条约、协定另有规定的,从其规定,但中国声明保留的条款除外。这个例外很窄、很具体,并不构成一项”只要业务上用得着就可以”的普遍许可。

第三点是范围。第十四条允许两类与外国企业业务有关的活动:与产品或者服务有关的市场调查、展示、宣传;与产品销售、服务提供、境内采购、境内投资有关的联络。法律、行政法规或者国务院规定从事这些活动须经批准的,应当取得批准。这份清单在形式上是穷尽的,比多数集团以为的要窄。

第四点是责任。因为没有法人资格,代表机构没有独立的财产,它的义务就是外国企业的义务;对代表机构作出的判决,实质是对母公司的判决。这一点的好处是:代表机构不需要靠注册资本来建立信用。坏处是:母公司直接暴露,少了子公司那层屏障,而这层屏障不会因为代表机构体量小而变厚。

还有第五点,是程序上的差异,也是产生最多往来函件的一点。代表机构的登记事项在省级办理,且设立、变更须由外国企业在登记机关指定的媒体上向社会公告。企业习惯了全国统一的登记系统加本地窗口;代表机构面对的是一位省级登记官员,外加一项公告义务。

材料清单的三层结构

条例第二十三条列了七项。市场监管部门的提交材料规范把它扩成更长的清单,而且更有用的是做了分组。按三层来读,这份清单就通了。

第一层是外国企业的主体资格:住所证明,以及存续两年以上的合法营业证明。材料规范对这一项写得很具体——指外国企业所在国家或者地区有关部门出具的、证明该企业存续两年以上的主体资格文件,或者其他有关营业证明。”两年”是整份清单里真正的实质筛子。它排除新设的母公司,也排除那些所在法域出不了”连续存续”这类证明的母公司。为中国业务专门搭一家新控股公司、再去申请代表机构的集团,往往不是在规划阶段发现这件事,而是在窗口发现。

同一层里还有外国企业的章程或者组织协议,以及同该外国企业有业务往来的金融机构出具的资信证明。材料规范在这里给了一处很实用的松绑:该金融机构为中国大陆地区境内金融机构的,无需公证认证。对一家境内实体已经在本土银行开户的集团来说,这一句话省掉的是一整条认证链。

第二层是授权与代表人:外国企业出具的对有权签字人的授权或者证明文件,对首席代表、代表的任命文件,以及他们的简历和身份证明。第十一条要求外国企业委派一名首席代表,首席代表在外国企业书面授权范围内可以代表外国企业签署代表机构登记申请文件;同时可以根据业务需要委派一至三名代表。所以那份授权文件不是形式件,它是让申请书上的签字有效的那件东西。

第十二条把几类人挡在门外:因损害中国国家安全或者社会公共利益被判处刑罚的;因从事这类违法活动,其所在代表机构被撤销设立登记、吊销登记证或者被责令关闭,且自该日起未逾五年的;以及国务院市场监督管理部门规定的其他情形。

第三层是驻在场所:对场所的合法使用权。第十五条允许外国企业自行选择驻在场所,同时允许有关部门根据国家安全和社会公共利益需要要求其调整,并及时通知登记机关。第十八条要求把登记证置于驻在场所的显著位置。所以场所材料干两件事:证明有权占用,以及固定登记证上将载明的那个地址。

有三项要求横跨三层。外文文件须附中文翻译,并附翻译单位或者翻译人员相关信息;依法须经批准才能设立的,要提交批准文件,并自批准之日起 90 日内申请设立登记;设立之后,外国企业须在登记机关指定的媒体上公告。

有一项值得单独提,因为它常被漏掉:按照中国缔结或者参加的国际条约、协定设立从事营利性活动的代表机构,还应当依法提交相应文件。第十三条那个例外是真实存在的,但它靠文件证明,不靠主张。

名称、证件与登记簿

从卷宗到代表机构的日常之间,隔着三条条文,通常被读一遍就忘了。

第一条是名称。第十条把构成和次序都规定了:外国企业国籍、外国企业中文名称、驻在城市名称,以及”代表处”字样。次序不是文风偏好。它意味着这个名字按顺序说明:母公司来自哪里、是谁、办事处驻在哪座城市、这个东西是什么性质。三类内容和文字被禁止:有损中国国家安全或者社会公共利益的;国际组织名称;法律、行政法规或者国务院规定禁止的。

