China Rewrote Its Cosmetics Law in 2021, and New Ingredients Started Moving Again

Ten in a decade

Between 2004 and 2014, the Chinese authorities approved ten new cosmetic ingredients. Not ten per year. Ten across the whole decade, in a market that had by then become one of the largest on earth.

That single figure explains more about Chinese skincare than any amount of commentary about consumer taste. If you cannot realistically get a new ingredient approved, you formulate from the inventory of ingredients that already exist, which everyone else also has. Differentiation stops being available in chemistry and moves entirely into packaging, price and story. An industry structured that way will produce marketing-led brands, because the regulation left no other lever to pull.

The first signal that something was changing came on 28 December 2020, when the National Medical Products Administration approved four new ingredients in a single day — as many in one announcement as the system had sometimes managed in years. Four days later, the regulation that made the change structural took effect.

What followed is a natural experiment, and the numbers are checkable. Across 2021 to 2024, China recorded 207 new cosmetic ingredient notifications, including those later withdrawn or cancelled, with a record 90 in 2024 alone. Cumulative filings had passed 272 by the first half of 2025, and more than a hundred further ingredients completed filing in the first half of 2026 alone. From ten per decade to roughly a hundred per year, with no change in the underlying chemistry available to anyone.

What CSAR replaced, and why the old framework was the bottleneck

The Regulations on Supervision and Administration of Cosmetics — CSAR — took effect on 1 January 2021, replacing the Cosmetics Hygiene Supervision Regulations that had governed the industry since 1989.

Consider what 1989 means in this context. That framework was written for a country whose domestic cosmetics industry barely existed and whose imports were negligible, and it was still the operative law when China became the world's second-largest beauty market. Every subsequent rule was a patch on a foundation designed for a different country.

CSAR's central move was to stop treating all cosmetics and all ingredients as one undifferentiated risk category. Finished products split into special cosmetics, which require registration — that is, pre-market approval — and general cosmetics, which require only filing. Sunscreens, hair dyes, perming products, freckle-removing and whitening products and products making new efficacy claims sit on the registration side. Ordinary cleansers, moisturisers and serums sit on the filing side.

The same logic was applied to ingredients, and that is where the throughput came from. New cosmetic ingredients are tiered by risk. The high-risk tier — preservatives, UV filters, colorants, hair dyes, and freckle-removing or whitening agents — still requires full registration and approval. A relatively high-risk tier covering anti-hair-loss, anti-acne, anti-wrinkle, anti-dandruff and deodorant functions requires notification. Everything else, which is most things, sits in a low-risk tier requiring only notification through an online dossier.

That is the whole reform in one sentence: the system stopped making a new humectant queue behind a new hair dye. It is unglamorous administrative engineering, and it moved an order of magnitude more material through the pipe than any amount of exhortation would have.

The three-year monitoring period, which just closed its first loop

The reform was not a deregulation, and this is the part usually left out of both the celebratory and the sceptical accounts.

An ingredient that completes registration or notification does not simply join the permitted list. It enters a compulsory three-year safety monitoring period, running from the date the first cosmetic using it completes its own registration or filing. During those three years the registrant or notifier must submit annual reports to the NMPA on the ingredient's use and its accumulated safety data.

At the end of the period the authority evaluates. No safety issue, and the ingredient is added to the Inventory of Existing Cosmetic Ingredients in China — the standing list that anyone may then formulate from freely. A safety problem, and the registration certificate or filing number is cancelled.

This is a genuinely well-designed mechanism, and it is worth saying so plainly. It front-loads market access and back-loads the evidentiary burden, which is close to the opposite of the American monograph model, where the burden is front-loaded so heavily that for two decades nobody attempted it. It also creates something regulators rarely obtain: a real-world usage and adverse-event dataset on a novel ingredient, gathered while the ingredient is actually in commerce rather than in a trial.

And the loop has now closed for the first time. On 19 August 2026 the NMPA announced that four new cosmetic ingredients had completed the three-year monitoring period without safety concerns and were being added to the inventory. Nine days before this article. That is what a functioning pathway looks like from the far end, and it is the first evidence that the back half of the design works rather than just the front.

