China’s private PMI survey provides a monthly view of manufacturing and services conditions. Previously known as Caixin PMI, it is now published under the RatingDog name. Learn to compare its components with the official NBS survey and examine what each says about demand relevant to China’s trading partners.

First check the manufacturing headline: 60×30%+50×25%+50×20%+50×15%+50×10%=53. Output remains 50, a separate component rather than another headline name.

Do not substitute headline for Output

S&P describes composite as GDP-weighted manufacturing output and services activity. Our weights expose the substitution error.

Scroll sideways to see the full diagram

Do not substitute headline for OutputFictitious Manufacturing Output 50 weighted 30% contributes 15. Services Business Activity 60 weighted 70% contributes 42. Composite 57. Manufacturing Headline 53 is the wrong input; substituting it gives 57.9.Composite needs the correct inputAll readings and sector weights are fictitiousManufacturing: Output 50Model weight: 30%Contribution: 15 pointsServices: Activity 60Model weight: 70%Contribution: 42 points15+42=57Manufacturing Headline 53 differs.53×30%+60×70%=57.9This uses the wrong input.
Our simultaneous monthly indices belong to one PMI family. Manufacturing: new orders 60; Output, employment, inverted deliveries and stocks 50. Headline weights 30/25/20/15/10 give 53. Sector weights 30/70 are invented, not Chinese GDP shares. Bars share one scale; indices are not production percentages. Method: S&P Global Output and composite; headline weights: S&P Global FAQ.
Correct weights cannot repair an incorrect index selection.

Copy the full field name before comparison. Simple mean 50, 60 gives 55 but ignores assumed sector shares. Neither 57 nor 55 means production grew 7% or 5%.

What the Caixin PMI actually is

PMI stands for Purchasing Managers’ Index. Caixin announced that its title sponsorship would end in July 2025, with the early-July edition its last. Later S&P Global releases use RatingDog China PMI. Purchasing managers report changes in several aspects of business since the previous month; current activity and a separate future-expectations question are different measures.

The series includes manufacturing and services surveys, plus a composite output index combining their activity measures. The composite covers the surveyed sectors rather than every part of the economy. The manufacturing headline is a weighted combination of new orders, output, employment, supplier delivery times and stocks of purchases; it does not measure the value of output in yuan.

The manufacturing headline uses new orders 30%, output 25%, employment 20%, inverted supplier delivery times 15% and stocks of purchases 10%. Composite Output combines manufacturing output and services activity using sector weights; do not average two headline names or substitute manufacturing headline for Output.

The 50 line — the border between expansion and contraction

For one hypothetical diffusion question, assume response shares after applicable weights: 35% higher, 40% unchanged and 25% lower. The index is 35 + 0.5 × 40 = 55 points. With 30% higher, 40% unchanged and 30% lower it is 50, although fewer than half report growth. These simple calculations precede seasonal adjustment and do not reconstruct the full manufacturing headline. A value of 55 does not mean output grew 5% or 55%.

Keep three comparisons separate. Our fictional current index 51.0, previous 50.5 and forecast 52.5 gives +1.0 point versus the 50 threshold, +0.5 versus the previous reading and −1.5 versus forecast. An expansion signal can therefore improve while missing a forecast. A fall from 52 to 50.5 weakens the expansion signal; exactly 50 does not clear the threshold. Converting an index to GDP growth requires an estimated model and matching period, not subtraction of 50. These distinctions also belong in fundamental analysis; no difference alone guarantees a price reaction.

Caixin versus the official NBS index — why two gauges

NBS and S&P Global publish separate China survey families with their own samples and procedures. Compare manufacturing with manufacturing for the same month and inspect the relevant component. NBS non-manufacturing is not automatically interchangeable with S&P Global services activity. A gap between headlines does not identify one economic cause.

Both the NBS and S&P Global manufacturing surveys use samples stratified by industry and company size. Both cover small firms as well as large ones; it is misleading to equate NBS with a state-owned core and Caixin with a private export margin. Their readings can differ because the samples and survey procedures differ. Compare the component indexes and their recent histories before assigning an economic explanation to a gap.

China-related demand and both currencies in AUD/USD

China’s demand matters to its trading partners. Australia exports commodities including iron ore and coal, while New Zealand’s exposure includes agricultural products such as dairy. A Chinese PMI surprise can affect expectations for that demand and broader risk sentiment. The resulting currency response also depends on domestic conditions, market expectations and positioning. These links are discussed in intermarket analysis.

A stronger-than-expected Chinese PMI can support confidence in demand from China, while a weak report can undermine it. AUD, NZD and other exposed markets may respond, but neither the order of their reactions nor their direction is guaranteed. The headline, components, prior expectations and broader news all matter. For another example of interpreting a business survey, see the US manufacturing ISM gauge.

Publication clock and instrument conditions

Check the S&P Global PMI release calendar for the exact series and date. Use its UTC time column and verify the current series name, rather than assuming the historical Caixin label identifies every present entry. Convert the entry to your own zone rather than relying on a Central European clock. Asia-Pacific trading conditions depend on the currency, venue and time, not on whether it is night where you live.

Measure the actual instrument’s spread and execution conditions rather than inferring them from your local hour. Important information does not establish a mandatory first-minute trade, a typical spike or a later London reversal. Record the report, forecast source and time, then compare a defined price window with simultaneous news. See how to use an economic calendar for dated scheduling.

A separate hypothetical manufacturing model keeps orders, output, employment and stocks at 50. If delivery times fall from 50 to 40, inversion gives 100 − 40 = 60. With their 15% weight, the headline rises from 50 to 0.85 × 50 + 0.15 × 60 = 51.5, solely through delivery delays. This does not prove stronger demand: supply disruptions can also lengthen delivery times. It assumes comparable prepared components, not a reconstructed panel or seasonal adjustment. The services activity index does not include this delivery component in the same way.

What to do after you close this page

  1. Identify the current series. Save the publisher, reference month and UTC calendar entry; historical Caixin labels may remain in third-party calendars.
  2. Compare compatible surveys. Match RatingDog and NBS manufacturing for the same month. A difference alone does not locate state-owned or private firms.
  3. Calculate all three differences. Keep threshold, previous reading and dated forecast comparisons separate; none determines the price direction.
  4. Inspect components and observations. Include delivery delays, orders and activity, then document spreads and other news in the chosen AUD/USD window. Observation without a position is a valid exercise.