代表机构应当以登记机关登记的名称从事业务活动。第三十八条把偏离这一条定为违法,责令限期改正,处 1 万元以上 3 万元以下罚款;逾期未改正的,吊销登记证。实务上这条规则咬人的地方往往出人意料:一家集团在全球统一用一个品牌名做市场,又让代表机构在门头或者文件上使用那个名字、省掉了登记的中文后缀,就是在没有主观故意的情况下偏离了登记名称。

第二条是证件。发两份:代表机构的登记证,以及首席代表、代表的代表证。第十八条要求把登记证置于驻在场所的显著位置。第十九条禁止伪造、涂改、出租、出借、转让这两种证件,并规定遗失或者毁坏的,应当在指定媒体上声明作废、申请补领。同一条还规定:登记机关依法作出准予变更登记、准予注销登记、撤销变更登记、吊销登记证决定的,原登记证和原代表证自动失效——这句话提醒的是,证件是当前状态的证明,不是一张永久的许可。

第三条是登记簿。第十七条要求登记机关把代表机构的登记事项记载于代表机构登记簿,供社会公众查阅、复制。这里有一处容易忽略的实务后果:交易对手、银行、法院去核的,是登记簿,不是墙上那张证。登记事项没有更新的代表机构,在登记簿上就是被错误描述的,哪怕手上的证看起来没问题——因为证跟着登记簿走,不是登记簿跟着证走。

与这两份证件相关的公告义务有两项,而且不对称。第二十条规定,代表机构设立、变更,由外国企业在登记机关指定的媒体上向社会公告;代表机构注销、或者被依法撤销设立登记、吊销登记证的,由登记机关进行公告。这种不对称是合理的:企业对”我来了”有公告的动力,对”我走了”未必有。

这两份证件还是所有下游手续的钥匙。第二十五条规定,代表机构、首席代表和代表凭登记证、代表证申请办理居留、就业、纳税、外汇登记等有关手续。就这一句,解释了证件为什么不止于悬挂。没有它们,首席代表无法办妥居留,代表机构无法完成税务登记,让母公司向办事处拨付经费的那笔外汇登记也开不出来。登记不是若干合规步骤里的一步,它是让其余步骤得以成立的那一步。

三条认证路径

母公司文件的认证,是卷宗最常卡住的地方,而 2023 年之后,规则实质变了。现在是三条路径,走哪条取决于母公司在哪儿。

第一条适用于《取消外国公文书认证要求的公约》缔约国的主体。中国于 2023 年 3 月 8 日加入该公约,公约于 2023 年 11 月 7 日对中国生效。自该日起,缔约国出具、拟在中国内地使用的公文书,只需办理该国主管机关签发的附加证明书,无需再办中国驻当地使领馆的领事认证。中国外交部是公约规定的附加证明书主管机关,并委托相关省、自治区、直辖市人民政府外事办公室以及部分市人民政府外事办公室,为本行政区域内出具的公文书签发附加证明书。中国的附加证明书采用贴纸形式,加贴银色国徽印鉴,支持在线核验。

第二条适用于非缔约国的主体,以及虽为缔约国但中国不承认其具有主权国家地位、或者与中国之间不适用公约的情形。这些主体的文件,须经所在国公证机关公证,再经中国驻该国使领馆认证。本国与中国没有外交关系的,链条更长:先经与我国有外交关系的第三国驻该国使领馆认证,再由中国驻该第三国使领馆认证。文件出自海外属地的,先在该属地办妥公证,再经该国外交机构认证,最后由我国驻该国使领馆认证。

第三条适用于香港、澳门和台湾地区的投资者。其主体资格文件按专项规定或者协议,提供当地公证机构的公证文件;香港、澳门的公证文件须经司法部专门机构转递,经该机构电子化流转的可以免予提交纸质件。条例第四十四条也规定,香港特别行政区、澳门特别行政区和台湾地区企业在中国境内设立代表机构的,参照本条例进行登记管理。