The requirement America does not impose

There is one element of CSAR that consistently surprises people who assume the direction of regulatory strictness runs one way, and it deserves its own section.

Under the CSAR framework, efficacy claims must be substantiated, with evaluation summaries published on the NMPA's platform. If a product claims to brighten, to firm, to soothe, to reduce the appearance of wrinkles, there is meant to be an evaluation behind it and a summary a member of the public can go and look at.

The United States imposes no equivalent. A cosmetic — as distinct from an OTC drug — may make performance claims without filing efficacy evidence with anyone. MoCRA, the 2022 overhaul, introduced a duty to adequately substantiate safety and to hold the supporting records, which is a meaningful change. It did not introduce an efficacy substantiation duty, and it created no public register of claim evidence.

So on the specific question of whether a brand must be able to show its work for the promise on the front of the box, the Chinese framework asks for more than the American one. That is not a comfortable finding for anyone whose model of global cosmetic regulation is a straight line from strict Europe to lax everywhere else, and it is exactly the kind of finding worth stating without softening.

It is also, to be fair to the American system, a requirement whose value depends entirely on enforcement quality and on what counts as an acceptable evaluation — a published summary is only as good as the study behind it. But the structural obligation exists in one place and not the other, and that is checkable.

Why an open pathway shows up in formulas within two cycles

The transmission from regulation to product is faster than people expect, because formulation is a slow art practised by teams who keep long wish-lists.

When the pathway is blocked, a formulator's options are the existing inventory, and every competitor has the same options. When it opens, the wish-list gets filed, and within roughly two product development cycles — call it two to three years — the new material appears in shipping formulas. That timeline lines up almost exactly with when Chinese domestic formulation started attracting serious attention outside China.

Look at what the category is now shipping and the character of it is different from the era of botanical-led hydrating creams. ↗Proya Source Force Repairing Essence is built on bifida ferment lysate with a multi-ceramide and cholesterol system; ↗Proya Ruby Wrinkless and Firming Essence combines retinol with niacinamide, peptides and ceramides in one formula, which is a formulation-stability problem before it is a marketing one; ↗Proya Double Effect Brightening Essence runs a niacinamide and tranexamic acid pairing alongside ergothioneine and ectoin. ↗Skintific 5X Ceramide Barrier Repair Moisture Gel is a barrier formula competing on lipid architecture.

None of that is exotic by international standards, and that is precisely the point. These are competent, current formulations of the kind that were structurally difficult to produce in a market where the ingredient list was frozen. ↗Niacinamide and ↗peptides are not new to anyone — but a formulation culture that can file what it needs behaves differently from one that cannot, and it shows up in the stability engineering and the combinations long before it shows up in a novel molecule.

If you want to check any of this yourself rather than take it on trust, our ingredient decoder will read a panel from any market and tell you what is actually in it.

The read that keeps both eyes open

It would be easy to file this as a straightforward good-news story about a regulator modernising, and that would be a naive edit.

The same reform package that raised the claims bar also sits inside an explicit industrial strategy. The July 2026 announcement widening animal-testing exemptions also carried measures encouraging domestic first launches of new products. A regulator that simultaneously polices an industry and is tasked with helping it win has an interest, and interests shape enforcement even when rules are written well. Anyone who would flag that dynamic in another country should flag it here.

The positive-list architecture is also a real constraint, not just an enabler. An ingredient not on the inventory cannot be used until it has been through the pathway, and the highest-risk tier — which includes every UV filter — still requires full registration and approval. A faster gate is still a gate.

And the numbers, while checkable, measure filings rather than quality. A hundred notifications a year tells you the pipe is open. It tells you nothing about whether the material coming through is interesting, and a good share of any country's new ingredient filings are incremental derivatives of things that already exist.

What survives all of that is the structural claim, and it is the one worth keeping. C-beauty's formulation turn is not primarily a story about national sentiment or about consumers deciding to buy local. It is a story about an administrative bottleneck being removed and an industry immediately doing what industries do when a constraint lifts. That explanation is duller than the cultural one, it is supported by a countable series, and it makes a testable prediction: if the pathway narrows again, the formulation will follow it down within two cycles. Watch the filing numbers rather than the marketing.