The China discount reversed in dollars. In yuan it never did.

The price American importers are quoted for Chinese goods fell to the lowest level in the twenty-two years the series has run, bottoming at 96.9 in July 2025. It has since recovered to 98.9, which reads as the exporter discount unwinding. It is not. Over those same eleven months the yuan strengthened 5.6% against the dollar — more than enough to account for the entire rise. Translated back into the exporter’s own currency, the price kept falling.

The series is the BLS Import Price Index by Origin for China (CHNTOT), monthly, not seasonally adjusted, indexed so that December 2003 equals 100. The newest observation is June 2026, published with a lag of roughly six weeks — considerably fresher than the customs trade flows, which run 89 days behind.

One property of the series does the work in everything below. BLS collects import prices as transaction prices — generally f.a.s. or f.o.b., depending on the practice of the industry — and excludes duties — the indexes are built to deflate the trade component of the national accounts, which counts tariffs as taxes rather than as part of the value of goods. BLS has published a separate note making the same point. CHNTOT is therefore not the landed cost of a Chinese good in an American warehouse. It is the price the Chinese exporter charges before any US duty is applied — the one public series that speaks to whether the exporter, rather than the importer, is giving something up.

The fall through 2025 was the steepest in the series

The index stood at 100.3 in December 2024 and fell in every one of the next seven observations — 99.9, 99.8, 99.3, 98.5, 97.7, 97.6 — to 96.9 in July 2025, a decline of 3.4%. That reading is the lowest of the 270 observations since the series began in December 2003.

The twelve-month rate gives the scale. The year to July 2025 shows −3.49%, the steepest twelve-month decline in 258 such comparisons — steeper than anything in the financial crisis, where the worst reading was −3.19% in the year to July 2009. The four most severe twelve-month declines in the history of the series all fall in 2025.

In dollars, the recovery is real and the run of declines has ended

The trough held for one month. From August 2025 the index rose in seven of the nine steps that followed, was flat in one and fell in one: 97.0, 97.1, 97.1, 97.2, 97.3, 97.5, 97.9, 97.7, 98.0, and 98.9 in June 2026.

The twelve-month rate ends a much longer run. It had been negative in 39 consecutive readings, from January 2023 through April 2026 — the last positive reading before that was December 2022, at +0.77%. It crossed back above zero in May 2026 at +0.31% and reached +1.33% in June, the 77.5th percentile of every twelve-month change the series has recorded.

The single-month move needs care, because these figures are not seasonally adjusted. Testing it against its own calendar month settles that: across the 23 May-to-June transitions in the series the mean change is +0.04 index points and the median is zero, with a standard deviation of 0.30. June 2026’s +0.90 is the largest May-to-June move of the 23, ahead of 2008’s +0.60. Whatever moved the index in June, the calendar is not it.

In yuan, nothing reversed

CHNTOT is denominated in US dollars. The yuan did not hold still underneath it.

On the monthly CNY/USD rate, the yuan went from 7.1741 per dollar in July 2025 to 6.7758 in June 2026 — a strengthening of 5.6%. Had Chinese exporters left their yuan prices completely unchanged, that move alone would have raised the dollar price of their goods by 5.9%. The observed rise was 2.1%.

Dividing one by the other gives the change in the exporter’s own currency:

Across the full eighteen months the yuan strengthened 6.9% against the dollar, which should have carried dollar-denominated Chinese export prices up 7.5%. They fell 1.4% instead. In yuan terms the price of Chinese goods sold to the United States is down 8.2% since December 2024, and the cutting did not pause when the dollar index turned in May.

This is a derived figure, not an observed one, and it should be read as a translation rather than as an invoice. Most US-China trade is contracted in dollars, so a yuan appreciation reaches the exporter’s margin on the currency’s timetable rather than the contract’s.

Measured against everything else America imports, Chinese goods stay cheap

Across the whole tariff era — January 2018, before the first Section 301 round, to June 2026 — the pre-duty dollar price of Chinese goods fell 1.2% while the price of everything else the United States imports, fuel excluded, rose 15.4%. The China index has been below its December 2003 base in every observation since January 2025 and sits at the 21.5th percentile of its own history. Its all-time high, 105.5, was set in February 2012.

That is what the aggregate series could not deliver. The all-commodities index alone cannot separate exporters cutting margin from buyers shifting to cheaper origins from composition change. Fixing the origin removes the second of the three.

The newest tariff is not in this data

The rate applying to Chinese goods moved four times in eighteen months. The IEEPA tariffs were held unlawful by the Supreme Court on 20 February 2026 in a 6-3 decision (CRS LSB11398). A 10% global tariff under Section 122 of the Trade Act of 1974 took effect on 24 February and lapsed on 24 July. USTR’s forced-labour Section 301 action then set two tiers, 10% and 12.5%, across the top 60 US trading partners (USTR fact sheet). China was placed in the higher tier. CBP’s implementing guidance is explicit: “articles the product of China will be assessed an additional ad valorem rate of duty of 12.5%” under HTSUS 9903.05.31, for entries on or after 12:01am eastern time on 24 July 2026 (CSMS #69326983).

June 2026 data ends five weeks before that duty applied. Nothing above reflects it. The July reading, due in mid-August, is the first observation that can.

The Stress Index reads 47.2 as of 28 July 2026 — the Normal band. Its global trade activity component sits at 10.4, the calm band, on an input observed 88 days ago. The volume side of world trade is registering as quiet while the price side at the Chinese border does something the headline dollar figure hides.

What this cannot settle

Composition. CHNTOT prices a basket of goods imported from China, and holding the origin fixed does not hold the basket fixed. If dearer items dropped out of Chinese sourcing through 2025 and have since returned, the index would trace the path above without any exporter changing a single quoted price. Separating a price movement from a mix shift needs item-level detail the published series does not carry.

Nor does a falling price prove a tariff is the cause. Weak domestic demand, overcapacity and competition for non-US buyers would all push the same direction, and none of them are separable here.

One gap in the record: October 2025 is blank in all three import-price series, so a month inside the tariff window is simply unobserved.

What to watch

The July 2026 CHNTOT reading, due mid-August. It is the first observation covering the 12.5% Section 301 duty, and the first test of whether the pattern above survives it.

The dollar reading against the CNY/USD rate, which averaged 6.7758 in June 2026 and stood at 6.7719 on 24 July. The two have been moving in opposite directions since July 2025. If the yuan stops appreciating and the index keeps climbing, the recovery becomes a genuine price increase rather than a translation effect.

The spread against IREXFUELS, 4.24% versus 1.33% over twelve months. It measures how far China-origin pricing is diverging from the rest of the US import basket, and it narrowed by roughly a third over the past year.

Correction, 14 October 2026: An earlier version said the 10% Section 122 tariff took effect on 24 June. The proclamation imposing it (91 FR 9339) made it effective on 24 February 2026, and it lapsed on 24 July. The sentence now gives both dates.