September 15, 2026
64,000 units separate the two versions of China’s August NEV retail figure. The preliminary number the CPCA published on September 3 was 1.069 million. The final release on September 8 was 1.005 million, a revision of about 6% in the direction that cuts against the story the first number generated. The preliminary figure is the one that circulated for five days, and it is the one most readers still have.
65.2% is the record penetration rate both versions produce, and 10.1% is how much NEV retail fell year over year to produce it. Total passenger-vehicle retail fell 23.6% to 1.541 million. A ratio rises when its denominator falls faster than its numerator, and the CPCA itself attributes the record partly to a near-halving of gasoline car sales. The record is real. So is the eighth consecutive monthly decline in the numerator, and a reader who only saw the first number has the record without the revision that sits underneath it.
698,000 battery-electric units, up 0.8% year over year, was the only powertrain category in China to grow at all in August. Plug-in hybrids fell 29.6% to 226,000. Extended-range vehicles fell 22.2% to 81,000. A market described as electrifying is one where two of its three electrified categories contracted by more than a fifth, and the third grew by less than one percent. The composition of the record matters as much as the record.
Four Series, One Country
888,000 passenger cars left China in August, up 77.8% year over year, and 518,000 of those were NEVs, up 154.7%. NEVs are now 58.4% of Chinese passenger-car exports, 18 percentage points higher than a year ago, and exports account for roughly a third of the 1.51 million NEVs Chinese plants shipped at wholesale in the month. GCBC has made the argument that domestic contraction and the export surge are one event. What August adds is scale. BYD led domestic NEV retail at 233,943 units, with Geely at 110,560 and Leapmotor at 84,874. Tesla China recorded 50,047.
60.6% is the August NEV share the CAAM published on September 10, against the CPCA’s 65.2%. Neither is wrong. The CAAM counts wholesale across all vehicle types including exports, where NEVs were 52.1% of what left the country, and the CPCA counts domestic passenger retail. Anyone comparing a Chinese penetration figure to a European or American one is usually comparing two of at least four available series, and the gap between the two Chinese numbers alone is 4.6 percentage points before any revision is applied.
6.674 million NEVs year to date, down 12.1%, against total passenger retail of 11.716 million, down 20.8%, is the same arithmetic run across eight months rather than one. The penetration rate is not an artifact of a single weak August. It has been produced all year by a gasoline market contracting faster than the electrified one. A share figure rising while both of its inputs fall describes a market shrinking unevenly, and the revision layer means even the inputs are provisional for a week after they appear.
The Same Week, the Same Problem, a Different Market
207.4 was the Manheim Used Vehicle Value Index at mid-August, down 1.2% from July and flat against August 2025. The finalized August index published September 8 came in at 208.2, down 0.9% from July and up 0.4% year over year. Non-adjusted prices went from down 0.2% year over year at mid-month to up 0.2% in the final. That is a sign change on a headline figure, produced entirely by the second half of the month.
Every component of the Manheim revision moved upward, and the China revision moved downward. The point is not that either was wrong. Mid-month and preliminary reads are honest estimates on partial data and they are published as such. The point is that the automotive calendar publishes most monthly series twice, the first read is the one that gets quoted, and nothing in the ordinary reporting cycle goes back to check.
1.8% is how far Manheim’s three-year-old index fell in August against a typical 0.6% for the month, while Cox chief economist Jeremy Robb noted nine and ten-year-old cohorts holding six percentage points above their long-term average. GCBC has covered that age spread as an affordability signal. The revision layer is what is new: the mid-month read understated the strength of the market in the back half of August, and the spread was wider than the first estimate implied.
71.5% is the NEV penetration rate the CPCA reported for September 1 to 6, on retail volume of 210,000 units, down 19% year over year, with NEV retail at 150,000, down 3%. That is a weekly preliminary and it will move. The full-month preliminary lands around October 1 and the final about a week later. Manheim’s complete September suite publishes October 7. In the one week examined here, one first read was 6% high and the other understated a year-over-year change by enough to reverse its sign. Anyone building a forecast off the first number should budget for the second.