由此有两个实务判断。

其一,走哪条路由母公司的法域决定,不由申请人的方便决定。一家母公司在新加坡、中间夹一家开曼控股公司的集团,得先定”哪家主体是登记意义上的外国企业”,而这个决定又决定了哪份文件走哪条路。把某一法域的公证文件拿去替代另一法域的附加证明书,是认证环节最常见的错误。

其二,2023 年的变化并没有缩短每一条链。它对缔约国是去掉领事认证这一步,链条变短;对非缔约国,链条和以前一样长,没有外交关系时第三国认证那一段仍然要走。把”中国现在适用附加证明书”读成一次普遍简化,恰恰会在本来就慢的那几个法域上失望。

它能做什么,不能做什么

关于代表机构最常被问的一个问题是:它能不能签合同、能不能收费。答案在第十三条、第十四条和第三十七条里,比问题所暗示的要窄。

代表机构可以做与外国企业产品或者服务有关的市场调查、展示、宣传;可以做与产品销售、服务提供、境内采购、境内投资有关的联络。这就是被允许的全部。宣传意义上的市场推广在里面;销售不在;收费提供服务不在;以自己名义取得收入不在。

越界的后果是分层的。第三十五条规定,代表机构违反规定从事营利性活动的,责令改正,没收违法所得,没收专门用于从事营利性活动的工具、设备、原材料、产品等财物,处 5 万元以上 50 万元以下罚款;情节严重的,吊销登记证。第三十七条规定,违反第十四条规定从事业务活动以外活动的,责令限期改正,逾期未改正的,处 1 万元以上 10 万元以下罚款;情节严重的,吊销登记证。

两条并读有信息量。营利性活动被处理得比超范围活动更重,说明”不得营利”被当作这种形态的内核,而不是一条技术性限制。

另一个相关问题是能不能雇人。能,但不是通常意义上的用人单位——它没有法人资格,不能作为劳动合同的一方。条例要求首席代表、代表以及工作人员遵守法律、行政法规关于出入境、居留、就业、纳税、外汇登记的规定;第二十五条规定代表机构、首席代表和代表凭登记证、代表证申请办理居留、就业、纳税、外汇登记等有关手续。用工结构因此建立在服务安排之上,而不是直接雇佣之上,那属于另一套实务。

第三个问题是能不能持有资产。母公司拨付的资金、自身需要的设备,它可以持有,第七条要求会计账簿把这两样都真实记下来。它不能做的是使用别的企业、组织或者个人的账户——这条针对的是把代表机构成本走到关联公司账上的做法。在同一座城市跑好几家代表机构的集团,账目上的分开必须是真分开,不能只是名义上分开。

四个时钟:设立、变更、年报、退出

这套制度里全是期限,而期限是最容易漏的部分,因为它们分散在四章里。

设立上,登记机关应当自受理申请之日起 15 日内作出是否准予登记的决定,作出决定前可以根据需要征求有关部门意见。准予登记的,自决定之日起 5 日内颁发登记证和代表证;不予登记的,5 日内出具登记驳回通知书,说明理由。依法须先经批准的,应自批准之日起 90 日内申请设立登记。

变更上,第二十六条要求登记事项发生变更即申请变更登记,第二十七条给时钟:自变更之日起 60 日内;变更事项依法须在登记前经批准的,自批准之日起 30 日内。第二十八条处理驻在期限届满而要继续经营的情形:应在期限届满前 60 日内申请。登记机关应当自受理之日起 10 日内作出决定,准予的自决定之日起 5 日内换发登记证和代表证。

第三十一条另设了一项备案义务,与变更登记不同:外国企业的有权签字人、企业责任形式、资本(资产)、经营范围以及代表发生变更的,应自变更之日起 60 日内向登记机关备案。登记与备案的区分值得记牢,因为两者的触发点不同,后果也不同。

年报上,第六条要求代表机构于每年 3 月 1 日至 6 月 30 日向登记机关提交年度报告,内容是指定的:外国企业的合法存续情况、代表机构的业务活动开展情况,以及经会计师事务所审计的费用收支情况。审计这一项是关键。没有单独建账的代表机构,出不了自己那份经审计的费用收支表,年报就从一项行政事务变成一项合规问题。

