01The statute — what Article 585 actually says
The 2021 PRC Civil Code, Article 585 (paraphrased):
If the agreed liquidated damages are lower than the actual loss caused, the People\u2019s Court or arbitration tribunal may, on a party\u2019s application, increase the amount; if the agreed liquidated damages are manifestly higher than the actual loss caused, the Court or tribunal may, on application, appropriately reduce the amount.
The keyword is "manifestly higher than" — not just "higher than".
LD above actual loss is permitted (freedom of contract). Only "manifestly higher" gets cut. The whole question becomes: what counts as "manifestly"?
02How "manifestly" is read in practice — the 130% rule of thumb
The old Contract Law interpretation said anything above actual loss × 130% was "manifestly higher" — that is the origin of the "30% cap" shorthand.
Civil Code does not name a specific ratio, but lower courts still apply these benchmarks:
- Total LD ≤ actual loss × 130%: safe band, courts almost never reduce.
- Actual loss × 130%–200%: at risk of reduction — bring loss evidence.
- Actual loss × 200%+: high probability of being cut back to ~130%.
So the real question is not "how high?" — it is "how well can I prove actual loss?" combined with "is my LD inside 130% of that?".
03How the sweet-spot numbers are derived (three LD types)
Different breach types use different formulas:
Derivation: for a US importer at 25–30% gross margin, one day of delay produces roughly 0.5–1% loss of contract value (line stoppage, missed slot, expedited air). 0.30–0.50% falls inside actual loss × 130%.
Critical: AQL clauses (Major 2.5 / Minor 4.0) must be the objective trigger. Without them, "non-conforming" becomes an endless argument.
Loss is hard to prove on these, so PRC courts tend to accept the contract default — provided the number is not absurd (under ~200% of estimated loss).
04Writing it too low also backfires
Article 585 cuts both ways: LD lower than actual loss can be increased by the court — but litigation cost may exceed the gain.
Common bad drafting:
- "0.05% per day": below industry baseline; court may raise it, but the case cost more than the recovery.
- "RMB 100 per day": a fixed amount becomes meaningless on a large PO — effectively no LD at all.
- "Breaching party shall compensate the non-breaching party for actual loss": dumps the entire evidentiary burden on you. Lose rate is high.
Right framing: write LD at 100–130% of actual loss. Skip the evidentiary fight, win fast, enforce easily.
05Tool — get your sweet-spot number in 5 seconds
Plug in monthly order value, expected delay days, and product category. The tool:
- Derives the "actual loss" baseline for your industry gross margin.
- Returns a defensible % / day range with a red-yellow-green rating.
- Saves multiple scenarios and shows a comparison chart for supplier negotiation.
These clauses only bite when they sit inside a bilingual agreement a PRC court will actually enforce. We give you the 17-point risk self-check, the LD sweet-spot calculator, and the PI vulnerability scanner.
Try the three tools on the main pageFAQ
Q1. Higher LD should scare the factory more, right?+
What scares them is enforceable LD. Write it too high and a PRC court cuts it back to ~130% of actual loss in litigation — net recovery is lower than a reasonable number, and the factory questions whether you really want a long-term relationship.
Q2. Can we copy a Taiwan or Hong Kong contract LD clause straight in?+
Not advisable. TW Civil Code Article 252 and HK common law LD rules differ from PRC Civil Code — especially on "reasonable pre-estimate of loss". TW / HK templates often produce numbers too high and get cut in PRC court.
Q3. If the factory has no assets, doesn’t any LD number become moot?+
Exactly why LD only works combined with (1) supplier due diligence on real assets, (2) payment structure with hold-back, and (3) venue at the factory’s local court. LD on paper is half the system; these three are the other half.
Q4. Does the "30% cap" still apply?+
The old Contract Law interpretation 30% is superseded by Civil Code’s "manifestly higher than actual loss" standard. 130% remains the working consensus in practice, but anchor on actual loss — do not apply 30% mechanically.
More in this series
An NDA only blocks disclosure. An NNN also blocks copying and back-door selling to your customers. In PRC courts, that gap is the whole game.
You signed the PI and wired the deposit, but the factory can still walk. Here are 12 gaps you must move into a bilingual master agreement.
Once the tool sits in their plant, leverage flips. How to draft it right up front, and how to claw it back when the factory refuses to release it.