退出上,第三十二条列了四种情形,各配 60 日:外国企业撤销代表机构;驻在期限届满不再继续从事业务活动;外国企业终止;代表机构依法被撤销批准或者责令关闭。第三十三条要求随申请提交税务登记注销证明、海关出具的相关事宜已清理完结或者该代表机构未办理相关手续的证明,以及国务院市场监督管理部门规定提交的其他文件。登记机关应当自受理之日起 10 日内作出决定。

漏掉这些时钟的后果,有两条值得写出来。第三十八条规定,未依条例提交年度报告、未按登记的名称从事业务活动、未按有关部门要求调整驻在场所、未依条例公告设立与变更情况、未依条例办理变更登记或者注销登记或者备案的,责令限期改正,处 1 万元以上 3 万元以下罚款;逾期未改正的,吊销登记证。第三十九条规定,从事危害中国国家安全或者社会公共利益等严重违法活动的,吊销登记证,且自吊销之日起 5 年内,设立该代表机构的外国企业不得在中国境内设立代表机构。

动笔之前的一张核对表

第一次设代表机构的集团,按下面这个次序走,能避开大部分本可避免的延误。

先确认将作为”外国企业”的那家主体存续已满两年,并确认其所在法域出得了证明这件事的文件。在动任何材料之前先定这家主体是谁,因为这个选择决定了认证路径;在有中间控股层的集团里,这个选择并不显然。

逐份文件把认证路径画出来。缔约国:按该国要求做公证,加附加证明书。非缔约国:公证加使领馆认证,没有外交关系时走第三国那一段。香港、澳门、台湾:当地公证,港澳另需司法部专门机构转递。

中文翻译和翻译单位信息,与原件同时准备。带着未翻译文件到窗口,不是”快好了”,是”还没开始”。

定首席代表和代表,并确认书面授权覆盖申请书上的那次签字。代表人数上限是三名,任命要按规范要求的文件形式落地。

落定驻在场所和证明有权使用的文件。确认登记证上将载明的就是那个地址,并且登记证能在那里显著悬挂。

把四个时钟记进日程:设立前置批准之后的 90 日;变更的 60 日与 30 日;每年 3 月 1 日至 6 月 30 日的年报窗口;注销的 60 日。

关于这种形态本身,还有两点。

其一,代表机构不是可以”转成”公司的跳板。没有转换程序,因为没有可转换的东西:没有法人、没有资本、没有股权。以后想要公司,就另设一家公司;愿意的话,把代表机构注销。两者是并行或者先后,不是变形。

其二,代表机构的文件负担是前置的,日常负担相对轻。几乎全部难度集中在第一次报送和每年的审计上。把第一份卷宗做对的集团,以及从第一个月就把代表机构的账与母公司分开记的集团,后面几年会过得平静。把第一次报送当成”窗口收什么就给什么”的集团,后面几年会一直在解释当初的缺口。

来源

  • 中国政府网·《外国企业常驻代表机构登记管理条例》(国务院令第 584 号,2010-11-19 公布,2011-03-01 施行):https://www.gov.cn/zhengce/2010-11/25/content_2602568.htm
  • 中国政府网·国务院公报版《外国企业常驻代表机构登记管理条例》(含 2013 年、2018 年修订表述):https://www.gov.cn/gongbao/content/2019/content_5468900.htm
  • 中华人民共和国外交部·《取消外国公文书认证要求的公约》将于 2023 年 11 月 7 日在中国生效实施:https://www.fmprc.gov.cn/wjbxw_new/202310/t20231023_11165858.shtml
  • 中国领事服务网·领事认证(附加证明书)简介:https://cs.mfa.gov.cn/zggmcgx/lsrz/jj/
  • 中国领事服务网·《取消外国公文书认证要求的公约》缔约国名单:https://cs.mfa.gov.cn/dygmd/202607/t20260724_11991225.html
  • 青岛市企业登记服务网·外国(地区)企业常驻代表机构设立登记办事指南(依据国市监注发〔2022〕24 号附件 3):https://qydjfw.qingdao.gov.cn/files/2026/08/20260821165015218.pdf

Professional Business Consulting for International Companies in China. | ylinsight | YINGLONG INSIGHT